Citizens' Perspective on Use of AI in Government | 2018 BCG Digital Government Benchmark

Discussion Paper
2018 BCG Digital Government Benchmark:
Citizens’ Perspectives on the Use of
Artificial Intelligence by Government
By Miguel Carrasco, Steven Mills, Adam Whybrew and Adam Jura
October 2018

Discussion Paper 1
THE BOSTON CONSULTING GROUP October 2018
Artificial Intelligence (AI) has the potential to transform government policy and delivery and
generate enormous value for citizens, but there is a lot of work still needed to build confidence
and trust needed to realize the benefits. Interest and adoption of AI is growing quickly, and many
governments are already implementing a surprisingly wide range of AI-enabled applications. In
China, for example, data from traffic lights, video from CCTV cameras and other sources are
being used to pre-emptively optimize traffic flows to reduce road congestion and clear a path for
police and emergency vehicles to improve response times. In the United Arab Emirates (UAE), a
customer service center is being manned by Rammas, a robot that uses AI to answer enquiries.
France uses AI and data from sensors embedded in tracks, trains and platforms to prioritize
maintenance activities and pre-empt service failures. All these capabilities are enabled by machine
learning algorithms, in which computers analyze large amounts of data to learn statistical patterns
and develop models that can be used to make accurate predictions.
Despite the obvious opportunities for efficiency and effectiveness, the role that AI, automation
and robotics should play in government policy and service delivery remains a contentious issue.
For example, how do you prevent algorithms based on historical data from perpetuating or
reinforcing decades of conscious or unconscious bias? When is it acceptable to use “black box”
deep learning models where the logic for decisions cannot possibly be explained or understood?
To gain some insights into citizen attitudes towards AI in government, BCG surveyed more than
14,000 Internet users across 30 countries to understand citizens’ perceptions about the use of AI
by government.
Overall, we found that citizens are supportive of governments using AI for some public services
such as traffic monitoring and optimization to reduce travel times and public transport, but are
more averse to its use in more sensitive human decision making environments such as medical
diagnoses or criminal justice decisions. There are significant variations across countries, but the
level of trust in government overall appears to be a critical factor when it comes to citizen
willingness to embrace new AI technologies. Our survey suggests that governments should
continue to invest in developing the skills and capabilities to understand and leverage the potential
of AI, to provide clear frameworks around ethical and acceptable use of AI, and to engage citizens
in the journey.

Discussion Paper 2
Understanding the Citizen’s View
BCG’s global survey on citizen attitudes towards use of AI in government was conducted as part
of the bi-annual Digital Government Benchmarking study. As part of this work, we asked a broad
cross-section of citizens to tell us:
 How comfortable are they with certain activities or decisions being made by a computer,
rather than a human?
 What concerns do they have about the use of AI by governments?
 To what extent do they agree or disagree with certain statements in relation to the impact
of AI on the economy and jobs?
Key findings:
 Citizens were most supportive of using AI for tasks like transport and traffic optimization,
predictive maintenance, and customer service activities. The majority did not support AI
for decisions in the justice system such as parole board recommendations or criminal
trials.
 Less developed economies and countries where there are higher perceived levels of
corruption, also tend to be more supportive of the use of AI. For example citizens in India,
China, and Indonesia tend to be most supportive of government applications of AI, with
the least support in countries like Sweden, Denmark and Estonia.
 Demographic patterns tend to mirror general attitudes toward technology, with millennials
and urban dwellers most supportive of government use of AI, while older age groups and
more rural and remote locations less supportive.
 Citizens were most concerned about the potential moral and ethical issues and a lack of
transparency in decision making, and there is significant anxiety around the potential
automation impact of AI on jobs and work in the future.
A Cautious Optimism Prevails
Citizens are generally positive on government use of AI but the level of support varies a lot
by use case, and is relatively weak overall. Across the 13 potential use cases covered in the
survey, citizens expressed a positive net perception for all of them except for activities relating to

Discussion Paper 3
the justice system (Exhibit 1). 51% of respondents disagree with using AI to determine innocence
or guilt in a criminal trial and 46% disagree with its use for making parole decisions. There is still
a high degree of empathy, subjectivity, and complexity in these decisions that most citizens feel
still should be still be left to human judgement and discretion.
Aside from this area though, the level of support for the use of AI is higher for many core
government decision-making processes such tax and welfare administration, monitoring fraud and
non-compliance and to a lesser extent traveler and visa processing. There was strong support for
less human-intensive and sensitive decisions such as traffic and transport optimization (See Side
Bar A), predictive maintenance of infrastructure, plant and equipment such as roads, trains and
busses, and customer service channels such as virtual assistants, avatars, and virtual and
augmented reality.
There was net positive support for AI use in medical diagnosis and image recognition (51% agree)
and it presents significant opportunities to improve the speed and accuracy of medical diagnoses
(See Side Bar B). But citizens were slightly less positive on AI to make medical treatment
recommendations (net perception of 10%) and citizen attitudes to medical use cases also vary a lot
by country. For example, citizens in India were the most supportive versus Estonia the least
supportive.
Side Bar A: Traffic congestion is a major challenge in major global cities and likely
to intensify in the coming years due to population growth, urbanization, and demand
for greater mobility. Governments across the world are already using a range of
technologies and approaches to employ AI to tackle congestion on the roads, shorten
journey times, and reduce vehicle CO2 emissions. The benefits are impressive.
Pittsburgh’s AI-powered traffic lights optimize traffic flow and have reduced travel
times by 15-20% and CO2 emissions by 20%. In China, vehicle traffic monitoring
notifies authorities of incidents in real time and helps them identify high-risk areas for
congestion and collisions. This has increased traffic speed by 15% and identified
traffic violations with 92% accuracy.

Discussion Paper 4
Overall, our observation is that the level of support is relatively low and the proportion who
neither agree nor disagree is relatively high. This suggests there is a level of indifference or lack
of awareness and that governments should do more to engage, educate and communicate with the
public about the potential use cases, benefits and trade-offs. In many cases, citizens may not even
be aware that their governments are already using AI to augment or automate some of these
activities.
Copyright©2018byTheBostonConsultingGroup,Inc.Allrightsreserved.
Level of support for use of AI in government varies by use case
22
22
22
21
21
20
20
19
19
18
18
12
12
24
23
23
21
21
19
19
17
17
17
15
11
10
25
22
22
21
18
19
15
14
15
14
11
8
8
-8
-9
-8
-9
-9
-9
-5
-6
-6
-6
-8
-9
-9
-9
-12
-14
-14
-17
-27
-33
-10
-7
-7
-6
-6
-5
-5
-4
-4
-4
-3
-3
-2
-8
-7
-6
-5
-4
-4
-100 1000 50-50
Determining tax assessments and tax payable
Identifying potential fraud or non-compliance in administration of government services
“Respondents” (%)
Making parole board decisions
Predicting break-downs and maintenance requirements for machinery and equipment
Determining eligibility for visas and immigration
Making recommendations for medical treatment
Assessing travellers for additional security screening
Determining innocence or guilt in a criminal trial
Medical imaging assessment and diagnosis (e.g. identifying cancer in medical images)
A virtual assistant to provide customer service and general enquiries
Matching job seekers with available jobs
Using real-time information to predict issues, optimize and re-route traffic
Determining eligibility and amounts for welfare and social security entitlements
Response options range from 1-7, where 1 = Totally disagree, and 7 = Totally agree. All responses are shown except for '4 - neutral'. Net perception reflects total agree less total disagree
Source: BCG 2018 Digital Government Benchmarking Study
Net perception
60%
57%
AgreeDisagree
53%
46%
43%
39%
32%
23%
23%
19%
10%
-15%
-22%
Exhibit 1
Use Case
Side Bar B: AI presents significant opportunities to improve the speed and accuracy
of medical diagnoses. For example, two of the most common eye diseases are age-
related macular degeneration and diabetic retinopathy. The Optical Coherence
Tomography (OCT) scans required to diagnose these diseases are highly complex, and
professionals require years of specialized training to analyze and interpret the results.
However, after training on fewer than 15,000 scans, Google’s DeepMind AI learning
algorithms were able to make referral suggestions for more than 50 critical eye
diseases with 94 percent accuracy. AI is already achieving and in some cases
exceeding human performance, enabling faster and more accurate diagnosis and better
treatment of patients with eye conditions.

Discussion Paper 5
Understanding where support for AI in government is highest
There are significant variations in support for AI in government across countries.
People in emerging markets tend to be more positive about government use of AI.
Somewhat surprisingly, citizens in mature economies were less positive on the use of AI than
those in emerging markets. For example, the top three most supportive countries were India,
China, and Indonesia. By comparison, citizens in countries that are well recognized as digital
government leaders like Estonia, Denmark and Sweden were actually least receptive to the use of
AI. In looking into this further, with the exception of Estonia, we found a strong positive
correlation in support for government use of AI and the frequency of usage and satisfaction with
digital government services. It suggests that the more citizens engage with government digitally
and the more satisfied they are with those services, the greater confidence and trust they have.
Support for government use of AI is strongly correlated with trust in government. Trust in
institutions is essential if government is to gain the support needed to rollout AI capabilities. We
found that the countries where citizens are most supportive of AI were India, China, Indonesia,
KSA and UAE. This is closely aligned with the countries that had the highest trust in government
on the 2018 Edelman Trust Barometer: China (1st), UAE (2nd), Indonesia (3rd) and India (4th).
Our analysis also found that less developed economies and countries where there are higher
reported or perceived levels of corruption, also tend to be more supportive of the use of AI
(Exhibit 2). Saudi Arabia (57th), China (77th), India (81st) and Indonesia (96th) This could be
interpreted as a preference by citizens for machine decision making over human decision making
where there is less confidence in the machinery of government for transparency and integrity in
decision making.

Discussion Paper 6
Millennials and urban dwellers are more supportive of the use of AI (Exhibit 3). Younger
people (aged 18-34) were the most supportive of the use of AI by government, with 57%
agreeing. Younger respondents were also less likely than older age groups to disagree with the use
of AI (21% versus 29% respectively). People living in densely populated urban areas were the
most supportive of the use of AI, with 61% agreeing with its use, compared to people living in
outer suburbs (52%), town/small town residents (50%), and regional, rural or remote residents
(43%).
Younger citizens and urban dwellers also expressed the fewest worries about AI in response to the
question: “What concerns you most about the use of AI by governments?” While it can be
difficult to attribute causality to the different views across cohorts, the results tend to mirror
population attitudes towards technology in other surveys, and usually reflects relative experience,
digital maturity, and adoption rates. This suggests that support use will naturally increase over
time, but also highlights the need to target communication messages to different audiences to
address concerns and build support for adoption.
Support for AI in government moderately correlated with corruption index
100
-20
60
80
60
40
20
40200
Support for AI (%)
Corruptionindex(2017)
UK
Australia
Denmark
NL
Singapore
Norway
Switzerland
Sweden Canada
Russia
Austria
Japan
Malaysia
South Africa
New Zealand
KSA
Indonesia
France
UAE
China India
Hong Kong
Estonia USA
Kazakhstan
South Korea
Germany
Poland
Argentina
Morocco
Source: BCG 2018 Digital Government Benchmarking Survey, Transparency International 2017 Corruption Index
Exhibit 2

Discussion Paper 7
Addressing citizen concerns about AI
Resolving the moral and ethical implications of AI usage are among the top concerns of
citizens, however the overall level of concern is probably lower than expected. When asked
about the potential concerns around the use of AI by government, some 32% of citizens expressed
concerns that significant moral or ethical issues have not yet been resolved, and 25% are
concerned about the potential for bias and discrimination. The other major concerns are the
perceived lack of transparency in decision making (31%), the accuracy of the results and analysis
(25%), and the capability of the public sector to use AI (27%). However, the overall level of
concern about government use of AI at this stage appears to be relatively low. This finding could
be because we are still in the early stages of adoption or there is a low level of awareness of the
potential risks or both.
However, citizens are very concerned about the impact of AI on jobs (Exhibit 4). When asked
about the implications of AI on the economy and society, citizens expressed significant concerns
about the availability of work in the future (61% agree), the need to regulate AI to protect jobs
(58% agree), and the potential impact of AI on jobs (54% agree). Respondents are relatively
evenly divided between agreeing and disagreeing on whether their countries’ tax, education, and
welfare systems are adequately preparing them for a world in which AI is pervasive. The same is
Younger people and people in urban environments are most supportive of AI
20
19
18
16
19
19
18
17
18
17
13
15-7
-8
-7
-7
-5
-5
-6
-6
-12
-9
-16
-16
0-60 60-30 30
19
19
30
Respondents(%)
50-59
60+
18-34
35-49
36
Age Location
20
19
18
16
22
18
17
14
19
15
15
13
-8
-8
-6
-6
-6
-4
-14
-12
-9
-19
90-30-60 0 30 60
Respondents(%)
Urban
-7 -8
Rural and remote
Town or small town
Suburb
Exhibit 3
AgreeDisagree AgreeDisagree

Discussion Paper 8
true on whether governments are doing enough to manage the impact of technology on the
workforce and economy, or to engage with the community on the impact of AI.
What This Means for Government
The results of our survey point to some interesting implications for governments as they consider
how they use and make policies in relation to AI.
Use case selection and addressing ethical moral issues are paramount to assuage citizen
concerns. Citizens are clearly worried about the removal of human discretion in certain decisions,
particularly in more sensitive domains such as health or justice. Perceptions of bias and
discrimination are major factors that affect use case selection. AI has the potential reduce human
biases – both cognitive and social – that influence human decision-making. In cognitive biases,
for example, humans rely disproportionately on the first piece of information they encounter,
rather than weighing all information dispassionately. Social biases are based on prior beliefs and
worldviews, and are sometimes manifested in discrimination. Algorithms have the potential to
eliminate both noise and bias from decision-making. For example, algorithms can weigh all inputs
exactly as instructed. However, eliminating bias is very difficult. AI learns from data, much of
which has been generated from human activity. Unfortunately, those activities include human
Concern about impact of AI on future jobs and the need for associated
regulation is front of mind for citizens
0 40 60-40 20-20 80
12%
10%-14%
17%
Our education system is adequately preparing our children for the jobs of the future
8%
-12%
-12%
16% 11%
13%
-10%
13%
Governments are doing enough to engage with the community on the impact of AI
Our current tax system is adequate to cope with growth of AI -8%
18%
I am not concerned because new jobs will be created to replace those that are eliminated
-9% 16% 8%
-11%
8%
-15%
-14%Governments are doing enough to manage the impact of technology on the workforce and the economy
8%
-10%
-9%
-4%
Governments should regulate the use of AI to protect jobs
-9%
20% 18%
-12%-7%
-8% 17%
10%
-6% 19% 18%
-4%
19%
-4%
11%
15%
I am concerned about the potential impact of AI on jobs
“Respondents” (%)
I am concerned there will not be enough work for everyone in the future -5% 19% 19% 23%
-7% 20%
-5%
-13% 16%Our current welfare and social security system is adequate to protect those affected by AI -9% -13%
-10%
-8%
Net perception
43%
43%
AgreeDisagree
38%
16%
7%
5%
3%
-1%
-1%
Exhibit 4
Statements

Discussion Paper 9
bias; thus it is possible to create systems that magnify and perpetuate prejudices that are already
present. Creating models free from that bias is still a significant technical challenge. For example,
a study in the US that showed the COMPASS system (used to estimate the likelihood of
recidivism in criminal offenders) has racial biases. Moreover, the “black box” approach of AI may
not meet government audit and non-discrimination requirements.
Governments will need to select carefully where they launch pilots and eventually scale them.
They should look to launch pilots where they can test and understand the potential application of
AI to a particular use case. Once successful, the focus should shift to scaling those pilots and
sharing best practices across the organization. Balancing the potential impact for citizens, the
reusability of a use case, and the opportunity to free up resources with the difficulty of
implementation should guide governments towards those use cases that will deliver the greatest
benefit from experimentation.
Governments should also consider how to involve citizens in these pilots. Switzerland, for
example, is exploring resettling asylum seekers in different parts of the country using AI. From
late 2018, the Swiss State Secretariat for Migration and the Immigration Policy Lab (IPL) will test
a new, data-driven method for assigning asylum seekers to cantons across the country. Asylum
seekers in the pilot will be assessed using an algorithm designed to maximize their chances of
finding a job. The algorithm will allow officials to resettle individuals in the canton that best fits
their profile, rather than allocate them randomly. The program will then follow these asylum
seekers over the next several years, comparing their employment rates to those of others who
entered the country at the same time. Pilots should be publicized, tracked, and reported on, not
only to demonstrate the value of AI but also to build public trust and create transparency.
Communication and education will play a large part in building this trust as governments roll out
increasingly advanced applications of AI to their policy and delivery environments.
(Re)Building trust and integrity in government institutions. Transparency is very important to
citizens, and should be high on the agenda for government. This means being clear about the
moral and ethical implications of AI, as well as on how will be used (and will not be used).
Governments need to underscore the continued role of humans in government decision making. In
some cases, AI can be used to make decisions but in many cases AI will be used to augment
humans and support better inform decision making. Checks need to be put in place, as well as

Discussion Paper 10
mechanisms through which citizens can raise concerns. And governments should measure and be
transparent about the quality of AI recommendations.
Appropriate oversight of AI will be key, if citizens are to have confidence. Legal frameworks may
need to be created or enhanced, an AI code of ethics and compliance should be created, and
transparency is needed on the predictability of algorithms to ensure their outputs are consistent
and reliable, given their inner workings are not known, as well as their robustness against
manipulation.
Regulation needs to be carefully thought out, balancing the need to limit government use of
individuals’ personal data and AI, while allowing government to innovate in use of the
technology. Putting rules and accountability frameworks in place will reassure citizens the AI is
being used responsibly and ethically. Given the importance of AI to future economic, geopolitical,
and security positioning, it is critical that governments have the support of their citizens for its
use.
Addressing genuine concerns about the future of work is critical to building citizen
acceptance of AI. The confluence of recent developments in big data, cloud-based computer
processing power, and neural network algorithms has fueled and accelerated the developments in
AI very quickly. There has been a lot of media attention and focus on the rise of robots and the
demise of humans. It is not surprising, therefore, that the potential for job losses as a result of AI
use emerges as one of the key concerns on the minds of citizens. The fear of “technological
unemployment” as machines become dominant in the economy is real; whether the threat of this
actually occurring is real is another matter entirely. Unless governments address head-on the fears
associated with potential job insecurity and uncertainty — through public dialogue and policies
that provide a safety net for those most affected—this could create a significant barrier to AI
development.
Citizens should be supported and empowered to navigate new career pathways through life-long
learning strategies and more tailored career guidance. Governments should prepare for potential
job losses through policy measures such as the expansion of social universal safety nets, provision
of more targeted retraining and upskilling programs, and the more effective job matching and job
placement services.

Discussion Paper 11
Government needs to build AI capabilities inside the public sector. As they adopt AI,
governments need educate themselves and prepare for wider AI rollout by building internal
capabilities and setting clear data strategies such as those being pursued by Singapore (See Case
Study). The technologies are evolving rapidly, which means ministers and public servants need a
basic understanding of AI (See BCG article: Ten things every manager should know about
artificial intelligence). While this may take time through recruitment and upskilling, they can
accelerate the acquisition of expertise by embarking on partnerships with companies or start-ups.
Identifying the right mix of current and future skills will be critical to enable government
organizations to scale up their AI-related efforts.
In a similar fashion, governments will need to bolster their data management capabilities, as the
old adage “garbage in, garbage out” rings very true for AI. Ensuring the availability of accurate
and reliable data is essential for machine learning techniques to deliver the correct decisions.
Sourcing data from within the organization, or potentially an external ecosystem should be high
on the list of priorities for governments in order to best “train” AI algorithms.
Governments need to focus their attention on educating citizens, creating transparency, and
putting in place programs and policies to support the rollout of AI in government. They also need
to begin adopting AI in a thoughtful way, soliciting feedback from citizens in the process to help
build citizen support for AI along the journey.
Case Study: Singapore Smart Nation and Digital Government Group
Singapore is using AI and advanced analytics in areas such as mobility, health, and public safety.
The Smart Nation and Digital Government Group (SNDGG) was formed in May 2017 and leads
efforts across government to integrate cutting edge technological capabilities into government
service provision.
The SNDGG promotes application of AI technologies across government; coordinating across
agencies, industry, and the public, developing digital enablers and platforms; and driving the
digital transformation of the public service.

Discussion Paper 12
As more government agencies employ more sophisticated technologies, governments need to take
action on data privacy. In Singapore, the Public Sector (Governance) Bill in January 2018
formalized agency data sharing frameworks, including the conduct of regular audits, removing
personal identifiers where appropriate, limiting access to sensitive personal data, and introducing
criminal offences for data-related offences.
Government uses of AI in Singapore are diverse and include:
 Assistive Technology, Analytics and Robotics for Aging and Healthcare – RoboCoach (a
robot) helps to provide physical and cognitive therapy to seniors who have suffered
strokes or have disorders such as Alzheimer's or Parkinson's
 Smart Homes - Smart devices available in some homes include the Elderly Monitoring
System to provide peace of mind to caregivers of the elderly, and the Utility Management
System that helps manage household utilities usage
 Preventing corruption in procurement - AI algorithms analyze HR and finance data,
procurement requests, tender approvals and workflows to pick up patterns to identify and
prevent potential corruption in government
 Matching jobseekers with positions - Machine learning and text analysis identifies skills
required for jobs and prioritize search results according to the relevance of the jobseeker's
skills
 Traffic management – The Expressway Monitoring and Advisory System uses technology
to detect accidents, vehicle breakdowns, and other incidents, and provides real-time travel
time information from the entry point of the expressway to selected exits
 Lamppost-as-a-Platform - Sensors on lamp posts monitor air quality and water levels,
count electric scooters in public places, and collect footfall data to support urban and
transport planning
* * *

Discussion Paper 13
About the Authors
Miguel Carrasco is a Senior Partner and Managing Director in the Sydney office of The Boston
Consulting Group and the leader of DigitalBCG in the Public Sector practice. You may contact
him by email at Carrasco.Miguel@bcg.com
Steven Mills is an Associate Director in the DC office of The Boston Consulting Group. You may
contact him by email at Mills.Steven@bcgfed.com
Adam Whybrew is an Associate Director in the Sydney office of The Boston Consulting Group.
You may contact him by email at Whybrew.Adam@bcg.com
Adam Jura is a Knowledge Expert in the Sydney office of The Boston Consulting Group. You
may contact him by email at Jura.Adam@bcg.com

Citizens' Perspective on Use of AI in Government | 2018 BCG Digital Government Benchmark

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Citizens' Perspective on Use of AI in Government | 2018 BCG Digital Government Benchmark

Similar to Citizens' Perspective on Use of AI in Government | 2018 BCG Digital Government Benchmark (20)

Recently uploaded

Recently uploaded (20)

Citizens' Perspective on Use of AI in Government | 2018 BCG Digital Government Benchmark