LHC
Artificial Intelligence: a Roadmap for California
Read the report at Little Hoover Commission ↗
Artificial Intelligence: A Roadmap for California
Report #245, November 2018
Milton Marks Commission on California State Government Organization and Economy
Little Hoover Commission Dedicated to Promoting Economy and Efficiency
Pedro Nava in California State Government
Chairman
Sean Varner
Vice Chairman/ Subcommitee
Member The Little Hoover Commission, formally known as the Milton Marks “Little Hoover”
Commission on California State Government Organization and Economy, is an
David Beier
Subcommitee Chair independent state oversight agency created in 1962. By statute, the Commission
is bipartisan and composed of nine public members, two senators and two
Iveta Brigis
Subcommitee Member assemblymembers.
Cynthia Buiza
In creating the Commission, the Legislature declared its p urpose:
Anthony Cannella
Senator
[T]o secure assistance for the Governor and itself in promoting
Chad Mayes economy, efficiency and improved services in the transaction of
Assemblymember
the public business in the various departments, agencies and
Don Perata instrumentalities of the executive branch of the state government,
and in making the operation of all state departments, agencies and
Bill Quirk
Assemblymember instrumentalities, and all expenditures of public funds, more directly
responsive to the wishes of the people as expressed by their elected
Richard Roth
Senator representatives. . .
Cathy Schwamberger
The Commission fulfills this charge by holding public hearings, consulting with experts
Janna Sidley and listening to stakeholders. During the course of its studies, the Commission may
Former Commissioners Who create subcommittees and conduct site visits.
Served During The Study
The findings and recommendations of the Commission are submitted to the Governor
Joshua LaFarga
and the Legislature for their consideration. Recommendations often take the form of
Helen Iris Torres legislation, which the Commission supports through the legislative process.
Commission Staff
Jeanne R. Wolfe
Executive Director (A)
Tamar Foster
Project Manager
Julissa de Gonzalez
Research Analyst
Contacting the Commission
Imran Majid
Research Analyst All correspondence should be addressed to the following:
With Gratitude To
Little Hoover Commission
925 L Street, Suite 805,
Mark Johnson
Graduate Student Intern Sacramento, CA 95814
(916) 445-2125
littlehoover@lhc.ca.gov
This report is available on the Commission’s website at www.lhc.ca.gov.
Letter From The Chair
November 27, 2018
The Honorable Toni Atkins The Honorable Patricia Bates
President Pro Tempore of the Senate Senate Minority Leader
and Members of the Senate
The Honorable Anthony Rendon The Honorable Marie Waldron
Speaker of the Assembly Assembly Minority Leader
and Members of the Assembly
Artificial Intelligence: A Roadmap for California is not merely a Commission report—it is a call for
action. Other states, cities and countries are surging forward with strategic plans to harness the
power of artificial intelligence in ways that will improve their economy, public health and safety,
jobs and environment. The race to develop and use AI for good is more akin to a marathon than
a sprint. It is fast paced and highly competitive, and one that California should be leading but is not.
Our state owns a lion’s share of the world’s economy and is home to Silicon Valley. But California
state government has yet to accept the AI challenge and truly begin the race. While some
departments are collaborating with the private sector and academia to develop and use new
technologies, the overwhelming majority are not preparing or strategizing for an AI world.
Imagine using AI applications to predict where fires may occur, detect early-stage wildfires, or guide
firefighters where best to fight a fire and save lives. Conceive of an environment where AI could promote
biodiversity and water conservation, and protect endangered species. See educators using AI to improve
student learning and increase graduation rates. Envision better detection of diseases, including cancer, and
more finely-tuned effective treatments. Certainly, such visions must be tempered with appropriate privacy
protections and robust laws aimed at preventing the misuse of data. In addition, this encouraging future, which
is presently knocking at our door, will require not just foresight but insight, not just political will but political
action, and not just one mind but a collaboration of minds in government, academia, and private industry.
Without question 2019 will test the new administration and Legislature with serious challenges from
wildfires to homelessness and skyrocketing health care costs. The justifiable cries to correct and solve these
momentous problems will compete with the pressing need for AI policymaking. But the Commission’s
study has found that lawmakers must not let this happen. Lawmakers must act with due diligence and
speed in 2019 to enact AI policies that will lead California into a future where AI is developed and used for
economic, social and environmental good. Challenges abound in this effort but so do vast opportunities.
Artificial Intelligence: A Roadmap for California provides the new Governor and Legislature with an AI
policymaking agenda that revolves around public engagement, building a human infrastructure, attacking
pressing social needs, and protecting core values—autonomy, responsibility, privacy, transparency
and accountability. In addition, the report offers lawmakers a high level plan that defines strategic AI
Letter From The Chair | 1
objectives and major steps to achieve those objectives. The recommendations are wide ranging and cover:
State government infrastructure and planning;
Education and training;
Collection of data;
Creation of an AI commission;
Incentives for AI investment, research and development;
Improving Employment Development Department (EDD) surveys; and
Apprenticeships for employees whose jobs and/or classifications may be displaced or
transformed by artificial intelligence.
The supporting materials take a deeper dive into the basis and reasons for the Commission’s
recommendations. There is also a listing of the top AI reports, experts, groups and books.
The Commission’s report thus provides a visionary, yet practical, roadmap for California’s ethical
development and use of artificial intelligence. Surely, as Peter F. Drucker once commented,
“[t]he best way to predict the future is to create it.” The urgency with which our policymakers
must act now cannot be overstated, since the race for AI superiority is already well under way.
Please know the Commission stands ready to provide any further assistance the new Governor or
Legislature may request on this topic.
Pedro Nava
Chair, Little Hoover Commission
2 | Little Hoover Commission
Contents
Commission Report
6
Introduction
..........................................................................................................
7
What is AI and What Opportunities Does it Offer?
.....................................
10
How Governments are Preparing for AI
........................................................
15
Agenda for Change
.............................................................................................
Supporting Materials and Resources
Overview of the Commission's Study 21
.....................................................
Sectors Impacted by AI .................................................................................. 22
How are Academia and Private Industry Using AI? ........................... 24
How can AI be Used for Public Service? ................................................. 26
What Risks Does an AI World Bring? ........................................................ 28
Anticipated Impacts of AI on Employment .......................................... 32
Are Californians Being Prepared for an AI World?............................. 36
Are California State Agencies Preparing for AI? ................................. 41
Are Lawmakers Preparing for AI? .............................................................. 42
Resources ............................................................................................................. 43
Contents | 3
Figures and Graphics
Figure 1: Weak and Strong AI ..................................................................... 7
Figure 2: Types of AI ................................................................................... 7
Figure 3: Automation and AI ...................................................................... 8
Figure 4: Expenditures on Education and Skills Training ............................. 9
Figure 5: Sizing the Prize: Which Regions Gains the Most from AI? ........... 12
Figure 6: Feasibility of Automation ............................................................ 13
4 | Little Hoover Commission
Page Intentionally Left Blank
5
Introduction
Artificial intelligence (AI) is already changing the can work together across the lines of political parties and
structure of goods and services in the economy, and with stakeholders such as workers, students, educators,
altering the nature of work. This has major implications researchers and the business community. One of the keys
for our workforce and opens critical questions about our to success will be to configure government, especially
human values like privacy. The fundamental difference the Executive Branch, to embed a comprehensive and
between these challenges and those faced by our cross cutting view of artificial intelligence. This report
ancestors in earlier disruptive eras, like the agricultural or urges action to create a new Cabinet level position
industrial revolutions, is the magnitude and pace on AI accompanied by an appropriate cross-agency
of change. commission. The state of California as the user of AI and
as the regulator of some AI products and services should
In the period between now and 2030, artificial fully engage the public in a robust discussion of AI and
intelligence will—according to a recent McKinsey serve as a vibrant example of best practices in the use of
report—increase global gross economic product by $13 AI. The opportunity to prepare for the AI world is now;
trillion.1 This increase roughly translates to an economic otherwise, California may find itself left behind.
effect in California of well more than $400 billion.2 The
impact of artificial intelligence on jobs stretches from The Governor and Legislature should:
job losses of 5 percent to job changes3 or alterations
of up to 60 percent.4 These percentages are of course Engage the public in broad conversations about the
estimates and, without more accurate data collection by responsible use of AI;
government, the exact impact on California jobs is hard
Identify the necessary training and other educational
to predict. It is, however, within the realm of possibility
opportunities to build the human infrastructure
that artificial intelligence could affect from 1 million to
11 million California jobs.5 In a future where artificial required for an AI-fueled economy;
intelligence is virtually interwoven in the fabric of the
Explore opportunities to form and leverage public
economy and daily activities, preserving a sense of
private partnerships to use AI to address societal
human control is vital.
challenges; and
AI poses four key decisions for California: (1) how to Examine legislative and policy strategies to preserve
support AI research and responsible AI use to grow the
and protect key public values: privacy, transparency
state’s economy; (2) how to take advantage of advances
and accountability.
in AI to enhance services to Californians; (3) how to
configure a new structure for lifelong education and The policy topics arising from artificial intelligence
training to respond to the inevitable disruption in the adoption the state must confront, plan for and solve will
tasks or content of work; and (4) how to protect its values be varied, complex and often daunting. To be clear, no
of privacy, transparency and accountability in this new single report or legislative session can address all these
economic era. issues. Rather this report urges the new Governor and
Legislature to begin these processes with a sense of
The magnitude of these challenges is immense. The urgency and an understanding that the emergence and
suite of actions by the next Governor and the incoming widespread use of artificial intelligence poses among
Legislature needs to be big, bold and scalable. the most important inflection points for the economy,
education and training, employment and societal values
The state of California will be well served if its leaders in the history of the state.
6 | Little Hoover Commission
What is AI and What Opportunities Does it Offer?
Introduction to Artificial Intelligence
Artificial intelligence (AI) is a quality of any computer
program (algorithms, data structures and data) that
can sense, reason, act and adapt like humans.6 While
many AI systems can also learn over time (e.g., a spam
filter, a recommendation system), not all can do so.
An expert system built to aid human reasoning, or an
autopilot that can land an airplane, are examples of AI
applications that do not learn. Whenever a computer
program’s performance gains near-humanlike abilities, or
better than human abilities, to sense, reason or act, we
Figure 2
tend to call that AI.
One of the most important sets of AI techniques,
which began to make progress in the 1990s, is machine
learning (ML). ML is a set of techniques (algorithms,
data structures and training techniques) used by
engineers to allow computer programs to get better
as they learn from new data, either in supervised
(human-trained) or unsupervised (machine-led) modes.
Many modern computer programs use ML techniques,
including search engines, spam filters, recommendation
systems, image classifiers, financial trading systems,
Figure 1 financial “roboadvisors,” scheduling agents, customer
service chatbots, environment and health monitors,
smartwatches, smartphones, smart speakers and
Two types of AI are commonly discussed: weak or narrow
many others. The use of ML has grown substantially
AI and strong or general AI. Weak AI are programs that
because the large data sets necessary for training are
are humanlike on narrow tasks for particular applications.
now accessible on the web, and the computing power
Examples include advanced video game engines,
required to use ML techniques is now affordable.
diagnostic technology for interpreting x-rays and anti-
collision technology in a car. This is very different from
There are at least five major computer engineering
strong or general AI, a smarter-than-human system that
approaches to ML.7 One that is particularly
functions across multiple areas of human thinking and
revolutionizing AI today is called deep learning (DL). DL
acting, as depicted in science fiction movies like HAL
is a ML technique for using brain-like learning algorithms
in 2001, Samantha in the movie Her, or Skynet in The
(neural networks) to filter and learn from large amounts
Terminator. In reality, narrow AI is already in broad use
of data, generate probabilistic responses and adapt to
today, while the arrival of general AI is likely several
new situations. In the early 2010s, DL began to make
decades away or more.
significant progress on previously very difficult problems,
like speech recognition, language translation, language
understanding, image recognition and spatial navigation.8
DL systems are particularly good at finding hidden values
What is AI and What Opportunities Does it Offer? | 7
and relationships in large sets of data, filtering out noise service cost, create wealth and are themselves new tools
to find signals, putting information together in novel that become cheaply or freely available to use in new
patterns, improving based on user feedback and adapting ways. The new tasks and capabilities that tomorrow’s
to new situations in creative ways. workers will gain by using AI-enabled products, services
and platforms in the workplace, the new jobs AI will
Governments, with their vast data sets and user base, are create and the new forms of training that AI will enable in
ideal customers for DL applications. Many are now using the workforce, are hard to predict.
DL for document discovery, automating documentation
and administration, sensor monitoring, cybersecurity Possible Benefits
and answering citizen queries with chatbots.9 DL
systems can have bias and be poorly adapted to their AI has the potential to revolutionize major areas of life for
tasks, particularly in their early versions. But they are economic, social and environmental good. The European
also learning systems, which allow their performance Commission, as a part of their Digital Single Market series
to get better and safer over time. Just as with human of reports, outlined specific examples of how AI might
brains, engineers can make DL systems more adaptive by assist in meeting social needs, including:
ensuring they begin with good algorithms, good training
and good data (broadly inclusive and representative). Reducing traffic accidents.
In the near term, DL based systems are expected to
Improving efficiency in the use of energy and
continue to do our most complex AI tasks, like driving
water resources.
cars, identifying faces, monitoring security networks,
answering questions in natural language and generally
Reducing risks of work-related injuries.
improving our ability to use and learn from data of all
types. In the longer term, they show the most promise to Detecting pollution and oil leaks.
become increasingly general in their problem solving and
adaptive abilities. Reducing trauma after surgery and helping
surgeons operate more precisely.
Reducing the use of pesticides and need
for irrigation.
Improving access to education.
Improving speed and accuracy in diagnosis
of diseases.
As for jobs, AI could make many easier by performing
mundane tasks, thus allowing humans to perform more
Figure 3
interpersonal and creative activities.11 Such a transition
It is worth calling out the relationship between may likely promote higher productivity and individual
automation, artificial intelligence and task or job creation job satisfaction. Bank tellers who no longer dispense
or elimination. Automation involves the replacement of cash are freed to improve the customer experience and
repetitive human physical or mental tasks with purpose- advise management on options for new services. Tony
designed technology. It has been occurring as long as Blair, executive chair of the Institute for Global Change
humans have had the ingenuity to mechanize certain and former UK Prime Minister, said, “AI will allow us to
forms of work. AI allows the automation of more complex do what it is that we are uniquely meant to do: to focus
types of human sensing, reasoning, acting or adapting, on high-level thinking, strategy, and paving the way for
with technology that can learn and improve over time. innovation.”12
Both automation and AI continually create, as well as
eliminate, tasks and jobs.10 Both automation and AI will AI will also drive the creation of new occupations,
free up human time and capital, reduce product and which may help to counterbalance job displacement.
8 | Little Hoover Commission
Investments in renewable energy and mitigation of
climate change may create demand for workers across
The 23 Asilomar Principles
several sectors, from manufacturing and construction to
installation. Growth in the population of people age 65 and
older may create demand for occupations in the health care California Assemblymember Kiley, with
sector, including doctors, nurses, home health aides and several co-authors, introduced Assembly
nursing assistants.13 Concurrent Resolution 215. The resolution
was enacted on September 7, 2018. Over
The possible benefits of AI are thus significant and should 1,200 AI and robotics researchers signed the
be broadly shared. Consequently, developers and software principles, which are intended to promote the
engineers must be provided proper incentives to develop AI safe and beneficial development of AI. Below
to perform good works. A balance must be struck between is a sampling:
AI investments by the private sector and AI training and
#3 Science-Policy Link: There should be
education for workers. In the United States, spending on
constructive and healthy exchange
formal education has substantially outpaced spending on
between AI researchers and
training. This is concerning because the advent of AI will
policymakers.
require a much more nimble workforce. Without proper
training, employees may find themselves without the skills #8 Judicial Transparency: Any involvement
necessary to succeed across a variety of sectors. by an autonomous system in judicial
decision-making should provide a
California’s future labor force therefore stands to benefit satisfactory explanation auditable by a
from additional worker training and education, and competent human authority.
workforce transition programs. These investments will lead
#12 Personal Privacy: People should
to increased public tax revenues and reduced reliance on
have the right to access, manage and
social safety net programs.14
control the data they generate, given AI
systems’ power to analyze and utilize
that data.
#15 Shared Prosperity: The economic
prosperity created by AI should be
shared broadly, to benefit all
of humanity.
#20 Importance: Advanced AI could
represent a profound change in the
history of life on Earth, and should
be planned for and managed with
commensurate care and resources.
#23 Common Good: Superintelligence
should only be developed in the service
of widely shared ethical ideals, and for
the benefit of all humanity rather than
Figure 4: Council of Economic Advisors. July 2018. “Addressing America’s one state or organization.15
Reskilling Challenge.”
What is AI and What Opportunities Does it Offer? | 9
How Governments are Preparing for AI
Artificial intelligence has the potential to change Other Countries
dramatically how governments do business, provide
public services, predict and respond to natural disasters, Many other countries are developing and implementing
conduct intelligence, improve national security and fight AI policies at a record pace. Below is a sampling of
wars. More importantly, governments must implement their efforts.
strategies and policies to encourage the private sector
to participate meaningfully in AI policy development. Europe: In April 2018, the European Commission
Consequently, countries around the globe and even states proposed a multi-pronged approach to boost investment
and cities in the United States are taking early steps to and set guidelines for AI among the European Union’s
understand and take advantage of AI technology and 28 member states. The Commission’s strategy includes:
applications. California can learn from and expand upon investing €20 billion in AI research and innovation by
recent advances made by the United States government, the end of 2020; modernizing education and training
other states and foreign governments. systems; supporting labor market transitions to prepare
for socio-economic changes; and establishing ethical
United States Federal Government guidelines for AI development that take into account
data protection, transparency, and other concerns. In
The Trump Administration in May 2018 pledged it would addition, the Commission appointed 52 experts to a
prioritize funding for AI research and development new High-Level Expert Group on Artificial Intelligence
as a means to economic growth and announced (AI HLEG) that includes members from academia,
the formation of the Select Committee on Artificial civil society, and private industry. The AI HLEG will
Intelligence, a task force under the National Science and provide recommendations for future AI-related policy
Technology Council, to promote a national approach development, including ethical, legal and societal
to the development and use of AI.16 Subsequently, the considerations. In addition, the AI HLEG will advise the
United States House of Representatives, Committee on Commission on mid to long-term AI challenges and
Oversight and Government Reform Subcommittee on opportunities and propose ethical guidelines.19
Information Technology issued a report in September
2018, Rise of the Machines, Artificial Intelligence France: After a six month review, the French government
and its Growing Impact on U.S. Policy. The report unveiled a national strategy for artificial intelligence
recognizes that the “United States cannot maintain its technology in 2018, including plans to invest €1.5
global leadership in AI absent political leadership from billion over the next five years. The review considered
Congress and the Executive Branch” and recommends possible applications of the technology and the role of
“increased engagement” by the federal government.17 government in several specific sectors: health, ecology,
Congressional members have introduced legislation transport and mobility, and defense and security.20
to better understand how rapid AI innovations may
benefit Americans. A bipartisan Congressional Caucus Switzerland: In a pilot program overseen by the Swiss
on Artificial Intelligence was created in 2017 to consider State Secretariat for Migration and the Immigration
the implications and opportunities of the technological, Policy Lab, refugees seeking asylum in Switzerland will
economic and social impacts of advances in AI.18 be placed in different parts of the country using an
algorithm designed to maximize their chances to find
employment.21
10 | Little Hoover Commission
Russia: Russia is investing in artificial intelligence, economic growth and productivity. To date the task force
particularly to support the nation’s military. Reports has considered state and national workforce trends,
suggest that the Russian military is designing autonomous the kinds of skills needed in a changing economy and
vehicles to guard its ballistic missile bases and an implications for Indiana’s universities and how changing
autonomous submarine capable of carrying nuclear economic forces may impact vulnerable populations.28
weapons. The Russian military is also experimenting with
battlefield robotics, such as remotely piloted tanks.22 Massachusetts: In November 2017, Massachusetts
established a Commission on Digital Innovation and
China: Like Russia, China invests heavily in artificial Lifelong Learning at the Massachusetts Institute of
intelligence to boost national security and increase Technology to examine how the state can expand
military capabilities. In 2017, China’s government issued a affordable, high-quality, employer-aligned education,
plan to lead the world in artificial intelligence technology training and lifelong learning opportunities in an era of
by committing to invest capital resources into AI research technological change.29
and development. The plan also provides guidance and
policy on artificial intelligence development.23 Weak Michigan: Michigan’s Department of Licensing and
privacy laws allow China to provide its AI industry with Regulatory Affairs uses artificial intelligence and machine
access to data on more than 1.4 billion citizens.24 To learning technologies to predict the likelihood of drug
ensure a workforce capable of supporting national AI related deaths and reveal other insights into the state’s
goals, Chinese schools include AI-related courses in opioid epidemic.30
primary and secondary education.25
New York: New York lawmakers in December 2017
South Korea: Known for cutting-edge industrial created a temporary, unpaid commission to investigate
robotics technology in its automotive, electronics and how to regulate artificial intelligence. Among other
semiconductor industries, South Korea plans to invest topics, the New York State Artificial Intelligence
₩2.2 trillion (U.S. $2 billion) by 2022 to strengthen Commission would study current state law related
its artificial intelligence research and development to artificial intelligence, examine the impact of AI
capability.26 In addition, the nation intends to establish on employment and the disclosure of confidential
six new research and development institutes, which information and the potential impact of the technology
will be focused on human resources, technology and industry of proposed regulations.31
infrastructure.
Vermont: In May 2018, Vermont’s Governor
Japan: Japan’s Strategic Council for Artificial Intelligence signed legislation to create a task force to develop
Technology released a formal artificial intelligence recommendations on the responsible growth of the
strategy in March 2017. The plan acknowledges the need state’s emerging technology markets, the use of
to develop technology researchers, engineers and data artificial intelligence in state government and the state’s
scientists capable of growing the industry sector. It also regulatory role in the field of artificial intelligence.32
outlines a multi-pronged strategy for government to
promote the infrastructure necessary to support artificial Washington: To stay up-to-date on workforce changes
intelligence capabilities, particularly in the areas of due to automation, AI technology and other factors, the
productivity, health, medical care, welfare and mobility.27 state of Washington established the Future of Work Task
Force. Among other responsibilities, the task force will
U.S. States & Select Cities inventory and assess industry trends, identify policies
to help Washington’s businesses and workers respond
Various states have a head start in addressing to rapid changes in technologies and create a policy
implications of AI. framework to support lifelong learning.33
Indiana: Indiana state leaders created The Future of Hawaii: Hawaii created a working group to consider
Work task force to study new technologies and devise whether to implement a universal basic income (UBI)
innovative workforce development strategies to drive program in Hawaii.34 Under a UBI program, every citizen
How Governments are Preparing for AI | 11
is granted a fixed income that is not dependent on their How is California Preparing
status in life. The UBI concept has received attention
and appeal, because it could minimize the impact of With the exception of the Governor’s Office of Business
automation and AI technology on jobs.35 and Economic Development (GO-Biz), Department of
Technology and the California Department of Food and
New York, NY: In January 2018, the New York City Council Agriculture, California’s state government has not kept
passed legislation to create accountability for algorithms pace with national, state and even city governments on
by establishing a task force to study how city agencies organizing efforts to address the implications of AI. Given
use algorithms to make decisions that impact people and its extraordinary intellectual and innovation resources
whether those systems appear to discriminate against and, as the fifth largest economy in the world, California
people based upon such protected characteristics as age, has the opportunity to marshal not only its public and
race, religion or gender.36 private resources to immediately begin addressing the
implications of AI but to begin taking advantage of AI
Pittsburgh, PA: The City of Pittsburgh, Pennsylvania is to enhance services to Californians and become the
using artificial intelligence, among other technologies, world leader in adopting thoughtful policies to ensure
to transform itself into a “future city.” Since 2015, city AI positively impacts the state’s economy and, more
officials have partnered with The Heinz Endowments to importantly, all of its constituents.
forge a new model of urban growth and development,
fueled by innovation, robotics, automation and artificial Although the proliferation of innovation in the AI field
intelligence. Dubbed P4 Pittsburgh–emphasizing the will impact and ultimately transform the labor market,
framework of people, planned place and performance the extent and timing of such impacts remains unclear.
–the project will explore how cities adapt to rapid Studies throughout the world reach different conclusions.
technological changes, while addressing social and Some researchers predict job losses due to automation,
economic inequality, environmental threats and the built while others anticipate job growth, and still others
environment.37 suggest the nature of work, specifically the tasks that
Figure 5: PricewaterhouseCoopers. June 2017. “PwC's Global Artificial Intelligence Study: Sizing the Prize.”
12 | Little Hoover Commission
comprise work, will change. The Commission heard to complete critical job tasks.41 Economists point to
testimony from diverse groups confirming the disparity of historical examples of new technologies that increased
such opinions in California. Notwithstanding the varying the demand for workers to focus on tasks that machines
opinions, a general consensus exists among experts that could not. An often cited example is the advent of the
lawmakers, educational institutions and private sector ATM machine, which performs routine banking tasks
policymakers must act quickly to influence not only the for customers, thus permitting bank tellers to provide
impact of AI on employment, but to propel California as a other, less routine, banking services to customers.42 In
leader in the transition of the labor force as a result of AI addition, automation and AI created human efficiencies
and accelerate statewide economic prosperity emanating may alleviate the impact of job attrition caused by baby
from AI. boomer retirements. Potential results include a higher
skilled workforce and higher median wage.
To predict trends in its workforce, the various state
departments have insufficient data and data collection To achieve these results, California must expedite the
processes to accurately predict how artificial intelligence adoption of policies addressing the impending new
may impact California’s workforce. The lack of accurate workforce demands. While the immediacy of adoption
data stems from outdated national employment models is critical, such policies should address access to training
on which state departments rely. The models predict among all demographics and skill levels in California.
workforce development based on historic trends.38 The state can partner with non-profit organizations
This type of information will not account for the rapidly to establish better online portals for access to skill
developing effects of automation on California’s future assessments, training resources and job search tools,
workforce. While the state requests additional data from such as the Council for Adult and Experiential Learning
private sector employers through voluntary surveys, low (CAEL). California also can actively notify workers once
participation in these surveys affects the reliability of they apply for unemployment insurance that they qualify
the information.39 California must develop an accurate for unemployment insurance benefits if they participate
method of predicting labor needs with the advent of AI in approved training.43 Provided it could expediently
and automation. For example, the California Community implement workforce training and retraining policies, the
Colleges Centers of Excellence (COE) for Labor Market state should prioritize collaborations with stakeholders
Research uses their own economic modeling to customize
state projections. COE conducts special projects to
better understand the workforce needs in evolving
and emerging industries. By conducting point-in-time
surveys and talking to employers in controlled ways, COE
has more accurate real time information of job trends.
Expanding the role of COE would require legislative
approval.40
The threat of job loss stemming from an increase in
automation and other AI-based technologies has elicited
doomsday scenarios featuring job-stealing robots. Others
predict a possibility that automation will create more
jobs than it would eliminate. Ultimately, it appears likely
that AI will accelerate the transformation of many jobs.
National labor market trends provide some insight on
how new technologies might change the labor market,
both nationally and in California. Researchers predict
job losses due to automation, but for more workers,
the nature of their job likely will change through
partial, rather than complete, automation of activities,
Figure 6: McKinsey and Company. July 2016. “Where Machines Could
or by working alongside robots and smart machines
Replace Humans – and Where They Can’t (Yet).”
How Governments are Preparing for AI | 13
from affected demographics, including broad access to 5. Assignment of a pre-market approval role for some or
the opportunities emanating from the implementation all AI-related products.
of AI. For example, to address changes in the trucking
There are competing interests in making choices in this
industry resulting from autonomous vehicles, U.C.
area. California should carefully weigh and assess the
Berkeley researchers suggest policymakers design
social and economic benefits of using AI against the
job-quality standards to ensure living wages and good
nature and extent of risks associated with bias. It will be
working conditions for drivers and create regulations to
important to recognize that a single state of the United
incentivize clean trucks to reduce pollution and other
environmental and public health costs.44 States can be both a leader and a potential roadblock.
California has led the way in matters like fuel efficiency
for cars, but if not careful, could end up creating a
Values: Autonomy, Responsibility, Privacy,
moat around AI that leaves the state and nation worse
Transparency and Accountability
off, particularly if regulations slow or reduce American
interest in this highly competitive global race. The easier
There is no doubt that the use of algorithms in decision
answers are likely also wrong. Too much regulation can
making poses a risk of error and bias. The public policy
block or deter adoption even when the use of AI can
question is what to do about these challenges and how
improve outcomes and lower costs. Just as regulators
to balance the benefits against the risks. For California
today in all fields must balance benefits against the
government, the first task is to help educate and inform
cost of risks, so it will be true here. The best approach
policymakers in state government, including within
for California in the short term is to learn more about
the executive branch and departments, agencies and
the promise of AI and its pitfalls long before taking
commissions, on how algorithms are created, tested
permanent and disruptive action.
and used. As a state with the best and strongest set of
research universities in AI, California should sponsor
research and development in this area.
Beyond that, an array of policy choices must be
confronted. These policy approaches range from intrusive
regulations that require permission before AI is used to
softer, non-governmental standards and self-regulation.
Some of the options California could consider include:
1. Industry based standards of disclosure and
explanation of algorithms (including relevant data
sources and programming assumptions and any
risks of bias due to the demographics of the human
programmers) aimed at improvements in design and
implementation to avoid unwarranted bias.
2. Legally mandated disclosure or algorithm explanations
of varying detail.
3. Public sector procurement rules, including selected
use of open source mandates for some state contracts.
4. Sector specific rules with focus on the risks in the most
sensitive fields such as criminal justice, health and
matters affecting privacy.
14 | Little Hoover Commission
Agenda for Change
AI Roadmap Accordingly, the Commission makes the following
recommendations to assist the Governor and Legislature
California is poised to be a natural frontrunner in the in these efforts:
race to prepare for AI, given its large economy and
location, which serves as the home base for cutting-edge Recommendation 1: The Governor should appoint
academia and private research into the development an artificial intelligence (AI) special advisor within the
and use of AI technology and applications. Yet, because Governor’s cabinet. This position should report directly
the field of AI is fast moving and changing, developing a to the Governor or to the Governor through his or her
strategic statewide response will require both innovative designee. Among other duties, the AI special advisor
and dynamic leadership. should:
California state government is underprepared to take the Coordinate with AI counterparts within state and
lead. It lacks any single clear leadership and focus on the local agencies, offices of other elected state officials
development and use of AI technology and applications and California’s higher education institutions.
to improve internal and external operations and services
within an ethical framework. Nor has it begun the Create a strategic plan that incorporates AI within
necessary work to forecast and prepare for the inevitable state government to improve and augment
changes the new technology will impose on the state’s internal and external operations and services, and
workforce and economy. This void could leave California also oversee the safe, beneficial and transparent
flatfooted in a highly competitive race for AI superiority deployment of AI technology and applications in
where only the winner takes all. state government. The plan should account for
changing job requirements due to AI technologies
However, the Governor and the Legislature have the and how to mitigate the impact of AI on jobs
opportunity to fill this void and can provide the vision, with apprenticeship, training and educational
coordination and sustained, unified effort for California opportunities.
state government to plan for and use AI technology
Provide the Governor with annual reports
for economic, social and environmental good. State
documenting any advances, deficiencies or other
government leaders should begin now to educate
relevant information regarding the planning and use
themselves about the implications, benefits and risks
of AI by state agencies and standalone departments.
of AI-powered technology. As a start, the Commission
urges the Governor and Legislature to take appropriate
Recommendation 2: The Governor and Legislature should
action and measures specific to the challenges and
require each state agency or standalone department
risks inherent with AI technology and applications
to designate a chief artificial intelligence officer. This
and begin to formulate a policy agenda in early 2019.
position should report to the agency secretary and
The goals of the agenda should be to grow the state’s
coordinate with the Governor’s AI special advisor. Among
economy, take advantage of AI to enhance services to
other duties, the chief AI officer should:
Californians, reconfigure its educational institutions to
provide necessary education and training, improve data
Develop, coordinate and implement the agency’s
collection and sharing while protecting sensitive data and
AI strategies.
promote privacy, transparency and accountability in the
development and use of AI. Consider the designation of similar positions
within the agency’s departments, particularly in
Agenda for Change | 15
departments which may be significantly impacted This plan should include steps necessary to expand access
by artificial intelligence. to and understanding of computer sciences in order to
upskill both current and future workers. Other scientific
Recommendation 3: The Governor and Legislature
disciplines related to AI such as psychology, mathematics,
should encourage the designation of a chief AI officer
statistics, ergonomics and engineering curriculum should
within independent offices to partner on developing a
also be revised to ensure that education and training
statewide AI strategy. Such offices may include:
methods, and skill assessment tools match up with future
workforce projections. Importantly, the study of ethics
The State Controller.
in the development and use of AI should become a core
competency for those seeking AI education and training.
The State Treasurer.
The Secretary of State. This plan should integrate public input and participation
of the private sector to make sure public institutions
The Trustees of the California State Universities. understand the scope, nature and pace of employment
changes anticipated by the private sector.
The University of California Board of Regents.
The plan should recognize and stress the vital importance
Recommendation 4: The Governor and/or Legislature
of consulting with and listening to the existing workforce,
should require each state agency, which includes
including union representatives. This will be especially
departments, boards, commissions and the like, to devise
important for state workers, because the state should
strategic plans that include the use and implementation
serve as an exemplar for employers that make AI
of AI technology and applications to improve and
beneficial for all stakeholders, including employees.
enhance the agency’s internal and external operations
and services within an ethical framework that promotes
A short term goal of the plan should be to provide details
the use of AI for economic, social and environmental
for apprenticeships and other on-the-job training that will
good.i The strategic plan should also include strategies to
facilitate the transition of workers into new and changed
minimize the impact of AI technology and applications on
workplaces. Broadly sharing the benefits—as embodied
jobs and job classifications.
in the Asilomar AI Principles—is already state policy,
so implementing these programs in state government
Recommendation 5: State government, including the
should be an early priority.
executive branch and independent entities such as local
school districts, government funded or subsidized training
Recommendation 6: The state of California—in
providers, the U.C. system and the California State
recognition of the central importance of relevant, useful
University (CSU) system, as well as regional workforce
and unbiased information—should implement changes to
development entities, should create and implement a
improve the collection of data on jobs and skills for future
comprehensive strategic plan to better assure that future
workers. Once collected and reviewed to assure validity
workers are prepared for the new skills required for
and utility, it should be state policy to operate on the
the jobs in a rapidly evolving workplace focused on the
basis that data should be widely and easily shared within
application of technology including AI.
all parts of state government provided privacy is assured.
A similar presumption should apply to sharing data with
the private sector when the benefits exceed the risks and
i The use of AI technology and applications may be particularly poignant for appropriate governance rules and compliance systems
agencies and/or departments that have missions critical to health, safety, are in place to protect personal privacy. The initial focus
law enforcement and the protection of California’s natural resources and
should be on data related to addressing issues of skills
wildlife. The impact of natural disasters, such as earthquakes, floods, storms,
and future job trends. This data should become useful
wildfires and droughts, can be catastrophic to people, the environment,
to improve the likelihood that public policy aims to focus
domestic animals and wildlife and infrastructure. Commissioners heard how
AI can improve emergency response and firefighting efforts. AI technology on equity, mitigating income inequality and ending unfair
and applications, thus, present opportunities that California can ill afford to distribution of government benefits, funding and services
ignore.
16 | Little Hoover Commission
based on impermissible discrimination and to avoid Recommendation 10:
unfair geographic effects. The Governor
California state
and Legislature
government employs
Recommendation 7: The Governor and Legislature should promote
more than 232,000
should create an AI commission made up of apprenticeships
employees.46
knowledgeable professionals and experts in the field of and other training
AI from private industry, governments, nonprofits, unions opportunities for
and academia, for the purposes of: (1) developing AI- employees in private
related demonstration projects for critical state services; industry, and state
(2) incorporating successful AI-related demonstration and local governments
projects into the state system; and (3) further advancing whose jobs and/or classifications may be displaced or
how data science can be effectively utilized by state transformed by AI technologies and applications.
government.
Public Engagement
Recommendation 8: The Governor and Legislature should
ensure an environment in California of continued and Public officials and policymakers in state government
sustained efforts to stimulate investment, research and are accountable to the people. Part of that responsibility
development into AI technologies and applications within is for them to engage and educate residents of the
an ethical framework that promotes AI for economic, state. Many matters before the Governor, including
social and environmental good. This effort should include executive departments and agencies and the Legislature
collaboration with stakeholders. are debated and decided by relatively few people and
often without full public transparency. In the case of
Recommendation 9: The Governor and Legislature artificial intelligence, because the scope and extent of
should improve the data collection related to at-risk its utilization, more must be done to inform and listen to
jobs. In particular, consideration should be given to the general public. This is true, in part, because virtually
requiring the Employment Development Department all current students will be entering a workforce shaped
(EDD) to develop an addendum survey to the Bureau of by the application of AI. Most current workers will see
Labor Statistics (BLS)’s survey that asks questions about some changes in the nature of the work they perform as
the impact on jobs of automation and, specifically, AI a result of AI. Some categories of jobs will be sufficiently
technology. To maximize the collection of data related disrupted that those jobs may disappear. The exact
to at-risk jobs, there should be improved evaluation of nature and extent of these changes is not known, but
data that, at a minimum, identifies how the impact of AI more than a handful should know about the potential for
technology and applications will differ by region, county, positive outcomes and about steps that could be taken to
employment sector, socio-economic group, educational mitigate any harm.
attainment, age group, gender and other appropriate
characteristics. In the case of artificial intelligence, there has been
substantial press coverage, but the level of fundamental
understanding remains low. Most of the debate has
been about the consequences of AI and not about what
The EDD uses the federal BLS employment
it is, how it can be applied or how to balance between
surveys, but the Commission’s study found
the potential for progress and risks. In the case of AI,
a two-fold problem with the BLS surveys:
we recommend California plan for and execute public
(1) many tech companies do not fill out
engagement using a suite of tactics that reach people
the surveys; and (2) the surveys do not ask
where they are most likely to accept news and facts.
questions specific to job loss or change due to
the automation of jobs.45
As a representative of the American Association for the
Advancement of Science put it: “We need to engage the
public in a more open and honest bidirectional dialogue
about science and technology and their products,
Agenda for Change | 17
including not only their benefits but their limits, perils, 4. What are the education and training requirements for
and pitfalls.”47 major current job categories and how well to existing
institutions of government meet those needs?
Public engagements needs to reach a broad cross
5. What is the gap workers will face between their
section of the public, be seen as fair and balanced
current skills and new job requirements over the next
and be structured to listen as much as inform. Public
20 years due to the changing nature of work arising
engagement consists of tools such as:
from AI?
Public hearings, both legislative and by agencies of
6. Should California establish a lifelong learning account?
government;
How should it be structured, who should run such a
Written, video, audio and online presentations and program, how much will it cost and how should it be
financed?
interactive techniques; and
Online surveys and other tools used by other nations 7. What changes, if any, should be made to the
unemployment insurance program to integrate into a
such as Denmark, France and the United Kingdom.
method for securing improved worker training?
These approaches include principles such as: (1) quality
8. What changes should be made with respect to better
of outcomes; (2) legitimacy of outcomes; and (3)
preparing students for AI in: (1) the University of
administrative efficiency. These principles have been
California system, (2) the California State University
embodied at the federal level in a set of principles
system, and (3) California Community College system?
of public engagement supported by federal law (the
Are any changes necessary in curriculum, financing
Administrative Procedures Act) and informed by a
tuition, the duration of enrollment or the types of
commitment to a framework of “responsible innovation”
degress or certificates offered?
that is “transparent, interactive process by which societal
actors and innovators become mutually responsible to
9. What role should the private sector play in worker
each other with a view to the (ethical) acceptability,
training? What role should state policy play in
sustainability, and societal desirability of the innovative
encouraging or subsidizing such training? How can the
process and its marketable products.”48
private sector assure training in more remote, rural or
hard to reach urban communities? What role should
Public engagement is central to the development of
apprenticeship programs play? What role should
a societal consensus. Past experience on vexing and
organized labor play in education and training?
complex scientific issues suggests that setting aside
funding for reviewing ethical, social and legal issues is 10. What changes should be made by state government to
vital.ii prepare the current and future workforce for AI? What
role should state government unions play?
Key Issues in Building a Human
Attacking Pressing Social Needs Using AI
Infrastructure
Several witnesses and experts told the Commission that
1. What is the scale and changing nature of work due to
social needs can be addressed and outcomes improved
AI? How many jobs will be gained and lost by type?
through the application of AI. Examples include access
to and use of state controlled data such as anonymous
2. What will be the pace of the changes in jobs due to AI?
electronic health data, especially from the University of
3. What types of jobs will change and when will they be California health system. Efforts are already underway to
most affected? use this testbed to derive insights into better diagnostic
and treatment options. Others have suggested the use
of AI to improve the identification of potential recipients
of social services, such as teenaged homeless youth.
ii The Human Genome Project provides an example: https://www.genome.
Yet others have pointed to many potential uses of AI
gov/10001618/the-elsi-research-program/.
18 | Little Hoover Commission
in responding to climate change, including planning to
optimize the use of intermittent energy sources such as
wind and solar.
This report does not urge any specific public private
partnership on these, or other, specific areas of societal
concern. Rather we urge state leaders to examine the
application of this tool to any appropriate field where it
fits and optimizes benefits to the residents of the state.
1. What areas of social need are most amenable to be
addressed using the tools and techniques of AI?
2. What are the optimal formulas for creating public
private partnerships? What should be the principles
of governance and accountability for such entities?
3. What role, if any, should the state play in financing
public private partnerships?
4. What examples exist in other jurisdictions or on
other topics from which California can learn?
Protecting Our Values
It is well accepted that AI and its use has privacy
implications. Political jurisdictions such as the European
Union have implemented new rules that cover AI. Closer
to home, the California Legislature passed and the The California Consumer Privacy Act of
Governor signed a new privacy statute, the California 2018 establishes consumer rights over their
Consumer Privacy Act of 2018. The best next step for personal information. The Act includes rights
the incoming administration and Legislature to examine related to disclosing personal information,
over time that statute to determine how effectively it is deleting personal information and stopping a
working and whether it is a good fit for AI-related issues. business from selling personal information.50
It will be useful to review the AI-related privacy
principles of non-profit groups, companies and their
trade associations, as well as those recently drafted
by Congressman Ro Khanna.49 These principles offer
additional insights into AI-related privacy questions.
Questions associated with transparency and
accountability are complex and fraught with risks of
unintended consequences. Moreover, rules in this area
potentially can be overly inclusive, under cover key issues
or insufficiently address unique challenges faced by
specific sectors (e.g., autonomous vehicles).
Agenda for Change | 19
SUPPORTING MATERIALS AND
RESOURCES
20 | Little Hoover Commission
Overview of the Commission's AI Study
Artificial intelligence (AI) technology and applications AI technology and applications early in 2019. The
are certainly here in our everyday lives from recommendations in the Commission’s report are
smartphones to digital personal assistants. AI, however, intended to assist the Governor and Legislature in that
has the very real potential to become something more effort. In this appendix, are supporting materials.
than a fun and entertaining technology. AI may very well
become immersed in every major area of our work and
personal lives: transportation, communications, finance,
health care services, emergency response services,
community planning, criminal justice and entertainment.
The list goes on and is only constrained by our
imaginations. Right now, the imaginations of researchers,
business executives, academics, software engineers and
the like into the potential uses and applications of AI-
based technology appears limitless.
There is a fierce race underway by governments, large
and small, to prepare for smarter and smarter machines.
Where is California in that race? This is the question
the Commission sought to explore in this study. The
Commission researched, reviewed and analyzed hundreds
of articles and white papers on the development, use
and trends of AI technology and applications. The
Commission also held one formal hearing in January 2018
and two advisory committee meetings in May and August
2018. The Commission heard testimony and received
comments from industry experts, government officials,
academia, business leaders, technology associations and
other stakeholders.
The Commission found that while California state
government has made some strides to use modern
technology to improve internal and external operations
and services, the government as a whole is not AI savvy
or well prepared for the inevitable changes that will be
brought by smarter and smarter machines. California
state government lacks an infrastructure, AI strategic
planning and targeted measures geared toward taking
advantage of AI opportunities and minimizing AI risks.
Given the speed with which AI advances are being made,
the Commission urges the Governor and Legislature
to take appropriate action and measures specific to
Overview of the Commission's Study | 21
Sectors Impacted by AI
Experts predict that rapid progress in the field of
All Roads
automation relying upon artificial intelligence will
continue, penetrating and redefining the nature of work Lead to
and the workplace. Such revolution will bring profound
changes for the way private industry and government AI
operates. Its transformative potential already is evident
across a number of industry sectors from agriculture and
health care to education and transportation.
Agriculture
The increasing availability
of data is paving the way
for robots to harvest firm
fruits, for algorithms capable
of accurately detecting and
classifying plant pests and
disease and other technologies
that can predict crop yield.
In the future, artificial
intelligence may be applied
to biotechnology and plant Education
breeding to improve the
Intelligent tutoring systems
efficiency of crop production.51
and online learning will allow
teachers to instruct students in
different languages or drastically
increase class size to reach more
students efficiently. Artificial
intelligence may be useful in
filling the gaps in subject areas in
which a teacher does not have a
particular expertise.52
22 | Little Hoover Commission
Manufacturing
Amazon, a company known for
manning its warehouses with
robots, is now using machines to
make critical inventory decisions,
rather than tasking employees
with decisions that can be
automated.55 Future applications
of artificial intelligence in
factories have the potential
to create new manufacturing
markets, improve workplace
safety and more accurately
Transportation
project market demand.56
Artificial intelligence has
spurred GPS-enabled smart
cars and enabled the adoption
of smartphone technology
that provides real-time sensing
and prediction of traffic, route
calculations and ride-sharing.
Health Care
Artificial intelligence has
the potential to help people
Energy
manage their own health care
Machine learning and analysis of
and maintain health through a
electricity-consumption patterns
healthier lifestyle. For example,
are helping to make informed,
the Smart belt has a built-in
real-time decisions about
magnetic sensor that tracks waist
maximizing energy efficiency.
size and tension to determine
Utilities can predict and meet
when users may have overeaten
constantly changing energy needs
and, if so, offers an alert.54
by analyzing data across electrical
grids. Machine learning algorithms
can forecast the occurrence of
solar flares, which can disrupt
power grids and satellites.53
Sectors Impacted by AI | 23
How are Academia and Private Industry Using AI?
The uses of AI in academia and the private sector veterans, which make up roughly 3.5 percent of cancer
range from absolutely fun to profoundly life changing. patients in the U.S.61
There are so many AI applications being used today, this
study could not adequately address them all. Therefore, Microsoft is another leader in developing AI technology.
the following provides a glimpse into an ever growing In its 2017 vision statement, the company dropped
landscape where the seeds for imagination flourish. a reference to “mobile-first and cloud-first world” in
exchange for “an intelligent cloud and intelligent edge
There are important AI business uses that improve how infused with AI.”62 Within Office, Microsoft uses features
certain tasks are performed. For instance, Adobe Sensei driven by AI, such as live translation of recorded speech.
is described as using AI and machine learning to help In 2017, Microsoft acquired Maluuba, a Canadian
users discover hidden opportunities and make tedious technology company founded in 2011 by students at the
processes faster. Sensei, a word in Japanese meaning University of Waterloo. The company creates programs
teacher, is also described as having three types of that use natural-language processing, allowing computers
intelligence: creative, content and experience.57 Adobe to understand words and develop reasoning capabilities.63
is working on enhancements to Sensei, such as making
it easier for marketers to search for images in Adobe’s Digital Democracy is a project of the Institute for
stock service without needing to use words to describe Advanced Technology and Public Policy at the California
what they are looking for. In a function called selective Polytechnic University, San Luis Obispo. The free online
similarity, Sensei will allow users to mask areas of an platform uses AI to create a searchable database of all
image and have the system find the related photos or statements made in state legislative hearings, such as
illustrations.58 in California and Florida, and can be used by citizens,
journalists, organizations and researchers. Digital
IBM is on the move with its Watson AI technology in Democracy goes beyond providing information, allowing
ways that promote business, fun and health. Last year, the use and sharing of information on media platforms
they announced a partnership with Rocket Fuel to embed like Facebook and Twitter. “Digital Democracy empowers
IBM’s Watson AI technology into Rocket Fuel’s systems to every Californian with unprecedented access to their
identify key words and sentiments that might help predict government. The result will be greater accountability.”64
the interest of consumers in certain advertisements.59
Last football season, IBM introduced IBM Insights, which One Concern is an AI company with one very important
is a tool its Watson AI technology uses to scan and parse concern: saving the planet from natural disasters. “By
the web for fantasy football advice on a weekly basis. 2030, 60 percent of the world’s population will live in
Watson can assist with deciding which quarterback to cities with 1.4 billion facing the highest risk of exposure
play or which flex player to start.60 Watson is not all fun to a natural disaster.” The company’s vision is to ensure
and games, however. In 2016, IBM showed that Watson “that everyone lives in a safe, sustainable, and equitable
may have a heart. As part of former Vice President Joe world.” The mission of the company may be even bigger:
Biden’s National Cancer Moonshot Initiative, IBM began “drive deep social impact through benevolent intelligence
collaborating with the U.S. Department of Veterans to save lives and livelihoods.” This lofty goal is not
Affairs aimed at providing access to precision treatment surprising given that CEO Ahmad Wani survived being
options for over 10,000 patients fighting cancer. marooned during the 2014 Kashmir flood for seven days,
Oncologists have used Watson for genomics to identify wondering if he would live or die.
mutations in cancer cells DNA for more than 2,700
24 | Little Hoover Commission
One Concern’s AI products allow governments and
other customers to predict and react to the impacts
of natural disasters and conduct multi-hazard analysis
of infrastructures.65 In addition, the company partners
with emergency operations centers in San Francisco,
Los Angeles and Woodside to provide instant
recommendations on response priorities and resource
allocation.66
Dutch agritech company Connecterra uses Google’s
TensorFlow, an open source machine learning framework,
to help farmers keep their cows healthy. Many call it
“Fitbit for cows.”67
How are Academia and Private Industry Using AI? | 25
How Can AI be Used for Public Service?
Although the risks and complexities of artificial intelligence must always be measured and considered, the question of
AI’s potential to do good for the economy, society and the environment, may be summed up in one word: limitless.
Artificial intelligence could improve substantially lives, jobs, businesses, health care, government services, education
and much more. This section does not attempt to explore all those ways. Instead, the focus is on a sampling of services
critical to California’s well-being and future.
Fighting Fires
Machine learning
algorithms can supply fire
crews with data-driven
predictions on when
and where forest fires
will occur.68 Surveillance Education
drones powered by
AI systems allowed
artificial intelligence can
school administrators
scan and assess images of
in Tacoma, Washington
the ground to accurately detect an early-stage wildfire.69
to identify at-risk
An AI tool developed by NASA can guide firefighters
students and intervene
by tracking their movements, identifying hazards and
to provide assistance.
facilitating communication.70
Within a few years, the
graduation rate was
raised by nearly 30
percent.71
Emergency Response
AI can assist government
agencies and scientists
to monitor tremors,
floods and other
Environment
possible natural hazards.
AI can also process AI could promote
and fast-track calls biodiversity and water
to dispatchers, while conservation, and protect
filtering out redundant endangered species.73
or less urgent requests.72 Machine learning can
identify the presence
of invasive species and
diseases in plants and
eliminate them.74
26 | Little Hoover Commission
HIV Prevention Among Homeless Youth
The USC Center for
Artificial Intelligence
in Society uses an AI
algorithm to select
homeless youth who
have the potential to be
influential leaders. They
are trained to deliver
and promote HIV prevention information. Compared
to traditional methods, the approach has resulted in Senior Citizens
homeless youth being 2.2 times more likely to have taken
an HIV test within the past 6 months.75
AI systems can detect when
an elderly person falls and
may alleviate caregiver
shortages for seniors with
dementia, Alzheimer’s
disease and depression.76
An AI-programmed robot
developed at the University
of Denver can hold conversations with patients, offer
medication reminders and lead light physical activity and
Health Care
cognitive games.77 By 2030, one in five Americans will
be of retirement age; in California, the number of adults
AI algorithms are already
65 years or older is projected to increase by nearly 60
augmenting human
percent between 2017 and 2030.78
experts in interpreting
x-rays, detecting
diseases and cancers
and suggesting more
finely-tuned treatments.
Mammograms can be
reviewed and translated 30 times faster with 99 percent
accuracy, reducing the need for unnecessary biopsies
and increasing the chances of early detection.79 AI
applications could save $150 billion annually for the
United States health care economy by 2026.80 Persons with Disabilities
AI systems that
understand images,
sounds and languages
can support people with
deafness and autism.81
How Can AI be Used for Public Service? | 27
What Risks Does an AI World Bring?
Artificial intelligence can offer significant quality of women and minority groups seldom had the opportunity
life improvements, but the risks and benefits of this to become CEOs.86
technology may not be evenly distributed across society.
AI systems may incorrectly label women as worse hires In addition, when designing AI systems, computer
then men. Workers in low-skill occupations may find their engineers may pass along human prejudices, assumptions
jobs replaced by a machine. If not properly addressed, and preconceived notions inadvertently or purposely.
artificial intelligence can widen inequality and violate For instance, the term “receptionist” could be linked to
human rights. This section discusses the legal, social and “female,” thus embedding a stereotype in the program.87
economic risks an AI world could bring. An algorithm to assess loan applicants may be designed
to consider both income levels and past repayments with
Bias and Discrimination: Isn’t This Only a heavier emphasis on income. Slanting the algorithm in
Human Problem? this way, while even seemingly justified, may discriminate
against people who tend to be in lower-income brackets,
Machine Bias and How it Can Lead to Unlawful such as women.88
Discrimination. Artificial intelligence empowers machines
to think like people. However, machines that utilize Machine bias can emerge “I want to see
artificial intelligence are prone to incorporating the biases because AI researchers, stories about
contained in data and of their human engineers. Machine engineers and developers people getting
bias occurs when different groups of people are treated are mostly male, generally hired at 55 or in
unequally by machines on the basis of their gender, race highly paid and technically
their late 40s. I
and other characteristics.82 For example, an AI algorithm educated. Their interests,
don’t see enough
that is taught to diagnose skin cancer using images of perspectives and life
of that in Silicon
white faces means the system will perform best for experiences are naturally
Valley.”91
Caucasians as opposed to African-Americans.83 Machine reflected in the technology
bias could lead, whether intentionally or unintentionally, they create and advance.89
to unequal and unfair treatment for certain socio- Theirs, however, is not the Foaad Khosmood,
economic, racial, religious and other similar groups. It only world view. Absent a rich senior research
may identify people of color as being at higher risk for diversity among AI innovators, fellow, Institute
committing a crime or systematically exclude people with the technology could fail to for Advanced
mental disabilities from being hired.84 Machine bias may achieve the economic, social
Technology and
perpetuate discriminatory outcomes and prevent society and environmental benefits it
Public Policy
from gaining the benefits of AI-powered technology. otherwise could, and, worst
case scenario, AI could end up
The Ways Machine Bias is Created. Data is the life blood causing more harm than good.90
of artificial intelligence, because to make decisions
and predications AI technology uses data. Therefore, The Implications of Machine Bias. Racial discrimination
artificial intelligence is only as good as the data it receives can live on AI platforms.92 A 2016 ProPublica investigation
and learns from.85 Data, however, may be incomplete into Florida’s criminal justice system found racial
or skewed. For instance, an AI algorithm could scan bias within an AI algorithm used to assess the risk of
historical data and conclude that white men are the most recidivism. The algorithm, known as the Correctional
likely to succeed as CEOs, because, until very recently, Offender Management Profiling for Alternative Sanctions
(COMPAS), was twice as likely to flag black defendants
2288 || LLiittttllee HHoooovveerr CCoommmmiissssiioonn
mistakenly as being at higher risk to commit future explain a decision, prediction or forecast in ways humans
crimes. COMPAS was also twice as likely to flag white can understand.101
defendants incorrectly as low risk re-offenders.93 COMPAS
is used by the California Department of Corrections Ryan Harkins, director of state affairs and public policy
and Rehabilitation and the Board of Parole Hearings. at Microsoft, told Commissioners that computer
ProPublica, however, did not evaluate California’s use of programmers can reveal the conclusions determined
the system.94 by algorithms used by AI machines, but they cannot
always explain how the judgment of the machine is
Machine bias could also amplify harmful stereotypes. shaped. To encourage transparency, the government
Examples include a camera that cautioned against could protect the propriety rights of companies that
taking the picture of a Taiwanese-American, because publically share the assumptions contained within the
the software erroneously believed she was blinking.95 data and algorithms used by AI machines. In addition, the
Google’s image labeling technology resulted in racist information required to be disclosed could include target
miscategorizations of two black people.96 AI algorithms variables and intended outcomes, such as whether a loan
do not require explicit data on race, gender or applicant has credit worthiness.102
socioeconomic status to exhibit bias. Commercial lending
and insurance algorithms, for example, may not ask direct Privacy: Does Machine Learning Have a
questions about race or class identity, but these tools Nosy Side?
frequently incorporate other variables like ZIP code that
would count against those living in poor communities.97 Users of AI systems may not fully understand how AI
AI systems might not work well for all populations and technology can potentially, and perhaps unintentionally,
can reflect prejudices deeply embedded in history intrude into areas of their life they believed were private.
and society. A woman in Oregon recently discovered that her Amazon
Alexa had recorded a conversation between her and her
Actions by Private Industry to Avoid Machine Bias. State husband without their permission or awareness. Alexa
and federal governments collect and possess immense sent the audio recording to a random person on their
data sets, including demographics, housing, migration, contact list. Amazon later acknowledged the incident, but
transportation, economics and retail. Representatives maintained that Alexa had been given verbal instructions
from technology companies have been lobbying state to send the recording.103
and federal governments to release more government-
collected data, arguing that to minimize and avoid bias In 2015, Google’s AI subsidiary, DeepMind, was found
in AI applications the use of broader and more reliable to have violated the data privacy laws within the United
data sets is needed.98 Otherwise, there is a risk that AI Kingdom. While developing a health care app, DeepMind
applications will have an overreliance on data collected acquired large amounts of sensitive public health data
from social media and apps. Data mined from such from United Kingdom hospitals, but patients were not
limited sources may result in lopsided benefits for asked if they consented to having their medical data
socioeconomically advantaged populations, since those processed by DeepMind.104
tend to have greater access to such devices
and services.99 Users of social media may not be aware of how law
enforcement can use AI to track their social media usage
Where AI applications are used to replace human and draw conclusions, correct or incorrect, about them.
decision making, many experts advocate for increased At a recent Assembly Joint Informational Hearing, Matt
transparency in the design of AI algorithms in order Cagle, a technology and civil liberties attorney for the
to determine and understand whether the basis of American Civil Liberties Union, described how automated,
the decision was fair, objective evidence-based and social media surveillance software employed by
well-reasoned.100 As machines continue to help with California law enforcement agencies led to discriminatory
decision making, some researchers have emphasized targeting. Fresno police used a product called Beware
the importance of developing explainable AI. The goal of that assigned color threat levels to residents based on
explainable AI is to ensure that machines can accurately their social media posts and other commercial records.
What Risks Does an AI World Bring? | 29
Law enforcement in San impact aspects of human life in both innocuous and
“There is a lot to Jose used social media substantial ways. Facebook applies facial recognition to
be gained by just surveillance to oversee suggest how to tag photos, and Apple allows a user to
having people Asian, Muslim and Sikh unlock their phone by verifying the user’s face. Smart
walk through protestors. Mr. Cagle TV’s and advertising firms use facial recognition to learn
the potential warned that these types of how to target consumers with specified products for
unintended AI systems could be used future consumption. Through facial recognition, churches
consequences to monitor and deport can track membership attendance and retailers can spot
immigrant communities. shoplifters entering a store.
(of AI-powered
He emphasized the need
technology).”106
for greater transparency, Despite its broad use, there is no national policy
accountability and regarding facial recognition.108 Several efforts at the
Jennifer King, director
oversight.105 federal level have led to mixed results. In 2014, the
of consumer privacy,
Department of Commerce’s National Telecommunication
Stanford University
Private Industry Seeks a and Information Administration led stakeholder meetings
Center for Internet
National Framework to regarding industry uses of facial recognition with the
and Society
Avoid Privacy Concerns. aim of creating voluntary privacy standards to protect
Private companies are consumers. The meetings brought industry and consumer
lobbying federal officials to privacy advocates together. Although the group issued
adopt a new national privacy law. The law would replace a best practices document, privacy advocates dropped
California’s privacy standards with rules that would give out of the meetings due to disagreement over what
private companies greater leeway over how personal constitutes adequate protections.109 A 2012 Federal Trade
digital information is handled. While other specifics of Commission (FTC) staff report included recommendations
the proposal are unclear, the primary goal is to create a that private industry design software using facial
single national framework for protecting sensitive data.107 recognition with consumer privacy in mind, develop
reasonable security protections for collected information
Facial Recognition: Can You Control Who and ensure sound methods for keeping and destroying
Uses Your Face? collected information.110
Facial recognition utilizes images and algorithms to Mired with controversy over the use of its facial
ascertain or verify the identity of an individual. A facial recognition system for surveillance by law enforcement,
recognition system measures the distinctive facial Amazon in July 2018 weighed in on government’s role
features in an image, transforms that information into in regulating the technology. After explaining that all
algorithms representing that person’s face and compares of Amazon’s customers agree not to engage in illegal
that to other faces already in the system’s repository. activity, the company stated that, “It is a very reasonable
Uses of facial recognition span various industries and idea, however, for the government to weigh in and
specify what temperature (or confidence levels) it wants
law enforcement agencies to meet to assist in their public
safety work.”111
To minimize machine bias, the collection of
Illinois, Texas and Washington are the only states with
demographic and socioeconomic data should
laws on facial recognition that limit its use to a certain
be meaningful, relevant and diverse, while
degree.112 Facebook’s tag feature is currently the subject
ensuring that sensitive data is protected
of a lawsuit in Illinois based on the state’s Biometric
against unauthorized access and disclosure.
Information Privacy Act. The law protects against the
Processes for detecting, correcting or
use of facial technology without informed consent by
removing corrupt or inaccurate records from
the consumer. The case will decide whether a user’s
record sets should also be implemented.
acceptance of Facebook’s terms and conditions signifies
consent.113 Local governments have moved toward
30 | Little Hoover Commission
limiting the use of facial recognition technology as it
relates to surveillance.114
In July 2018, Microsoft’s President Brad Smith penned
a blog calling for government regulation of facial
recognition technology. Mr. Smith issued a call for an
expert bipartisan commission that could consider issues
regarding facial recognition’s use by law enforcement,
minimum performance levels of accuracy and consent of
its use.115
What Risks Does an AI World Bring? | 31
Anticipated Impacts of AI on Employment
Recent research by the global management consulting the impact of automation on the nature of work–self-
firm, McKinsey and Company, found that the checkout machines in grocery stores or ATM machines
primary driver of workforce transformation would in banks–its projections do not account for significant,
be the displacement of job activities and tasks due long-term changes in the labor force due to smarter and
to automation.116 After reviewing 800 occupations, smarter machines.119
McKinsey estimated that at least one-third of work
activities could be automated in 60 percent of The Employment Development Department (EDD)
occupations. Collection and processing of data, and the is the state’s equivalent of BLS. The EDD publishes
performance of predictable physical tasks like operating short-term (2-year) and long-term (10-year) regional
machinery or preparing fast food, are tasks more likely to and state labor projections to identify labor trends.120
be replaced by automation. Jobs involving large amounts On an annual basis, EDD submits statistical reports to
of labor, such as paralegal work and accounting could be BLS. While BLS does not have stringent methodology
impacted by automation, while jobs involving managing requirements for the reports, EDD uses BLS software,
people, applying expertise or interacting socially will be and thus BLS methodology and assumptions, for the
less subject to automation.117 reports. Accordingly, EDD’s analysis of data and forecasts
of economic trends are intertwined with the arguably
Although it is certain that automation and AI-based faulty presumptions embedded in the BLS system. That
technology will change the labor market, it is less problem alone is cause for worry, but there is more.
clear how and when those changes will occur. Some EDD gathers some of its data from voluntary employer
expert witnesses told the Commission they anticipated surveys; yet, many tech firms decline to respond. This
that new technologies would have a serious and lack of participation means that EDD’s projections for
detrimental impact on jobs while others were more future labor force needs may be inaccurate.121
optimistic. Regardless, there was general consensus
that policymakers can positively influence the future of There is not a separate unit within EDD that conducts
employment, but only if they take appropriate action data analysis or produces reports on why certain
and soon. occupations in California have changed.122 It should be
noted the employment security department in the state
California: What the Data Shows and Does of Washington designed its own economic modeling
Not Show instead of using the prepackaged BLS software. The
department strives to use methodology that will best
The federal Bureau of Labor Statistics (BLS) publishes capture labor trends contained within specific industries,
10-year estimates of national employment by industry while still meeting BLS reporting requirements.123
and occupation by analyzing current and historical
economic data. The bureau’s projections rely on a Labor Markets: California May Find
series of assumptions to create a 10-year estimation of
Clues There
future jobs; however, the Commission learned that BLS
assumptions are not necessarily consistent with a real-
To date, much attention has been given to the threat
world economy. The BLS model assumes there will be no
of job losses stemming from an increase in automation
changes in the future business cycle; broad social and
and other AI-based technologies. A 2013 study out of
demographic trends will continue; and the supply and
Oxford University predicted about 47 percent of U.S.
demand of labor will remain in equilibrium.118 Though
employment is at risk due to these technologies.124
labor market information analyzed by BLS accounts for
32 | Little Hoover Commission
Software entrepreneur and author of Rise of the Robots, of demographic and policy demands, which may help
Martin Ford predicts a jobless future where machines to counterbalance job displacement. For example,
take over more and more of the tasks that comprise investments in renewable energy and mitigation of
typical jobs.125 Microsoft co-founder Bill Gates believes climate change may create demand for workers across
job loss will be devastating and suggests a robot tax to several sectors, from manufacturing and construction to
modulate the impact.126 Elon Musk, CEO of Tesla, mirrors installation. Similarly, growth in the population of people
those concerns and advocates for a universal basic age 65 and older may create demand for occupations in
income.127 the health care sector, including doctors, nurses, home
health aides and nursing assistants.134
Yet, many experts predict moderate job loss. The
Information Technology and Innovation Foundation As new AI technologies proliferate the workplace, it is
estimates that about eight percent of jobs are at high likely that more people will find themselves working
risk of automation by 2024.128 A 2017 study by McKinsey alongside smart machines. Robots are transforming a
and Company estimated that automation could result number of industries, particularly in industrial settings.135
in the loss of 39 million jobs and displacement of Cobots, or collaborative machines that work alongside
approximately 23 percent of workers in the United States human operators, are able to perform specific routine
by 2030.129 In addition, the Information Technology and tasks, thus freeing up employees to perform more
Innovation Foundation contends that the 2013 study by sophisticated work.136 Some predict that demand for
the University of Oxford is flawed and said it is “alarmist” cobots will increase in industries facing employee
to believe the world’s economies face a tsunami of retirements, like agriculture, manufacturing and
job destruction.130 Governing Magazine noted in 2017 construction.137
that “the biggest unknown at this point is whether
automation will eliminate more jobs than it creates.”131 Some economists, though, are sounding warning
alarms for wages. They predict that technology-driven
Some economists believe labor market changes will hollow out the middle class,
that technological eliminating middle-class jobs such as bookkeeping,
innovations will create clerical work and production. The consequence will be
“Today we are in the
more jobs. They point job polarization where, on one end, there is an increase
midst of just such
to historical examples of in demand for lower-paid, low-skill jobs and, on the other
a panic cycle about
new technologies that end, an increase in demand for higher-paid, high-skill
artificial intelligence
increased the demand jobs aided by computers.138 These economists argue that
(AI) , and much of the
for workers to focus job polarization is already happening and has led to the
panic swirls around
on tasks that machines stagnation of median wages.139
its potential impact
could not. An often cited
on jobs, inequality
example is the advent of What the Commission Heard: Optimism
and other economic
the ATM machine, which
and Concern
performs routine banking outcomes.”132
tasks for customers, thus
In his written testimony to the Commission, Robert
permitting bank tellers to
Robert Atkinson, Atkinson, president of the Information Technology and
provide other, less routine,
president, Information Innovation Foundation, provided several examples of how
banking services to
Technology and states can and should do a better job of enabling workers
customers. An economist
Innovation Foundation to get “better” skills. States can partner with nonprofit
at the Boston University
organizations to establish better online portals for access
School of Law found that
to skill assessments, training resources and job search,
between 1982 and 2012
such as the Council for Adult and Experiential Learning
employment grew faster in occupations that made more
(CAEL). California also can actively and clearly notify
use of computers, thus allowing employees to focus
workers once they apply for unemployment insurance
on other aspects of their work.133 McKinsey predicts
that they qualify for unemployment insurance benefits
that new occupations will be created as a consequence
Anticipated Impacts of AI on Employment | 33
if they are in approved training. “When worker skills are Ms. Bernhardt told Commissioners policymakers now
more developed, worker adjustment from dislocation have a range of options to shape the speed, direction and
becomes easier,” Mr. Atkinson wrote.140 scope of transformation. She points out that technical
feasibility of automation is not the same as adoption
At the Commission’s May 2018 advisory meeting, and implementation of new technologies. “A wide range
several labor representatives from the California Nurses of factors (economic, social, political, organizational)
Association, Teamsters, United Food and Commercial drive the scope and pace of technology adoption, which
Workers (UFCW) and Service Employees International is often slower and less complete than one would
Union (SEIU) California discussed the potential negative predict based on feasibility alone,” she testified at the
impacts of AI-powered technology. For example, Tia Orr, Commission’s January 2018 hearing. “The lesson from
SEIU director of government affairs, discussed the failed past technological transformation is that public policy
launch of a new Medi-Cal eligibility system, which was plays a significant role in determining the trajectory of
designed to automate enrollment for beneficiaries. On adoption and impact.” She concluded that large-scale
the first night of rollout, the system crashed and humans automation may not occur in the United States in the
were needed to intervene and contact customers. “There near or medium term, especially if public policy prioritizes
is no incentive right now for good acting,” Ms. Orr said. investments that support economic growth, innovation,
“There does have to be a conversation on what we’re job quality and workforce transition.145 Ms. Bernhardt
willing to automate. We have to ask ourselves: at what also told the Commission:
point do we be bring in the human?”141
All of the attention to date has been on
Annette Bernhardt, director of the low-wage work automation. But new technologies have many
program, U.C. Berkeley Center for Labor Research and other direct effects on tasks–deskilling or
Education, told Commissioners that wages, benefits, and upskilling existing ones, creating new ones–
hours worked all may be impacted by the adoption of as well as a slew of indirect effects, such as
AI technologies. Specifically, she testified that “workers enabling outsourcing and changing the job
of color, immigrant workers, and women are dependent matching process. These are all important
upon the type of routine/manual occupations that effects for policymakers to understand. In
are at high risk of automation, or at risk of significant particular, it is likely that incremental changes
upskilling (requiring more education) or deskilling in task content and skill requirements will affect
(resulting in lower wages).” New technologies are likely more workers in the coming years than large-
to have significant effects on the distribution of jobs by scale automation events.146
race, gender, immigration status, education level, and
geography, Ms. Bernhardt said.142
California labor representatives are concerned that “Some workers will have to be retrained to
job displacement due to automation and other AI work alongside AI directed machines, while
technologies may exacerbate income inequality.143 “We others will have to be redeployed within
really need to look at the quality of jobs,” California Labor the company or elsewhere in the economy;
Federation Policy Coordinator Sara Flocks told legislators businesses have a vital role to play in aiding
in a March 2018 hearing on the promises and challenges
these transitions. This will require changes in
of AI for California. “We see this [problem of inequality]
skills, mindsets and culture as we transition
when you get rid of a UPS trucking job or you automate a
into a world where ‘coworkers’ include
trucker’s job and then they go be an Amazon gig worker.
machines as well as other people.”147
You’re going to take a good union job and all of a sudden
it’s a low-wage, temporary, part-time, seasonal job. We’re
McKinsey Global Institute
not going to be able to rebuild our middle class with
that.”144
34 | Little Hoover Commission
The Real Story on AI’s Impact: There is
Disagreement, Except on the Need for
Policy Action
As discussed above, there is no consensus on how and
when automation and, in particular AI technology and
applications, will change the workforce. There is also
disagreement on what that change will look like.
“The question is: can we actually transition
This disagreement may be best demonstrated in
people from what they’re doing now to what
discussions concerning driverless cars. Researchers
they will need to do in the future. This will
with the Economic Policy Institute contend that there
affect all of us. All of us will have a significant
is no proof automation has caused wage inequality or
number of activities that machines will be able
stagnation and dispute its impact on overall joblessness.
to do. Then the question is not about mass
Instead of making workers obsolete, they believe “rapid
unemployment but mass redeployment.”150
technological advances are associated with better, not
worse, economic outcomes for working Americans.”148
Michael Chui, partner, McKinsey Global Institute
Most recently, a September 2018 report from the U.C.
Berkeley Center for Labor Research and Education and
Working Partnerships U.S.A. explored how self-driving
technology could transform the trucking industry.
Researchers predicted autonomous trucks could replace
as many as 294,000 non-specialized long-distance drivers,
eliminating both some of the highest-paid career trucking
positions and many more moderate-wage driving jobs.
Those researchers, however, also predicted that
automation could create many new driving jobs that
require humans to navigate local streets and handle non-
driving tasks. Local and delivery drivers currently receive
low wages and often have poor working conditions. The
U.C. Berkeley researchers envisioned a new scenario
where autonomous trucks could almost constantly be on
the highway and human drivers could work locally and
go home each night. This strategy could save the trucking
industry billions each year.
The U.C. Berkeley researchers found that proactive
public policy is needed to ensure that the benefits of
automation are spread in the transportation industry.
They suggest policymakers could design job-quality
standards to ensure living wages and good working
conditions for drivers and create regulations to incentivize
clean trucks to reduce pollution and other environmental
and public health costs.149 Of course, those are only
suggestions that may never occur without lawmakers
taking action.
Anticipated Impacts of AI on Employment | 35
Are Californians Being Prepared for an AI World?
California has three public university systems: the to address them in new ways and yet provide concrete
community colleges, state universities and the strategies for tackling these problems.”156
University of California. The Master Plan for higher
education in California, which was originally adopted While the California Community College System (CCCS)
in 1960, provides in relevant part that the community has colleges that offer computer science associate
colleges will offer lower-division instruction that is degrees, none are specific to artificial intelligence. In
transferable to four-year colleges, provide remedial and addition, the CCCS does not have any program similar or
vocational training and grant associate degrees and equivalent to CAIS or the Center for Human-Compatible
certificates. The University of California is to serve as Artificial Intelligence. The California Department of
the state’s primary public research university and grant Education has adopted the state’s first-ever computer
baccalaureate, master’s, doctoral and other professional science standards for K-12 grades, and those standards
degrees. The California State University (CSU) is to include instructional content related to artificial
focus on instruction in the liberal arts and sciences and intelligence for grades 9 through 12.
grant baccalaureate and master’s degrees. Subsequent
legislation, however, has permitted CSU to offer The California Department of Technology offers a nine-
doctorates under limited circumstances.151 week training program for state employees working in
the information technology field that includes such topics
In addition, Senate Bill 850 (Block), which was approved as data analytics and representation. In addition, the
into law in September 2014, established a statewide Governor’s Office of Business and Economic Development
baccalaureate degree pilot program at not more than (GO-Biz) sponsors a cyber-challenge event that provides
15 community college districts with one baccalaureate cybersecurity education and training for young people,
degree program each, to be determined by the but there is no event specific to artificial intelligence.
Chancellor and approved by the Board of Governors.152
Participating community college districts must meet
specified requirements, including, but not limited to,
offering baccalaureate degree programs and program
curricula not offered by the California State University
or the University of California, and in subject areas with The Bureau of Labor Statistics (BLS) estimates
unmet workforce needs, as identified and documented that the number of computer and information
by the district. The law by its own terms becomes research science positions will increase by
inoperative on July 1, 2023.153 19 percent between 2016 and 2026.157 The
California job market for computer and
In 2016, the University of California, Berkeley opened the information research scientists is projected
Center for Human-Compatible Artificial Intelligence.154 to increase by 22 percent between 2016 and
The primary focus of the new center is to ensure that 2026.158
AI systems are beneficial to humans.155 In 2016, the
University of Southern California, a private school,
created the Center on Artificial Intelligence for Society
or CAIS. Milind Tambe, the co-director of the center,
said, “Wicked social problems such as homelessness are
incredibly complicated. AI provides us the opportunity
36 | Little Hoover Commission
California Community Colleges deliverance systems
to the instructional
The CCCS is a postsecondary education system and space, major 7,081,091 assignable
includes the Board of Governors of the California renovations will be square feet is
Community Colleges, the Chancellor’s Office, 72 required.”The five- needed at a cost
community college districts and 114 community year plan includes of $10.9 billion
colleges.159 The CCCS serves more than 2.1 million 23.6 million assignable to accommodate
students. A seventeen-member Board of Governors, square feet to be current and future
appointed by the California Governor, oversees CCCS.160 modernized over the enrollment at CCCS.
The board exercises general supervision over the next five years at a The total unmet
system of community colleges and provides leadership cost of $11.5 billion needs and costs for
and direction through planning and policymaking.161 and relates only to facilities are $30.4
The Consultation Council, which is comprised of 18 buildings more than billion.168
representatives from stakeholder groups such as trustees, 40 years old and
students, administrators and business officers, meets reported as needing
once per month to review and evaluate policy proposals major renovation.167
and related issues impacting the CCCS.162
An opportunity thus exists to expand upon the CCCS
An important component to the CCCS success is computer sciences curriculum by including classes and
alternative methods of instruction, such as distance degrees specifically related to the development and use
learning. “Many districts are actively pursuing online of artificial intelligence and applications. This addition
courses as a method of instruction in order to provide would complement and expand upon the new computer
greater access for students as well as reducing the need science standards for grades K-12. Any such courses
for new facilities.”In the 2016-17 school year, there were should be taught within a framework that promotes AI
164,855 full-time equivalent students who took distance for economic, social and environmental good.
education, which is 14 percent of the total full-time
The Strong Workforce Program
equivalent students (1,183,240).163
In 2016, Assembly Bill 1602 (Committee on Budget)
Most, if not all, community colleges offer a degree or
created the Strong Workforce Program, which expands
certificate related to computer science, but there are
and improves career education, also known as career
some 50 subcategories related to computer science,
technical education or CTE, throughout the state with
ranging from programming to helpdesk support.164
the CCCS leading the effort.169 Year one served in large
The College of Marin and Butte College, for instance,
part as an implementation year, with colleges and
offer associate degrees in computer science. Both
regions planning, preparing and collaborating to begin
programs prepare students for fields as software or
laying the foundation for “more and better” career
hardware programmers.165 Associate degrees specific
education programs. The CCCS activities in the first year
to technologies driven by artificial intelligence are not
were many and included: increased collaboration and
offered, however, in the CCCS. In addition, CCCS does
visibility of career education programs, and greater
not have any program equivalent to CAIS or the Center
focus on programmatic decision-making. For year two,
for Human-Compatible Artificial Intelligence. In fact, a
the CCCS recommends continuing the development
search on the California Community Colleges website for
of regional structures, strengthening connections to
“artificial intelligence” revealed no results.166
partners and promoting the use of data for community
college leadership, administration and faculty to better
The 2019-20 Five-Year Capital Outlay Plan states, in
understand the areas in which they are succeeding
relevant part, “[D]ue to technological advances, California
and where they need improvement related to student
Community Colleges needs to incorporate more
outcomes.170
sophisticated technology into its facilities so the system
can deliver state-of-the-art instructional programs.
To make buildings smarter by providing cabling and
Are Californians Being Prepared for an AI World? | 37
The state’s Economic and Workforce Development defined as “the process of converting a program design
(EWD) Program supports the Strong Workforce into an accurate and detailed representation of that
Program by furnishing education, training and other program in a suitable language.”174 The operation of the
services that advance the economic growth and global law is subject to an appropriation in the annual Budget
competitiveness of California and its regional economies. Act or another statute.175
By providing grant awards, EWD provides the financial
support needed for the workforce development of In September 2016, the SBE began the process of
employers, workers and students. The majority of grants implementing this law into standards, and, on July 26,
go to the “ecosystem of intrapreneurs” who are sector 2018, the IQC and SSPI recommended to the SBE a draft
experts and navigators working to build or strengthen of the Computer Science Standards.176 On September
business and industry links with community colleges to 7, 2018, the SBE approved the standards, which are
ensure that community college programs move and form California’s first ever computer science standards.177
at the speed and need of business and industry.171 The proposed standards include the goal that second
graders will learn to use common computer hardware
The Strong Workforce Program investments for year one and software and be able to create simple computer
show that in 2016-17 funds were invested in a variety programs and debug errors in an algorithm. The
of sectors across the seven macro-regions. The seven standards for high school include more complicated
macro-regions were: Bay Area, Central/Mother Lode, coding and data analysis.178
Inland Empire/Desert, Los Angeles/Orange County,
North/Far North, San Diego/Imperial and South Central For grades 9 through 12, the draft standards include
Coast. The sectors included advanced manufacturing, standards related to describing how artificial intelligence
advanced transportation and renewables, agriculture, drives many software and physical systems. The following
water and environmental technologies, clean energy is the descriptive statement for the standard:
(Proposition 39), education, health, global trade and
logistics and public safety. Technologies in those sectors Artificial intelligence is a sub-discipline of
that use artificial intelligence were not expressly computer science that enables computers
identified.172 Accordingly, this gap presents an opportunity to solve problems previously handled
for lawmakers to ensure that the Strong Workforce by biological systems. There are many
Program is inclusive of businesses that develop and/or applications of artificial intelligence, including
use AI technology and applications. computer vision and speech recognition.
Students research and explain how artificial
California’s First-Ever Computer Science intelligence has been employed in a given
Standards system. Students are not expected to
implement an artificially intelligent system in
The State Board of Education (SBE) is required to adopt order to meet this standard.
statewide academically rigorous content standards,
pursuant to recommendations of the Commission for the For example, students could observe an
Establishment of Academic Content and Performance artificially intelligent system and notice
Standards, in core curriculum areas. On September where its behavior is not human-like, such
30, 2014, Governor Brown signed Assembly Bill 1539 as when a character in a videogame makes a
(Hagman) into law, which required the California mistake that a human is unlikely to make, or
Department of Education (CDE), Instructional Quality when a computer easily beats even the best
Commission (IQC), on or before July 31, 2019, to human players at a given game.
“consider existing computer science content standards,
which include, but are not limited to, the national K-12 Alternatively, students could interact with a
computer science content standards developed by the search engine asking various questions, and
Computer Science Teachers Association, and consider after reading articles on the topic, they could
content standards that include, but are not necessarily explain how the computer is able to respond
limited to, standards for teaching coding.”173 Coding is to queries.179
38 | Little Hoover Commission
The draft also includes standards related to the The California Computer Science Strategic
implementation of algorithms that use artificial
Implementation Plan Panel
intelligence to overcome simple challenges. The
descriptive statement provides: The California Computer Science Strategic
Implementation Plan Panel, which is within the CDE, is
Artificial intelligence algorithms allow recommending the state adopt a high school graduation
a computer to perceive and move in requirement and a teacher certification pathway for
the world, use knowledge, and engage computer science, and that it expand scholarship
in problem solving. Students create eligibility and loan forgiveness for computer science
a computational artifact that is able teachers in low-income districts. The Department of
to carry out a simple task commonly Education is expected to turn those recommendations
performed by living organisms. into an implementation plan, which the board is
Students do not need to realistically scheduled to consider for adoption in March 2019.183
simulate human behavior or solve a
complex problem in order to meet this State Government
standard.
California Department of Technology. The California
For example, students could implement Department of Technology (CDT) offers a nine-week
an algorithm for playing tic-tac-toe that program, called the Digital Services Innovation Academy
would select an appropriate location for (DSIA), for state employees working in the information
the next move. technology field. The course is designed to introduce
students to the tools and skills required to transform a
Alternatively, students could implement business problem from concept to product using design
an algorithm that allows a solar- and development principles found in modern digital
powered robot to move to a sunny service development. Topics include data analytics and
location when its batteries are low.180 representation, UI/UX design, development methods,
ideation and communication. Students also do a digital
The computer-science standards are primarily technical services challenge assignment that provides them the
in nature and do not include courses on ethical opportunity to apply the concepts and knowledge they
considerations related to the development and use of AI have learned during the academy. The curriculum,
or how AI can, and should, be used for economic, social however, does not expressly address the use or
and environmental good.181 This is an educational gap implementation of technologies driven by artificial
that should be remedied in future legislation. intelligence.184
The Governor’s Office of Business and Economic
Development. The effects of emerging technologies on
the labor market require businesses and government
to consider how to prepare the next generation
Other states have folded computer science of workers. The Governor’s Office of Business and
into the K-12 curriculum, including Ohio, Economic Development (GO-Biz) sponsors the California
Virginia, Utah and Wisconsin. Virginia was Cyber Innovation Challenge (CCIC) as part of an effort
the first.182 to prepare young people for the needs of future
jobs involving cybersecurity. The CCIC is a statewide
competition for high school students allowing them
to demonstrate their skills and learn about careers in
cybersecurity.185
Are Californians Being Prepared for an AI World? | 39
Students reached the CCIC via an at-large application
process or through their success at regional competitions
taking place in Los Angeles, Sacramento, San Diego and
the Bay Area. The number of teams in the CCIC has
grown each year from eight to 20, comprised of up to
five students each. The CCIC includes timed cybersecurity
challenges intended to replicate real world scenarios.
The CCIC received financial support from the California
Workforce Development Board and relied on
partnerships with the California Community Colleges
Chancellor’s Office, the Governor’s Office of Emergency
Services, the Department of Technology, the California
Military Department and the California Department of
Justice.186 GO-Biz also teamed up with the CyberCalifornia
Coalition—made up of businesses, academic institutions
and state agencies—since the competition’s inception
in 2016. Among its various goals, CyberCalifornia strives
to strengthen the state’s leadership in cybersecurity and
connect the state’s workforce development system with
the needs of employers.187 GO-Biz also collaborates with
institutions of higher learning to serve as hosts, which
thus far have included Sacramento City College and
California Polytechnic State University.
The CCIC represents how industry, government and
academia can work together to encourage the workforce
of the future to enter careers in emerging technologies.
The competition exposes young students to critical
training that will help meet the demands of the changing
labor market.
40 | Little Hoover Commission
Are California State Agencies Preparing for AI?
In March 2018 lawmakers held two informational GO-Biz works “We want to hear
with 15 regional
hearings to consider how artificial intelligence from the private
innovation hubs
technologies may change California’s economic and social
sector about
across the state
landscape. The first focused on automation applications
how we can
to stimulate
in government and other settings, and included
help pioneers be
partnerships, economic
discussions around balancing convenience, bias and
successful.”192
development and job creation
privacy, as well as implications of AI on the workforce and
public policy.188 The second focused on ways technology around specific research
clusters, including AI.191 Part Sid Voorakkara,
and automation may impact worker rights and job
quality.189 of the office’s role is to bring deputy director
industry entrepreneurs to of external affairs,
Sacramento to meet with state Governor’s Office
In July 2018, the Little Hoover Commission surveyed
representatives from 84 California state agencies about regulators and lawmakers to of Business
the applications and implications of artificial intelligence help inform policy decisions and Economic
in their organizations. The Commission received around the opportunities of Development
using new technologies.
25 responses. The majority of survey respondents
(80 percent) reported that their department is not
The Department of Technology is beginning
currently using AI, although nearly 23 percent said
to consider ways AI can make it easier for
their department intends to use AI within the next
citizens to communicate with government
decade. Several said their department is currently using
and conduct online research by, for
technology to automate tasks. The Public Employment
instance, embedding AI chatbots on CA.gov
Relations Board uses AI technology to automatically
websites and expanding the department’s open data
process templates for notification, the Office of
platform.193
Administrative Law uses it to calendar due dates of
assignments based on statutory requirements and the
Office of the Secretary of State uses it to operate a The Department of Food and Agriculture
chatbot, a program built to automatically engage with is involved in burgeoning efforts led by the
received messages. University of California Agricultural and
Natural Resources to use a broad range of
The California Department of Corrections agricultural technology to solve some of the
and Rehabilitation and the Board of Parole major agricultural challenges facing the state, including
Hearings, although not responding to the climate change, drought and invasive species.
survey, use AI technology to operate the
Correctional Offender Management Profiling
for Alternative Sanctions (COMPAS) tool to determine if The February 2018 California Public Sector
an inmate should be paroled.190 CIO Academy sponsored by Government
Technology featured a panel discussion
on intelligent and cognitive automation,
including how government technology
leaders can incorporate a range of solutions
into their work, from simple bots to complex
machine learning.194
Are California State Agencies Preparing for AI? | 41
Are Lawmakers Preparing for AI?
During the 2017-2018 session, Assemblymember personal information that the business collects about the
Kiley, with several co-authors, introduced Assembly consumer; (2) the categories of sources from which the
Concurrent Resolution 215. This measure passed both information is collected; (3) the purposes for collecting
Houses and was filed with the Secretary of State on or selling the information; and (4) the categories of third
September 7, 2018. It expressed the support of the parties that have received the information.
Legislature for the 23 Asilomar AI Principles, as guiding
values for the development of artificial intelligence and Governor Brown’s 2018-2019 budget did not expressly
of related public policy. These principles were developed reference the impact AI-powered technologies will have
by leading AI researchers, economists, legal scholars, on the economy and jobs, but the budget establishes tax
ethicists and philosophers during a conference hosted credits, and training and education programs that provide
by the Future of Life Institute. Over 1,200 AI and robotics opportunities for Californians to transition smoothly into
researchers signed the principles, which are intended to an evolving labor market where smarter and smarter
promote the safe and beneficial development of artificial machines will exist. The budget expanded California’s
intelligence.195 Earned Income Tax Credit to cover individuals who are
between ages 18 and 24 and over age 65. The income
Senate Bill 1470 (Stern) would establish the Commission level at which the credit phases out completely was
on the Future of Work, which would consist of eight increased to $24,960.
members with knowledge of, and expertise in, the
following areas: work, workforce development, labor, The budget included a $120 million allocation to create an
technology, robotics or artificial intelligence. The bill was online community college to provide vocational training,
held in the Assembly Appropriations Committee. career advancement and credentialing opportunities to
fill current workforce gaps. In addition, $150 million was
Assembly Bill 2596 (Cooley) would declare that California assigned to build K-12 career technical education classes
needs a “triple-bottom-line” strategy that simultaneously as part of the Strong Workforce Program. The California
advances an inclusive economy through equity, prosperity Education Learning Lab was established to increase
and sustainability while responding to market forces by, learning outcomes and close achievement gaps through
among other things, encouraging the development within adaptive learning technologies in scientific, technology,
California of new technologies that yield more equitable, engineering and math (STEM) classes.
efficient, sustainable communities and affordable
qualities of life. The bill was vetoed by Governor Brown The California Apprenticeship Initiative New and
on September 25, 2018. In his veto message, Governor Innovative Grant Program, also created by Governor
Brown stated, “I don’t believe an ongoing costly study and Brown’s budget, funds opportunities in developing
report will provide any additional benefit to these efforts.” industry areas, such as advanced manufacturing, where
apprenticeships are either non-existent or not fully
On June 28, 2018, Assembly Bill 375 (Chau) was approved established. The budget allocated additional grant funding
by the Governor and filed with the Secretary of State. for public private partnerships to prepare students for
This bill declares, in relevant part, that people desire high skilled STEM careers and for the development of
privacy and more control over their information and they online programs that lead to short-term, industry-valued
should be able to exercise control over their personal certificates by community college districts.
information. Beginning January 1, 2020, the bill will grant
consumers the right to request a business to disclose
the following: (1) the categories and specific pieces of
42 | Little Hoover Commission
Resources
Top Reports 5. Oxford Martin School, September 2017, The
Future of Employment: How Susceptible Are Jobs
To Computerisation? This report examines the
1. Council of Economic Advisers, July 2018,
expected impacts of future computerisation on
Addressing America’s Reskilling Challenge. This
U.S. labor market outcomes. It concludes that
white paper emphasizes the importance of
about 47 percent of total employment in the U.S.
collaborations between employers, workers and
is at risk of computerisation.
educational institutions to address America’s
reskilling challenges.
6. Ryan Calo, August 2017, Artificial Intelligence
Policy: A Primer and Roadmap. The paper provides
2. Executive Office of the President, October 2016,
a way to understand the implications of AI on
Preparing for the Future of Artificial Intelligence. This
services, economy and law. It raises key policy
report surveys the current state of AI, its existing
questions while underscoring the many benefits
and potential applications and the questions that
AI will bring.
progress in AI will raise for society and public
policy. It also recommends specific actions,
including collaboration, development of best 7. Tata Communications, September 2018, Cognitive
practices in AI and retraining. Diversity: AI and the Future of Work. Based on
the opinions of 120 global business leaders, this
report envisions a positive impact of AI on the
3. Gallup Inc., January 2018, Optimism and Anxiety:
future workplace. Business leaders believe AI
Views on the Impact of Artificial Intelligence and
will diversify human thinking, enhance human
Higher Education Response. This paper describes
collaboration and help workers become more
key findings from an AI survey. The survey
agile, curious and nimble.
included questions regarding perceptions of
AI’s impact on people’s lives and their work or
education, and potential interventions from 8. Thomas Kalil, March 2017, AI for Good: Maximizing
higher education, government and private the Economic and Societal Benefits of AI. This
industry. white paper analyzes how government can set
ambitious goals to ensure the responsible and
beneficial development of AI. It provides specific
4. McKinsey Global Institute, December 2017, Jobs
areas where AI can achieve economic and
Lost, Jobs Gained: Workforce Transitions in a Time
societal benefits.
of Automation. This report summarizes previous
research completed by McKinsey on the future
workforce impacts of automation and AI and 9. U.S. House of Representatives, September
analyzes the factors that will lead to job loss or 2018, Rise of the Machines: Artificial Intelligence
creation. It predicts employment growth through and its Growing Impact on U.S. Policy. This paper
a series of economic models and assumptions. summarizes the main findings of hearings held
by the Subcommittee on Information Technology
of the House Committee on Oversight and
Government Reform, which includes testimony
from academia, industry and government.
Resources | 43
10. World Economic Forum, September 2018, The 5. Fei-Fei Li, Director, Stanford University Artificial
Future of Jobs Report 2018. This report examines Intelligence Lab. Ms. Li’s invented ImageNet and
the implications of technological change on the ImageNet Challenge, a large-scale dataset
the future of work. The report suggests that that has contributed to computer vision and
businesses take an active role in reskilling and deep learning in AI. She also directs the Stanford
upskilling workers. University Artificial Intelligence Lab, an academic
center that builds smart algorithms.
11. Google. AI for Social Good. This website
provides various examples of how AI can solve
6. Laura Tyson, Faculty Director, U.C. Berkeley
humanitarian and environmental challenges.
Institute for Business and Social Impact. Ms.
Tyson is an economist who has examined and
12. Markle Foundation, 2018, How California’s Next written about domestic and international policy
Governor Can Create Opportunities for Workers and matters, such as the impacts of automation.
Businesses. This report discusses opportunities for
Californians in an automated economy.
7. Lenny Mendonca, Co-Chair, California Forward.
Mr. Mendonca is the co-chair of California
Top Experts Forward, a nonprofit organization that aims to
reform policies regarding fiscal and government
accountability. As a senior partner emeritus
1. Annette Bernhardt, Senior Researcher, U.C.
at McKinsey & Company, he founded the
Berkeley Institute for Research on Labor and
organization’s U.S. state and local public sector
Employment. Ms. Bernhardt has collaborated
practice and helped dozens of government,
with union organizations to examine the impacts
corporate and nonprofit clients solve their most
of automation on jobs and studied policies to
difficult management challenges.
improve wages and working conditions at the
local, state and national levels.
8. Olaf Groth, Program Director for Digital Futures,
Hult International Business School. Mr. Groth
2. Jack Clark, Strategy and Communications Director,
researches AI, automation, the Internet of Things
OpenAI. Mr. Clark’s work focuses on setting AI
and the 4th Industrial Revolution. He has testified
policy and strategy. He frequently participates in
in front of the California State Assembly, where
fact-finding studies and forums related to AI and
he emphasized the importance of governing AI so
its impact on policy, ethics and security.
that it empowers humans.
3. Deirdre Mulligan, Faculty Director, U.C. Berkeley
9. Robert Atkinson, President and Founder,
Center for Law and Technology. Ms. Mulligan’s
Information Technology and Innovation
researches information policy, law, privacy
Foundation. Mr. Atkinson primarily works on
and security. Her work explores the legal and
issues related to technological innovation, global
technical means of protecting values such as
competitiveness and productivity growth. He
privacy, freedom of expression and fairness in
has served on various federal boards and task
emerging technical systems.
forces, including the Commission on Workers,
Communities and Economic Change in the New
4. Eric Horvitz, Distinguished Scientist and Managing
Economy.
Director, Microsoft Research, Redmond.
Mr. Horvitz’s researches the principles and
applications of machine intelligence. He also
has examined how humans and machine can
complement each other, creating opportunities
for collaboration and enhanced productivity.
44 | Little Hoover Commission
10. Susan Athey, Senior Fellow, Stanford Institute for 6. Silicon Valley Leadership Group. The Silicon Valley
Economic Policy Research. Ms. Athey’s researches Leadership Group represents more than 375 of
the economics of the internet and digitization. Silicon Valley’s most respected employers on
She currently teaches classes about the issues, programs and campaigns that affect the
economics of the internet and digital markets, economic health and quality of life in Silicon
platform markets and internet search. Valley, including tech and innovation.
7. Stanford University Center for Internet and Society
Top Groups
(CIS). CIS brings together scholars, academics,
legislators and students to study the interaction
1. AI Now Institute. The AI Now Institute at New of new technologies and the law and to examine
York University is an interdisciplinary research how the synergy between the two can either
center dedicated to understanding the social promote or harm public goods like free speech,
implications of AI across four areas: rights innovation, or privacy.
and liberties, labor and automation, bias and
inclusion and safety and critical infrastructure.
8. TechNet. TechNet promotes the development
of entrepreneurship, mobile commerce and the
2. Center for Information Technology Research in the next wave of innovation in the new economy.
Interest of Society (CITRIS). CITRIS is sponsoring The organization believes that establishing an
a multi-campus research initiative to explore a innovation-friendly policy framework is the key to
broad range of machine applications that can the competitiveness of the technology industry.
benefit people and society.
9. UC Berkeley Laboratory for Automation Science and
3. CompTIA. CompTIA supports the success of IT Engineering (AUTOLAB). The AUTOLAB is a center
businesses across the full IT channel, from the for research in robotics and automation. Students
largest vendors on one end to smaller businesses pursue and work on projects related to computer
providing IT hardware, software and services on assisted surgery and automated manufacturing,
the other end. among other robotic applications.
4. Information Technology and Innovation 10. USC Center for Artificial Intelligence in Society
Foundation. The Information Technology and (CAIS). CAIS works to share ideas about how AI
Innovation Foundation (ITIF) focuses on issues can be used to tackle the most difficult societal
at the intersection of technological innovation problems. The organization conducts AI research
and public policy, such as innovation and and projects that can be used to benefit low-
competitiveness, information technology and resource communities in the United States and
data and trade and globalization. globally.
5. Partnership on AI to Benefit People and Society
Top Books
(PAI). A non-profit organization that strives to
advance the public’s understanding of artificial
1. Ajay Agrawal, Joshua Gans, and Avi Goldfarb,
intelligence, PAI also aims to set societal and
April 2018, Prediction Machines: The Simple
ethical best practice for AI research.
Economics of Artificial Intelligence. This book
suggests how companies can set strategies, how
governments can design policies and how people
can plan their lives in an AI world.
Resources | 45
2. Darrell M. West, May 2018, The Future of Work: 7. Martin Ford, July 2016, Rise of the Robots:
Robots, AI, and Automation. This book argues Technology and the Threat of a Jobless Future.
that society needs to rethink the concept of jobs, This book describes how automation is changing
reconfigure the social contract, move toward a the economy, undermining work and reshaping
system of lifetime learning and develop a new lives. It examines what accelerating technology
kind of politics that can deal with economic means for the economic prospects of society,
dislocations. now and in the future.
3. Erik Brynjolfsson and Andrew McAfee, January 8. Paul R. Daugherty and H. James Wilson, March
2016, The Second Machine Age. This book 2018, Human + Machine: Reimagining Work
examines how digital technology is transforming in the Age of AI. The book examines how AI
work and society. The authors recommend is transforming all of the business processes
revamping education to prepare people for the within an organization—whether related to
next economy instead of the last one, designing breakthrough innovation, everyday customer
new collaborations that pair processing power service or personal productivity habits.
with human ingenuity and embracing policies
that make sense in a radically transformed
9. Pedro Domingos, February 2018, The Master
landscape.
Algorithm: How the Quest for the Ultimate
Learning Machine Will Remake Our World. This
4. John Markoff, August 2015, Machines of Loving book reviews the inside the learning machines
Grace: The Quest for Common Ground Between that power Google, Amazon and smartphones.
Humans and Robots. This book considers: will It assembles a blueprint for the future universal
machines help us, or will they replace us? It learner—the Master Algorithm—and discusses
offers a detailed and rich history of the field of what it will mean for business, science and
robotics and AI and emphasizes the importance society.
of including a human element into these
technologies.
10. Yuval Noah Harari, September 2018, 21 Lessons
for the 21st Century. This book examines
5. Kai-Fu Lee, September 2018, AI Superpowers: how today’s most urgent issues will evolve as
China, Silicon Valley, and the New World Order. technology advances faster, forcing society and
Due to unprecedented developments in AI, policymakers to ask important questions in order
dramatic changes will happen sooner than to survive.
many people expect. The book urges both
the U.S. and China to accept and embrace the
great responsibilities that come with significant
technological power.
6. Mariya Yao, Adelyn Zhou, and Marlene Jia, June
2018, Applied Artificial Intelligence: A Handbook
For Business Leaders. This book is a practical
guide for business leaders to leverage machine
intelligence to enhance the productivity of their
organizations and the quality of life in their
communities.
46 | Little Hoover Commission
Notes 2017. “Five Ways Artificial Intelligence is Already
Changing Government.” http://www.govtech.
com/computing/5-Ways-Artificial-Intelligence-Is-
1. Jacques Bughin et al. September 2018. McKinsey Already-Changing-Government.html.
Global Institute.”Notes From the AI Frontier:
Modeling the Impact of AI on the World Economy.” 10. James Manyika et al. McKinsey Global Institute.
https://www.mckinsey.com/featured-insights/ December 2017. “Jobs Lost, Jobs Gained:
artificial-intelligence/notes-from-the-ai-frontier- Workforce Transitions in a Time of Automation.”
modeling-the-impact-of-ai-on-the-world-economy. https://www.mckinsey.com/featured-insights/
future-of- organizations-and-work/what-the-
2. California Department of Finance. May 4, future-of-work- will-mean-for-jobs-skills-and-
2018.”California State Gross Domestic Product wages. Accessed May 4, 2018.
(GDP), 1963-2017.” http://www.dof.ca.gov/
Forecasting/Economics/Indicators/Gross_State_ 11. Gordon Ritter. Recode. “AI can Take Over
Product/documents/BBStateGDP_000.xls. Also, Our Mundane Tasks.” https://www.
International Monetary Fund. October 2018. recode.net/2017/10/18/16492156
“World Economic Outlook Database. “ https:/www. coaching-cloud-future- work-jobs-
imf.org/external/pubs/ft/weo/2018/02/weodata/ artificial-intelligence-ai-enterprise-
weoselgr.aspx. Accessed November 20, 2018. employee-training. Accessed October
15, 2018.
3. James Manyika et al. January 2017. McKinsey
Global Institute. “A Future That Works: 12. Tata Communications. September 2018. “Cognitive
Automation, Employment, and Productivity.” Diversity: AI & The Future of Work.” https://
https://www.mckinsey.com/featured-insights/ www.tatacommunications.com/wp-content/
digital-disruption/harnessing-automation-for-a- uploads/2018/09/Report_Cognitive-Diversity_AI-
future-that-works. and-The-Future-of-Work.pdf.
4. James Manyika et al. See endnote 3. 13. James Manyika et al. See endnote 10.
5. These figures are based on estimates by a 2013 14. Council of Economic Advisors. July 2018.
Oxford study that 47% of total U.S. employment “Addressing America’s Reskilling Challenge.”
is at risk of being automated and estimates by a https://www.whitehouse.gov/wp-content/
2017 McKinsey study that estimates less than 5% uploads/2018/07/Addressing-Americas-Reskilling-
of jobs will be fully automated. They also assume Challenge.pdf.
18,783,600 employed in California based on figures
from the California Employment Development 15. Future of Life Institute. January 2017. “Asilomar AI
Department. Carl Benedikt Frey and Michael Principles.” https://futureoflife.org/ai-principles/.
Accessed September 10, 2018.
A. Osborne. September 17, 2013. “The Future
of Employment: How Susceptible Are Jobs to
16. The White House. May 10, 2018. “Artificial
Computerisation?” https://www.oxfordmartin.
Intelligence for the American People.” https://www.
ox.ac.uk/downloads/academic/The_Future_
whitehouse.gov/briefings-statements/artificial-
of_Employment.pdf. Also, James Manyika et
intelligence-american-people/. Will Knight. MIT
al. See endnote 3. Also, California Employment
Technology Review. May 10, 2018. “The White
Development Department. 2018. “California Profile:
House Says a New AI Task Force Will Protect
Employment and Wages.”
Workers and Keep America First.” https://www.
https://www.labormarketinfo.edd.ca.gov/cgi/ technologyreview.com/s/611118/the-white-house-
databrowsing/localAreaProfileQSResults.asp?selec says-a-new-ai-taskforce-will-keep-america-first-and-
tedarea=California&selectedindex=0&menuChoice protect-its-workers/.
=localAreaPro&state=true&geogArea=0601000000.
November 15, 2018. 17. Subcommittee on Information Technology
and Committee on Oversight and Government
6. Seema Singh. Towards Data Science. May Reform. September 2018. U.S. House of
26, 2018. “Cousins of Artificial Intelligence.” Representatives. “Rise of the Machines
https://towardsdatascience.com/cousins-of- Artificial Intelligence and its Growing Impact on
artificial-intelligence-dda4edc27b55. U.S. Policy.” https://oversight.house.gov/wp-
content/uploads/2018/09/AI-White-Paper-.pdf.
7. Pedro Domingos. Basic Books. September
22, 2015. “The Master Algorithm: How the 18. U.S. House of Representatives.
Quest for the Ultimate Learning Machine Will “Congressional Artificial Intelligence Caucus.”
Remake Our World.” https://artificialintelligencecaucus-olson.
house.gov/.
8. Roger Parloff. Fortune. September 28, 2016. “Why
Deep Learning is Suddenly Changing Your Life.” 19. Digital Single Market. “High-Level Expert Group
http://fortune.com/ai-artificial-intelligence-deep- on Artificial Intelligence.” https://ec.europa.eu/
machine-learning/. digital-single-market/en/high-level-expert-group-
artificial-intelligence.
9. William D. Eggers. Government Technology. May 4,
Notes | 47
20. Cédric Villani. March 8, 2018. “For a Meaningful Drug Overdoses in Michigan.” https://www.
Artificial Intelligence: Towards a French and michigan.gov/snyder/
European Strategy.” https://www.aiforhumanity.
fr/pdfs/MissionVillani_Report_ENG-VF.pdf. 31. New York State Senate. “2017-18 Legislative
Session. Assembly Bill A8821.” https://www.
21. Stanford University. May 30, 2018. “Switzerland nysenate.gov/legislation/bills/2017/A8821.
Launches Program to Test AI for Refugee
Integration.” https://phys.org/news/2018-05- 32. Vermont General Assembly. “Act 137
switzerland-ai-refugee.html. (H.378), 2018.”https://legislature.vermont.
gov/bill/status/2018/H.378.
22. Michael Horowitz. Texas National Security
Review: Volume 1, Issue 3. May 2018. “Artificial 33. Washington State Legislature. “SB 6544
intelligence, International Competition, and the (2017-2018).” http://apps2.leg.wa.gov/billsu
Balance of Power.” https://tnsr.org/wp-content/ mmary?BillNumber=6544&Year=2017&BillN
uploads/2018/05/TNSR-Vol-1-Iss-3_Horowitz.pdf. umber=6544&-Year=2017.
23. Michael Horowitz. See endnote 22. 34. Hawaii State Legislature. “29th Legislature,
2018. HCR89.” https://www.capitol.hawaii.
24. Janosch Delcker. Politico. July 23, 2018. “Germany’s gov/Archives/measure_indiv_Archives.aspx?
Falling Behind on Tech, and Merkel Knows It.” billtype=HCR&billnumber=89&year=2017.
https://www.politico.eu/article/germany-falling-
behind-china-on-tech-innovation-artificial- 35. Business Insider. June 15, 2017. “Hawaii
intelligence-angela-merkel-knows-it/. Accessed Just Became the First U.S. State to Pass a
September 7, 2018. Bill Supporting Basic Income.” https://www.
businessinsider.com/hawaii-basic-income-
25. Meng Jing. South China Morning Post. May 3, 2018. bill-2017-6.
“China Looks to School Kids to Win the Global AI
Race.” https://www.scmp.com/print/tech/china- 36. New York City Council. “Legislative Research
tech/article/2144396/china-looks-school-kids-win- Center. File # Int 1696-2017.” http://
global-ai- race. Accessed September 7, 2018. legistar.council.nyc.gov/LegislationDetail.
aspx?ID=3137815&GUID=437A6A6D- 62E1-
26. Synched Review. May 16, 2018. “South Korea 47E2-9C42-461253F9C6D0.
Aims High on AI, Pumps $2 Billion into R&D.”
https://medium.com/syncedreview/south- 37. P4 Pittsburgh. http://p4pittsburgh.org/.
korea-aims-high-on-ai-pumps-2-billion-into-r-d-
de8e5c0c8ac5. 38. Bureau of Labor Statistics. February 3,
2018. “Employment Projections: Overview.”
27. Strategic Council for Artificial Intelligence https://www.bls.gov/opub/hom/emp/home.
Technology. March 31, 2017. “Artificial intelligence htm. Accessed April 30, 2018.
Technology Strategy.” http://www.nedo.go.jp/
content/100865202.pdf 39. Amar Mann, Branch Chief and Supervisory
Economist, Economic Analysis and
28. Gary Bertoline. November 8, 2017. “Reinventing Information, U.S. Bureau of Labor Statistics
Education: The Impact of Automation, Innovation West Regional Office. Sacramento, CA.
& Technology.” https://www.in.gov/dwd/ May 24, 2018. Little Hoover Commission.
files/Future%20of%20Work%20Taskforce%20 Advisory committee meeting.
Presentation%20-%20Bertoline.pdf Also, Indiana
Department of Workforce Development. “SWIC: 40. Laura Coleman, Statewide Director, California
Future of Work Taskforce.” https://www.in.gov/ Community Colleges Centers of Excellence for
dwd/3162.htm. Labor Market Research. March 15, 2018. Personal
communication with Commission staff. Also,
29. Office of Governor Charlie Baker of Massachusetts. Laura Coleman, Statewide Director, California
November 20, 2017. “Baker-Polito Administration Community Colleges Centers of Excellence for
Establishes New Commission on Digital Learning.” Labor Market Research. May 24, 2018. Little
https://www.mass.gov/news/baker-polito- Hoover Commission. Advisory committee
administration-establishes-new-commission-on- meeting.
digital-learning. Also, Office of Governor Charlie
Baker of Massachusetts. April 23, 2018. “Governor 41. Michael Pooler. Financial Times. November 19,
Baker Swears In New Commission on Digital 2017. “Robot Army is Transforming the Global
Innovation and Lifelong Learning.” https://www. Workplace.” https://www.ft.com/content/
mass.gov/news/ governor-baker-swears-in-new- f04128de-c4a5-11e7-b2bb-322b2cb39656.
commission-on-digital- innovation-and-lifelong-
learning. 42. The Economist. June 25, 2016. “Special
Report: The Impact on Jobs–Automation and
30. Office of Governor Rick Snyder of Michigan. Anxiety.” https://www.economist.com/special-
Newsroom.Press Releases. April 11, report/2016/06/25/automation-and-anxiety.
2018. “New Data Assessment Provides
Information About Patterns of Opioid Use, 43. Robert Atkinson, President, Information Technology
48 | Little Hoover Commission
and Innovation Foundation. January 25, 2018. for the Earth.” https://www.pwc.com/gx/en/
Written testimony to the Commission. sustainability/assets/ai-for-the-earth-jan-2018.pdf.
44. Steve Viscelli. Center for Labor Research and 54. Pricewaterhouse Coopers. June 2017. “What
Education, University of California, Berkeley, and Doctor? Why AI and Robotics Will Define New
Working Partnerships USA. September 2018. Health.” https://www.pwc.com/gx/en/industries/
“Driverless? Autonomous Trucks and the Future of healthcare/publications/ai-robotics-new-health/
the American Trucker.” http://www.driverlessreport. ai-robotics-new-health.pdf.
org/. Accessed September 6, 2018.
55. Spencer Soper. Bloomberg. June 13, 2018.
45. Margo Gonzales, Manager, Labor Market “Amazon’s Clever Machines Are Moving From
Information Division, and Paul Wessen, Senior the Warehouse to Headquarters.” https://www.
Economist, Employment Development Department. bloomberg.com/news/articles/2018-06-13/
Sacramento, CA. May 24, 2018. Little Hoover amazon-s-clever-machines-are-moving-from-the-
Commission. Advisory committee meeting. Also, warehouse-to-headquarters.
Amar Mann. See endnote 39.
56. Peter Dorfman. Autodesk. January 3, 2018.
46. State Controller’s Office. October 2018. “State “3 Advances Changing the Future of Artificial
Employee Demographics.” https://www.sco. Intelligence in Manufacturing.” https://www.
ca.gov/ppsd_empinfo_demo.html. Accessed autodesk.com/redshift/future-of-artificial-
November 15, 2018. intelligence/.
47. Alan Leshner, Chief Executive Officer, American 57. Adobe. “Adobe Sensei.” www.adobe.com/sensei.
Association for the Advancement of Science. html. Accessed September 8, 2018.
February 14, 2003. “Public Engagement with
Science.” Science, Vol. 299, Issue 5609. Accessed 58. B2B News Network. April 10, 2018. “Adobe Shows
November 26, 2018 at http://science.sciencemag. how Sensei AI Technology is Helping Marketers
org/content/299/5609/977.full. Search for Stock Images Without Words.” http://
www.b2bnn.com/2018/04/adobe-sensei-ai/.
48. Rene Von Schomberg. 2013. “A Vision of Accessed September 8, 2018.
Responsible Innovation.” http://www.pacitaproject.
eu/wp-http://www.pacitaproject.eu/wp- 59. Marty Swant. Adweek. March 24, 2017. “Adobe
content/uploads/2014/04/von-Schomberg-RRI- and IBM Are Rolling Out More Artificial Intelligence
owenbookChapter.pdf. Accessed November 19, Tools for Brands.” https://www.adweek.com/
2018. digital/adobe-and-ibm-are-rolling-out-more-
artificial-intelligence-tools-for-brands/. Accessed
49. Hamza Shaban. Washington Post. October 5, September 8, 2018.
2018. “Silicon Valley Congressman Unveils
an Internet Bill of Rights.” https://www. 60. Andre Snellings. ESPN. September 2018. “Fantasy
washingtonpost.com/technology/2018/10/05/ Insights with Watson.” http://www.espn.com/
silicon-valley-congressman-unveils-an-internet- fantasy/football/story/_/page/insights090718/
bill-rights/?utm_term=.3edde88df030. fantasy-football-fantasy-insights-watson. Accessed
Accessed November 20, 2018. September 8, 2018.
50. California Consumer Privacy Act of 2018 (Civ. Code, 61. Kyle Wiggers. July 19, 2018. “IBM Watson Health
§§ 1798.100 through 1798.199). and the VA Extend Partnership to Fight Cancer.”
https://venturebeat.com/2018/07/19/ibm-watson-
health-and-the-va-extend-partnership-to-fight-
51. Ryan Rakestraw and Amrit Acharya.
McKinsey & Company. September 5, cancer/. Accessed September 8, 2018.
2017. “Can Artificial Intelligence Help
Feed the World?” https://www.forbes. 62. Bernard Marr. Forbes. October 2, 2017. “The
com/sites/themixingbowl/2017/09/05/ Amazing Ways Microsoft Uses AI to Drive
can-artificial-intelligence-help-feed-the- Business Success.” https://www.forbes.com/
world/#334a05d146db. sites/bernardmarr/2017/10/02/the-amazing-
ways-microsoft-uses-ai-to-drive-business-
52. Stanford University. One Hundred Year Study success/.
on Artificial Intelligence: Report of the 2015-
2016 Study Panel. September 6, 2016. “Artificial 63. Jay Greene. The Wall Street Journal. January 13,
Intelligence and Life in 2030.” http://ai100. 2017. “Microsoft Acquires Artificial-Intelligence
stanford.edu/2016-report. Also, Adam Avery. The Startup Maluuba.” https://www.wsj.com/articles/
Atlantic. 2017. “Artificial Intelligence Promises microsoft-acquires-artificial-intelligence-startup-
a Personalized Education for All.” https://www. maluuba-1484338762. Accessed September 8,
theatlantic.com/sponsored/vmware-2017/ 2018.
personalized-education/1667/.
64. Digital Democracy. “About Digital Democracy.”
53. Celine Herweijer et al. Pricewaterhouse Coopers. https://www.digitaldemocracy.org/about-digital-
January 2018. “Fourth Industrial Revolution democracy. Accessed September 8, 2018.
for the Earth: Harnessing Artificial Intelligence
Notes | 49
65. One Concern. “What We Do.” https://www. AI-Assisted Care (PAC).” http://med.stanford.
oneconcern.com/product/. Accessed September edu/ cerc/research/new-pac.html.
8, 2018.
77. Susan Hall. Health Tech Magazine. December
66. Jason Shueh. Government Technology. March 25, 27, 2017. “How Artificial Intelligence Impacts
2016. “One Concern: Applying Artificial intelligence Senior Care.” https://healthtechmagazine.net/
to Emergency Management.” http://www. article/2017/12/how- artificial-intelligence-impacts-
govtech.com/em/disaster/One-Concern-Applying- senior-care. Also, Jon Stone and Justin Beach.
Artificial-Intelligence-Emergency-Management. University of Denver. October 26, 2017. “Companion
html. Also, Ray Mueller, CEO, Vice President of Robot Helping Patients with Alzheimer’s.” http://
Government, One Concern. July 13, 2018. Personal news.du.edu/companion-robot- helping-patients-
communication with Commission staff. with-alzheimers/.
67. Google. “AI for Everyone: Inside TensorFlow, 78. Jonathan Vespa et al. U.S. Census Bureau.
Our Open-Source Machine Learning Platform.” March 2018. “Demographic Turning
https://www.ai.google/stories/tensorflow/ Points for the United States: Population
Accessed September 8, 2018. Projections for 2020 to 2060.” https://www.
census.gov/content/dam/Census/library/
68. Marty Graham. Dell Technologies. February publications/2018/demo/P25_1144.pdf.
20, 2018. “How Big Data is Helping Predict and Also, Department of Finance Demographics
Fight Wildfires.” https://www.delltechnologies. Research Unit. http://www.dof.ca.gov/
com/en-us/perspectives/how-big-data-is- Forecasting/Demographics/. Accessed
helping-predict-and-fight-wildfires/. December 29, 2017.
69. Mary Beth Faller. Arizona State University. June 12, 79. Pricewaterhouse Coopers. See endnote 54.
2018. “Students will Compete for $100,000 Prize
in International Microsoft Pitch This Summer.” 80. Matthew Collier et al. Accenture. December
https://asunow.asu.edu/20180612-arizona- 15, 2017. “Artificial Intelligence: Healthcare’s
impact-asu-team-uses-artificial-intelligence- New Nervous System.” https://www.
detect-wildfires. accenture.com/t20171215T032059Z__w__/
us-en/_acnmedia/PDF-49/Accenture-Health-
70. Emily Matchar. Smithsonian Magazine. Artificial-Intelligence.pdf.
September 9, 2016. “This NASA-Developed AI
Could Help Save Firefighters’ Lives.” https:// 81. Brad Smith. May 7, 2018. “Using AI to Empower
www.smithsonianmag.com/innovation/nasa- People with Disabilities.” https://blogs.microsoft.
developed-ai-could-help-save-firefighters- com/on-the-issues/2018/05/07/using-ai-to-
lives-180960398/. empower-people-with-disabilities/.
71. Ryan Harkins, Director of State Affairs and 82. AI Now Institute. October 2017. “AI Now 2017
Public Policy, Microsoft. August 23, 2018. Little Report.” https://ainowinstitute.org/AI_Now_2017_
Hoover Commission. Advisory committee Report.pdf. Accessed September 6, 2018.
meeting.
83. Ryan Calo. August 9, 2017. “Artificial Intelligence
72. Rich McKay. October 23, 2017. “How AI Boosts Policy: A Primer and Roadmap.” https://lawreview.
Emergency Response in the New Age of Super law.ucdavis.edu/issues/51/2/Symposium/51-2_Calo.
Disasters.” https://www.ibm.com/blogs/ pdf.
industries/ ai-boosts-emergency-response-new-
age-super- disasters/. 84. World Economic Forum. March 2018. “How to
Prevent Discriminatory Outcomes in Machine
73. USC Center for Artificial intelligence in Society. Learning.”http://www3.weforum.org/docs/
“PAWS: Protection Assistant for Wildlife Security.” WEF_40065_White_Paper_How_to_Prevent_
https:// www.cais.usc.edu/projects/wildlife- Discriminatory_Outcomes_in_Machine_Learning.pdf.
security/. Accessed September 6, 2018.
74. Celine Herweijer et al. See endnote 53. 85. Dina Bass and Ellen Huet. Bloomberg. December
4, 2017. “Researchers Combat Gender and Racial
75. Eric Rice and Milind Tambe, Co-Directors, USC Bias in Artificial Intelligence.” https://www.
Center for Artificial Intelligence in Society. January bloomberg.com/news/articles/2017-12-04/
25, 2018. Written testimony and testimony to the researchers-combat- gender-and-racial-bias-in-
Commission. artificial-intelligence.
76. John Markoff. Washington Post. October 18, 86. Jonathan Vanian. Fortune. June 25, 2018.
2017. “Artificial intelligence Could Improve “Unmasking A.I.’s Bias Problem.” http://fortune.com/
How We Age.” https://www.washingtonpost. longform/ai-bias-problem/.
com/news/theworldpost/wp/2017/10/18/
artificial-intelligence-could-improve-how- 87. AI Now Institute. See endnote 82.
we-age/?utm_term=. b8e6b84d662c;.
Also, Stanford Medicine. “Partnership in
50 | Little Hoover Commission
88. World Economic Forum. See endnote 84. users- raises-a-data-privacy-problem/.
89. AI Now Institute. See endnote 82. 104. James Vincent. The Verge. July 3, 2017. “Google’s
DeepMind and UK hospitals Made Illegal Deal
90. Jackie Snow. MIT Technology Review. February for Health Data, Says Watchdog.” https://www.
14, 2018. “We’re in a Diversity Crisis: Cofounder theverge.com/2017/7/3/15900670/google-
of Black in AI on What’s Poisoning Algorithms deepmind-royal-free-2015-data-deal-ico-ruling-
in Our Lives.” https://www.technologyreview. illegal.
com/s/610192/were-in-a-diversity-crisis-black-in-
ais-founder-on-whats-poisoning-the-algorithms-in- 105. Matt Cagle, Technology and Civil Liberties
our/. Attorney, American Civil Liberties Union. March 6,
2018. Testimony before California State Assembly,
Assembly Privacy & Consumer Protection
91. Foaad Khosmood, Senior Research Fellow, Institute
Committee, Assembly Select Committee on
for Advanced Technology and Public Policy. August Emerging Technologies and Innovation.
23, 2018. Little Hoover Commission. Advisory
committee meeting. 106. Jennifer King, Director of Consumer Privacy, Stanford
University Center for Internet and Society. July 9,
92. Kate Crawford and Ryan Calo. Nature. October 2018. Personal communication with Commission
13, 2016. There is a Blind Spot in AI Research. staff.
https://www.nature.com/news/there-is-a-
blind-spot-in-ai- research-1.20805#/ref-link-2. 107. Cecilia Kang. New York Times. August 26, 2018.
“Tech Industry Pursues a Federal Privacy Law,
93. Julia Angwin, Jetal. ProPublica. May 23, 2016. on Its Own Terms”. https://www.nytimes.
“Machine Bias.” https://www.propublica.org/ com/2018/08/26/technology/tech-industry-
article/machine-bias-risk-assessments-in-criminal- federal-privacy-law.html.
sentencing.
108. Jared Bennett. The Daily Beast. July 31, 2017.
94. California Department of Corrections “Facebook: Your Face Belongs to Us.” https://www.
and Rehabilitation. “COMPAS Assessment thedailybeast.com/how-facebook-fights-to-stop-
Tool Launched.” https://www.cdcr.ca.gov/ laws-on-facial-recognition. Accessed August 13,
rehabilitation/docs/FS_COMPAS_Final_4-15-09. 2018.
pdf. Accessed December 28, 2017.
109. Andrea Peterson. The Washington Post. June
16, 2015. “The Government’s Plan to Regulate
95. Ryan Calo. See endnote 83.
Facial Recognition Tech is Falling Apart.” https://
www.washingtonpost.com/news/the-switch/
96. Ryan Calo. See endnote 83. wp/2015/06/16/the-governments-plan-to-regulate-
facial-recognition-tech-is-falling-apart/?utm_
97. Christina Couch. PBS. October 25, 2017. “Ghosts term=.0ca2a0008dbf. Accessed August 13, 2018.
in the Machine.” http://www.pbs.org/wgbh/nova/
next/tech/ai-bias/. 110. Federal Trade Commission. October 22, 2012.
“FTC Recommends Best Practices for Companies
98. Dr. Ben Buchanan, Postdoctoral Fellow, Belfer That Use Facial Recognition Technologies.”
Center Cybersecurity Project, Harvard University. https://www.ftc.gov/news-events/press-
April 18, 2018. Testimony before U.S. House of releases/2012/10/ftc-recommends-best-
Representatives, Committee on Oversight and practices-companies-use-facial- recognition.
Government Reform, Subcommittee on Information Accessed August 14, 2018.
Technology. Hearing on “Game Changers: Artificial
Intelligence Part III.”https://oversight.house.gov/ 111. Matt Wood. June 1, 2018. “Some Quick Thoughts
wp-content/uploads/2018/04/Buchanan-Harvard- on the Public Discussion Regarding Facial
Statement-AI-III-4-18.pdf. Recognition and Amazon Rekognition This Past
Week.” https://aws.amazon.com/blogs/aws/some-
99. AI Now Institute. See endnote 82. quick-thoughts-on-the-public-discussion-regarding-
facial-recognition- and-amazon-rekognition-
this-past-week/. Also, Matt Wood. July 27, 2018.
100. World Economic Forum. See endnote 84.
“Thoughts on Machine Learning Accuracy.” https://
aws.amazon.com/blogs/aws/thoughts-on-machine-
101. Cliff Kuang. New York Times. November 21, 2017.
learning-accuracy/?tag=cnet-vig-news-20. Accessed
“Can AI be Taught to Explain Itself?”https://www.
August 13, 2018.
nytimes.com/2017/11/21/magazine/can-ai-be-
taught-to-explain-itself.html.
112. Dawn Kawamoto. Government Technology. July 24,
2018. “Facial Recognition Gets Better as the Debate
102. Ryan Harkins. See endnote 71.
About Its Merits Gets Louder.” http://www.govtech.
com/products/Facial-Recognition-Gets-Better-as-
103. Youyou Zhou. Quartz. May 25, 2018. “An Oregon the-Debate-About-its-Merits-Gets-Louder.html.
Family’s Encounter with Amazon Alexa Exposes the Accessed August 14, 2018.
Privacy Problem of Smart Home Devices.” https://
qz.com/1288743/amazon-alexa-echo-spying-on-
Notes | 51
113. Andrea Peterson. See endnote 109. and Innovation Foundation. August 7, 2017.
“Unfortunately, Technology Will Not Eliminate Many
114. Tom Simonte. Wired. May 22, 2018. “Few Jobs.” https://itif.org/publications/2017/08/07/
Rules Govern Police Use of Facial-Recognition unfortunately-technology-will-not-eliminate-many-
Technology.”https://www.wired.com/story/few-rules- jobs.
govern-police-use-of-facial-recognition-technology/.
Accessed August 14, 2018. 131. Mike Maciag. Governing Magazine. August 2017.
“Is Automation Anxiety All Hype?” http://www.
115. Brad Smith. July 13, 2018. “Facial Recognition governing.com/topics/mgmt/gov-automation-jobs-
Technology: The Need for Public Regulation research.html.
and Corporate Responsibility.” https://blogs.
microsoft.com/on-the-issues/2018/07/13/ 132. Robert Atkinson. See endnote 43.
facial-recognition- technology-the-need-for-
public-regulation-and- corporate-responsibility/. 133. The Economist. See endnote 42.
Accessed August 13, 2018.
134. James Manyika et al. See endnote 10.
116. James Manyika et al. See endnote 10.
135. Michael Pooler. Financial Times. November 19,
117. James Manyika et al. See endnote 10. 2017. “Robot Army is Transforming the Global
Workplace.” https://www.ft.com/content/
118. Bureau of Labor Statistics. See endnote 38. f04128de-c4a5-11e7-b2bb-322b2cb39656
136. Peggy Hollinger. Financial Times. “Meet the Cobots:
119. Amar Mann. See endnote 39.
Humans and Robots Together on the Factory Floor.”
https://www.ft.com/content/6d5d609e-02e2-
120. Margo Gonzales and Paul Wessen. See endnote 45. 11e6-af1d-c47326021344.
121. Amar Mann. See endnote 39. 137. CB Insights. July 18, 2018. “Smaller
Collaborative Robots Are Disrupting The
122. Ryan Shiroi, Research Program Specialist, Robotics Industry.” https://www.cbinsights.
Employment Development Department. March 20, com/research/collaborative-robots-reinvent-
2018. Personal communication with Commission industries/?utm_ source=CB+Insights+N
staff. ewsletter&utm_campaign=5a9dea0dc1-
ThursNL_07_26_2018&utm_
123. Gustavo Avilés, Manager, Workforce Information medium=email&utm_term=0_9dc0513989-
and Technology Services, Washington 5a9dea0dc1-88449373.
Employment Security Department. Sacramento,
CA. May 24, 2018. Little Hoover Commission. 138. David Rotman. MIT Technology Review. June 12,
Advisory committee meeting. 2013. “How Technology Is Destroying Jobs.” https://
www.technologyreview.com/s/515926/how-
124. Carl Benedikt Frey and Michael A. Osborne. technology-is-destroying-jobs/.
See endnote 5.
139. The Economist. See endnote 42.
125. Financial Times. “Best Business Books: The Rise
of the Robots by Martin Ford.” https://ig.ft.com/
140. Robert Atkinson. See endnote 43.
sites/business-book-award/books/2015/winner/
the-rise-of-the-robots-by-martin-ford/.
141. Tia Orr, Director of Government Affairs, Service
Employees International Union California. May
126. Kevin Delaney. Quartz. February 17, 2017. “The
24, 2018. Little Hoover Commission. Advisory
Robot that Takes Your Job Should Pay Taxes, Says
committee meeting.
Bill Gates.” https://qz.com/911968/bill-gates-the-
robot-that- takes-your-job-should-pay-taxes/.
142. Annette Bernhardt, Director of the Low-Wage Work
Program, UC Berkeley Center for Labor Research
127. Catherine Clifford. CNBC. November 4, 2016.
and Education. January 25, 2018. Written testimony
“Elon Musk: Robots Will Take Your Jobs,
to the Commission.
Government Will Have to Pay Your Wage.”
https://www.cnbc.com/2016/11/04/elon-musk-
143. Sara Flocks, Policy Coordinator, California
robots-will-take-your- jobs-government-will-
Labor Federation. May 24, 2018. Little Hoover
have-to-pay-your-wage.html.
Commission. Advisory committee meeting.
128. Robert Atkinson. See endnote 43.
144. Sara Flocks, Policy Coordinator, California Labor
Federation. March 6, 2018. Assembly Standing
129. James Manyika et al. See endnote 10. Committee on Privacy and Consumer Protection
and Assembly Select Committee on Emerging
130. Robert Atkinson, Information Technology Technologies and Innovation. Joint Informational
Hearing.
52 | Little Hoover Commission
145. Annette Bernhardt. See endnote 142. technology/computer-and-information-research-
scientists.htm. Accessed September 7, 2018.
146. Annette Bernhardt. See endnote 142.
158. Projections Central: State Occupational Projects.
147. Jacques Bughin et al. McKinsey Global Institute. “Long Term Projections.” Search Terms: Area:
June 2017. “Artificial Intelligence The Next California; Occupations: Computer and Information
Digital Frontier?” https://www.mckinsey.com/~/ Research Scientist. http://www.projectionscentral.
media/McKinsey/Industries/Advanced%20 com/Projections/LongTerm. Accessed September 7,
Electronics/Our%20 Insights/How%20artificial%20 2018.
intelligence%20can%20 deliver%20real%20
value%20to%20companies/MGI-Artificial- 159. California Community Colleges Chancellor’s
Intelligence-Discussion-paper.ashx. Office. June 2018.”2019-20 Five Year Capital
Outlay Plan.” http://extranet.cccco.edu/Portals/1/
148. Lawrence Mishel and Josh Bivens. Economic ExecutiveOffice/Consultation/2018_agendas/July/4-
Policy Institute. May 24, 2017. “The Zombie Robot Attachment-1-Five-Year-Plan-Proof.pdf. Accessed
Argument Lurches On.” https://www.epi.org/ September 6, 2018.
publication/the-zombie-robot-argument-lurches-
on-there-is-no-evidence-that-automation-leads-to- 160. Education Code, section 71000.
joblessness-or-inequality/.
161. Community College League of California.
149. Steve Viscelli. See endnote 44. September 1999. “Community College
Governance. A Joint Statement of the Board of
150. Michael Chui, Partner, McKinsey Global Institute. Governors and CCCT.” https://www.ccleague.org/
Sacramento, CA. January 2018. Testimony to the sites/default/files/pdf/CCGovernanceJntStmt.pdf.
Little Hoover Commission. Accessed September 7, 2018.
151. Legislative Analyst Office. July 2010. “What 162. California Community Colleges.
is California’s Master Plan for Higher “About The Consultation Council.”
Education?” http://californiacommunitycolleges.
https://lao.ca.gov/sections/education/ed- cccco.edu/ChancellorsOffice/
basics/What-CA-Master-Plan-For-HE.pdf. ConsultationCouncil.aspx. Accessed
Accessed September 7, 2018. September 5, 2018.
152. California State Legislature. “Senate Bill No. 163. California Community Colleges Chancellor’s Office.
850.” Added Article 3, commencing with Section See endnote 159.
78040 of the Education Code. http://leginfo.
legislature.ca.gov/ faces/billNavClient.xhtml?bill_ 164. California Community Colleges. “Find a
id=201320140SB850. Accessed September 7, Program at a California Community College.”
https://californiacommunitycolleges.cccco.
2018.
edu/FindaProgram. Accessed September 7,
2018.
153. California State Legislature. See endnote 152.
165. Butte College. “Academic Programs Butte
154. Amina Khan. Los Angeles Times. September 1,
College 2017-2018 Catalog.” http://www.
2016. “Two California Universities Open Centers
butte.edu/academicprograms/program_
Dedicated to Artificial Intelligence Research.”
details.php?year=7&program_id=661. Also,
http://www.govtech.com/education/Two-
College of Marin. “Degree Information.”
California-Universities-Open-Centers-Dedicated-
https://netapps.marin.edu/Apps/Directory/
to-Artificial-Intelligence-Research.html. Accessed DegreeSearch.aspx?Keyword=computer%20
September 5, 2018. science. Accessed September 6, 2018.
155. Jeffrey Norris. August 29, 2016. “UC Berkeley 166. California Community Colleges. www.cccco.
launches Center for Human-Compatible edu. Accessed September 6, 2018.
Artificial Intelligence.” http://news.berkeley.
edu/2016/08/29/center-for-human-compatible-
167. California Community Colleges Chancellor’s Office.
artificial-intelligence/. Accessed September 5,
See endnote 159.
2018.
168. California Community Colleges Chancellor’s Office.
156. Marc Ballon. USCNews. August 31, 2016. “Artificial See endnote 159.
Intelligence: Are We Facing a Future of Robots
Running Wild?” https://news.usc.edu/106565/ 169. California Community Colleges Chancellor’s Office
artificial-intelligence-for-a-better-tomorrow/. “Strong Workforce Program: Annual Report 2018.”
Accessed September 7, 2018. http://californiacommunitycolleges.cccco.edu/
Portals/0/Reports/2018-swp-annual-ada.pdf.
157. Bureau of Labor Statistics. April 13, 2018. Accessed September 6, 2018.
“Occupational Outlook Handbook: Computer
and Information Research Scientists.” https:// 170. California Community Colleges Chancellor’s Office.
www.bls.gov/ooh/computer-and-information- See endnote 169.
Notes | 53
171. California Community Colleges
Chancellor’s Office. March 1, 2018. 188. Assembly Privacy & Consumer Protection
“Economic and Workforce Development Committee and Assembly Select Committee on
Program: 2018 Annual Report.” http:// Emerging Technologies & Innovation Joint Hearing.
californiacommunitycolleges.cccco.edu/ March 6, 2018. State Capitol, Sacramento, CA.
Portals/0/Reports/2018-EWD-Annual-Report-
ADA.pdf. Accessed September 6, 2018. 189. Senate Labor and Industrial Relations Committee
and the Assembly Committee on Labor and
172. California Community Colleges Chancellor’s Office. Employment Joint Hearing. March 20, 2018. State
See endnote 171. Capitol, Sacramento, CA.
190. California Department of Corrections
173. Education Code, section 606054.
and Rehabilitation. See endnote 94.
174. See endnote 173.
191. California Governor’s Office of Business and
Economic Development. “iHub Program.” http://
175. See endnote 173. www.business.ca.gov/Programs/Innovation-and-
Entrepreneurship/IHubs. Accessed September 9,
176. California State Board of Education. “Computer 2018.
Science Education.” https://www.cde.ca.gov/
be/st/ss/computerscicontentstds.asp. Accessed 192. Sid Voorakkara, Deputy Director, External
September 6, 2018. Affairs, Governor’s Office of Business and
Economic Development. August 23, 2018.
177. Janet Weeks. California Department of Little Hoover Commission. Advisory committee
Education. September 7, 2018. “California meeting.
Adopts First-Ever Computer Science
Standards.” https://www.cde.ca.gov/nr/ne/ 193. Scott Gregory, Deputy Director and State
yr18/yr18rel56.asp. Accessed September 7, Geographic Information Officer, California
2018. Department of Technology. September 11, 2018.
Personal communication with Commission staff.
178. California State Board of Education. See endnote 176.
194. California Public Sector CIO Academy.
February 27-28, 2018. “The Next Frontier
179. California State Board of Education. See endnote 176.
in Government Transformation: Intelligent
and Cognitive Automation.” http://www.
180. California State Board of Education. See endnote 176. govtech.com/events/California-Public-
Sector-CIO-Academy.html. Accessed
181. California State Legislature. “Assembly Bill No. 1539.” September 6, 2018.
Chapter 876. An act to add Section 60605.4 to the
Education Code. http://leginfo.legislature.ca.gov/ 195. Future of Life Institute. See endnote 15.
faces/billNavClient.xhtml?bill_id=201320140AB1539.
Accessed September 7, 2018.
182. Corinne Lestch. EdScoop. January 9, 2018. “These
States Embraced Computer Science Education
in 2017.” https://edscoop.com/these-states-
embraced-computer-science-education-in-2017.
Accessed September 7, 2018.
183. Diana Lambert. EdSource. September 6, 2018.
“California Adopts First Computer Science
Standards for K-12 Students.” https://edsource.
org/2018/californias-first-computer-science-
standards-set-for-approval/601985. Accessed
September 6, 2018.
184. California Department of Technology. “Digital
Services Innovation Academy.” https://cdt.ca.gov/
whats-new- in-office-of-digital-innovation/. Accessed
September 5, 2018.
185. California Governor’s Office of Business and
Economic Development. “Cybersecurity.” http://
www.business.ca.gov/Programs/Innovation-
and-Entrepreneurship/Cybersecurity. Accessed
September 7, 2018.
186. Danielle Borrelli. June 26, 2018. “North
Hollywood High School Wins Title at California
Cyber Innovation Challenge at Cal Poly.” https://
calpolynews.calpoly.edu/news_releases/2018/
june/cyber_challenge. Accessed August 8, 2018.
187. CyberCalifornia. “About Us.” http://cybercalifornia.
biz/about/. Accessed August 9, 2018.
54 | Little Hoover Commission
Little Hoover Commission Members
Chairman Pedro Nava (D-Santa Barbara) Appointed to the Commission by former Speaker of the Assembly John Pérez
in April 2013 and reappointed by Speaker of the Assembly Anthony Rendon in 2017. Government relations
advisor. Former state Assemblymember from 2004 to 2010. Former civil litigator, deputy district attorney
and member of the state Coastal Commission. Elected chair of the Commission in March 2014.
Vice Chairman Sean Varner (R-Riverside) Appointed to the Commission by Governor Edmund Brown Jr. in
April 2016. and reappointed in January 2018. Managing partner at Varner & Brandt LLP where he practices as
a transactional attorney focusing on mergers and acquisitions, finance, real estate and general counsel work.
David Beier (D-San Francisco) Appointed to the Commission by Governor Edmund G. Brown Jr. in
June 2014 and reappointed in January 2018. Managing director of Bay City Capital. Former senior officer of
Genentech and Amgen. Former counsel to the U.S. House of Representatives Committee on the Judiciary.
Serves on the board of directors for the Constitution Project.
Iveta Brigis (D-Los Gatos) Appointed to the Commission by Governor Edmund G. Brown Jr. in April 2017. Lead for
the open sourcing people operations program at Google Inc. since 2014 and looks after re:Work, Google’s
initiative to open source data-driven HR practices.
Cynthia Buiza (DTS-Los Angeles) Appointed to the Commission by Speaker of the Assembly Anthony Rendon in
October 2018. Executive Director of the California Immigrant Policy Center since 2016. Former Policy and
Advocacy Director for the Coalition for Humane Immigrant Rights of Los Angeles.
Senator Anthony Cannella (R-Ceres) Appointed to the Commission by the Senate Rules Committee in January
2014. Elected in November 2010 and re-elected in 2014 to represent the 12th Senate District. Represents
Merced and San Benito counties and a portion of Fresno, Madera, Monterey and Stanislaus counties.
Assemblymember Chad Mayes (R-Yucca Valley) Appointed to the Commission by former Speaker of the Assembly
Toni Atkins in September 2015. Elected in November 2014 to represent the 42nd Assembly District.
Represents Beaumont, Hemet, La Quinta, Palm Desert, Palm Springs, San Jacinto, Twentynine Palms, Yucaipa,
Yucca Valley and surrounding areas.
Don Perata (D-Orinda) Appointed to the Commission in February 2014 and reappointed in January 2015 by the
Senate Rules Committee. Political consultant. Former president pro tempore of the state Senate, from 2004
to 2008. Former Assemblymember, Alameda County supervisor and high school teacher.
Assemblymember Bill Quirk (D-Hayward) Appointed to the Commission by Speaker of the Assembly
Anthony Rendon in 2017. Elected in November 2012 to represent the 20th Assembly District. Represents
Hayward, Union City, Castro Valley, San Lorenzo, Ashland, Cherryland, Fairview, Sunol and North Fremont.
Senator Richard Roth (D-Riverside) Appointed to the Commission by the Senate Rules Committee in February
2013. Elected in November 2012 to represent the 31st Senate District. Represents Corona, Coronita, Eastvale,
El Cerrito, Highgrove, Home Gardens, Jurupa Valley, March Air Reserve Base, Mead Valley, Moreno Valley,
Norco, Perris and Riverside.
Cathy Schwamberger (NPP-Calistoga) Appointed to the Commission by the Senate Rules Committee in April
2018. Associate general counsel for State Farm Mutual Automobile Insurance Company Company, where
she leads State Farm’s legislative and regulatory affairs in the 10 western states.
Janna Sidley (D-Los Angeles) Appointed to the Commission by Governor Edmund Brown Jr. in April 2016. General
Counsel at the Port of Los Angeles since 2013. Former deputy city attorney at the Los Angeles City Attorney’s
Office from 2003 to 2013.
Full biographies available on the Commission’s website at www.lhc.ca.gov.
“Democracy itself is a process of change, and satisfaction
and complacency are enemies of good government.”
Governor Edmund G. “Pat” Brown,
addressing the inaugural meeting of the Little Hoover Commission,
April 24, 1962, Sacramento, California