Professional Documents
Culture Documents
Public Trust in Ai
Public Trust in Ai
Public Trust in Ai
In Artificial
Intelligence
Australian insights
2020
October 2020
KPMG.com.au
2 Trust In Artificial Intelligences
KPMG Advisors
James Mabbott, Richard Boele, Ali Akbari,
Rossana Bianchi and Rita Fentener van Vlissingen.
Acknowledgements
We are grateful for the insightful input,
expertise and feedback provided by members
of the Trust, Ethics and Governance Alliance
at The University of Queensland, particularly
Dr Ida Someh, Associate Professor Martin
Edwards and Professor Matthew Hornsey,
KPMG Partners Phillip Sands, Scott Guse, Joel
Di Chiara and Leo Zhang, as well as domain
expert input by Greg Dober, Mike Richmond
and Professor Monica Janda.
Citation
Lockey, S., Gillespie, N., & Curtis, C. (2020).
Trust in Artificial Intelligence: Australian
Insights. The University of Queensland
and KPMG Australia. doi.org/10.14264/b32f129
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Contents
Executive summary 2
Introduction 5
Appendix 52
Endnotes 53
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Executive
summary
Artificial Intelligence (AI) is the
cornerstone technology of the Fourth
Industrial Revolution and is enabling
rapid innovation with many potential
benefits for Australian society (e.g.
enhanced healthcare diagnostics,
transportation optimisation) and
business (e.g. enhanced efficiency
and competitiveness). The COVID-19
pandemic has accelerated the uptake of
advanced technology, and investment in
AI continues to grow exponentially1.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Trust is central to the acceptance Australians have low trust in AI Australians expect AI to be
of AI, and is influenced by four systems but generally ‘accept’ regulated and carefully managed
key drivers or ‘tolerate’ AI
The large majority of Australians
Our results confirm that trust Trust in AI systems is low in Australia, (96%) expect AI to be regulated,
strongly influences AI acceptance. with only one in three Australians but most either disagree (45%) or
There are four key drivers that reporting that they are willing to trust are ambivalent (20%) that current
influence citizens’ trust in AI AI systems. Almost half of the public regulations and laws are sufficient
systems: 1) beliefs about the (45%) are unwilling to share their to make the use of AI safe. This
adequacy of current regulations and information or data with an AI system powerfully highlights the importance
laws to make AI use safe, 2) the and two in five (40%) are unwilling of strengthening the regulatory and
perceived uncertain impact of AI to trust the output of an AI system legal framework governing AI.
on society, 3) the perceived impact (e.g. a recommendation or decision).
Most Australians (66%) expect
of AI on jobs, and 4) familiarity and Many Australians are not convinced
government oversight of AI, with
understanding of AI. AI systems are trustworthy. However,
co-regulation between industry
they are more likely to perceive
Of these drivers, the perceived and government also a popular
AI systems as competent than
adequacy of current regulations option (60%).
designed to operate with integrity and
and laws is clearly the strongest.
humanity. All Australians expect AI governance
This demonstrates the importance
challenges to be carefully managed.
of developing adequate regulatory While many in the community
The public view data challenges
and legal mechanisms that people are hesitant to trust AI systems,
such as fake online content (70%),
believe protect them from the risks Australians generally accept (42%)
surveillance (69%), data privacy
associated with AI use. or tolerate (28%) AI, but few approve
(69%), and cyber-attacks (67%) to
(16%) or embrace (7%) AI.
be the most likely to impact large
Australians have the most numbers of Australians in the near
confidence in Australian universities future. More than half also viewed
and research institutions, as well as disease misdiagnosis (56%), HR
defence organisations, to develop bias (56%), and technological
and use (51 – 55%) and regulate unemployment (51%) as likely to
and govern AI (45 – 46%). In impact Australian society.
contrast, Australians have the
least confidence in commercial
organisations to do this. Only
about a quarter (24%) have high or
complete confidence in commercial
organisations to develop and use AI,
and less than one in five (19%) have
high or complete confidence in them
to regulate and govern AI. This may
be due to the fact that most (76%)
believe commercial organisations
innovate with AI for financial gain,
rather than for societal benefit.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Australians expect organisations Australians feel comfortable with Australians want to know more
to uphold the principles of some but not all uses of AI at work about AI but currently have low
trustworthy AI awareness and understanding
Most Australians (65 – 79%) are
of AI and its uses
The public has very clear comfortable with the use of AI at
expectations of the principles work for the purposes of monitoring Only 51% of the public have heard
and related practices they expect organisational security, and task about AI in the past year, and most
organisations deploying AI systems automation and augmentation. (61%) report a low understanding of
to uphold in order to be trusted. However, they are less comfortable AI, including how and when it is used
Australians almost unanimously with the use of AI for monitoring in everyday applications. For example,
expect AI systems to meet high and evaluating employees, or in even though 78% of Australians
standards of: recruitment and selection. report using social media, 59% of
them are unaware that social media
– performance and accuracy Most Australians (59%) disagree that
apps use AI. The good news is that
– data privacy, security AI will create more jobs than it will
most Australians (86%) want to know
and governance eliminate. In the event that their jobs
more about AI. Considered together,
are automated, Australians clearly
– transparency and explainability the results suggest there is both a
expect advanced notice (93%),
– accountability need for, and an appetite for,
retraining opportunities (92%), and
– risk and impact mitigation a public AI literacy program.
redeployment (89%). This suggests
– fairness that where AI has negative impacts,
– human oversight. Australians expect support. Meeting
these expectations will require
Most Australians (more than 70%) strategic long-range workforce
would be more willing to use AI planning and investments in re-skilling
systems if assurance mechanisms by organisations and government.
were in place, such as independent
AI ethics reviews, AI ethics
certifications, national standards
for transparency, and AI codes of
conduct. Organisations can directly
build trust and consumer willingness
to use AI systems by supporting and
implementing these mechanisms.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Introduction
Artificial Intelligence (AI) is an increasingly ubiquitous part of the
everyday lives of Australians that is transforming the way we live and
work 4. AI is used in a range of applications, such as calculating the
best travel route to take in real-time, predicting what customers will
buy, identifying credit card fraud, helping diagnose disease, identifying
people from photos, and enabling self-driving vehicles.
The benefits and promise of AI for The public’s trust in AI technologies is HR and financial investment.
society and business are undeniable. vital for its continued acceptance. If These domains represent common
AI is helping people make better AI systems do not prove to be worthy applications of AI that relate to
predictions and informed decisions, is of trust, their widespread acceptance citizens, employees and consumers.
enabling innovation, productivity gains and adoption will be hindered, and
This research provides insights into
and improved efficiency, and lowering the potential societal and economic
the key drivers of trust, community
costs. It is helping protect physical benefits will not be fully realised.
expectations and confidence in
and financial security (e.g. through
Despite the central importance of trust the regulation of AI, expectations
fraud detection) and facilitating the
for the widespread use and acceptance of the management of societal
current global fight against COVID-19.
of AI in society, to date little is known challenges associated with AI,
The risks and challenges that AI poses about the Australian community’s trust as well as Australians’ current
for society are equally undeniable. in AI or what influences it. Instead, understanding and awareness of AI.
These include the risk of codifying and current thinking has been informed by Importantly, the findings provide a
reinforcing unfair biases, infringing supposition, speculation and surveys in clear understanding of the practices
on human rights such as privacy, other jurisdictions. and principles Australians expect
spreading fake online content, organisations to use to responsibly
This national survey is designed to
technological unemployment and develop and ethically deploy AI in
understand and quantify Australians’
the dangers stemming from mass society and the workplace.
trust in and support of AI, and to
surveillance technologies, critical AI
benchmark these attitudes over time. Collectively the research insights
failures and autonomous weapons.
By taking the first deep dive into provide an evidence-based pathway
These issues are causing public
the question of trust, this research for building and maintaining the trust
concern and raising questions about
provides a comprehensive and and acceptance of AI systems by
the trustworthiness and regulation
nuanced understanding of Australians’ the Australian public. The insights
of AI systems 6.
overall trust in AI systems, as well are relevant for informing policy and
as in specific AI applications in the practice across all three sectors of
domains of healthcare, policing, government, business and non-profits.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
How we Sample
conducted 2,575
the research
Gender
51% 49%
We used a research panel to Female Male
State
NT
<1% QLD
WA
10% SA 19%
8% NSW
31%
ACT
VIC
27% 2%
TAS
3%
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Year 11 or lower
68%
Metropolitan
22%
Regional
Completed Year 12
12%
19%
Vocational/trade/ technical qualification
9%
Rural
1%
Remote
28%
Undergraduate degree (e.g. Bachelors)
31%
Postgraduate degree (e.g. Masters, PhD)
10%
Identification as an ethnic
13% have taken at least one university-level course in computer science
minority group
7% have a computer science or engineering degree
Income
3%
10% Prefer
Yes $0-$24,999
not to
say 27%
$25,000-$49,999
25%
87%
No $50,000-74,999
19%
$75,000-$99,999
43%
Current
Survey
44%
Australian National
University survey*
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Do Australians
trust AI?
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Overall, the results indicate the Australian public are on the specific purpose of the AI system. The public is
ambivalent about trusting AI. In particular, the public somewhat supportive of AI use in general. Of the specific
question whether AI is designed to operate with AI applications we examined, the public is most supportive
integrity and humanity. Further, the public’s support of its use in healthcare (e.g. disease diagnosis) and least
for the development and use of AI systems depends supportive of its use in human resources.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
As shown in Figure 2, Australians are slightly less willing Figure 2. Willingness to rely on and share
to share information with AI systems (45% unwilling, information with AI systems
mean 3.5/7), than to rely on the output of AI systems (40%
Rely on AI output
unwilling, mean 3.7/7). The same pattern emerged for trust % Unwilling
in AI systems in general as for specific AI systems. 40 29 31
% Neutral
Share information with AI system % Willing
45 27 28
Many Australians do
not view AI systems as Ability
trustworthy. However, they Relates to the perceived reliance, performance and accuracy
are more likely to perceive of AI output.
AI as competent than to be
designed to operate with
integrity and humanity. Integrity and humanity
We assessed the key components Relates to perceptions that the AI is developed based on
of trustworthiness. sound ethical principles (e.g. fairness), is transparent about
the data it collects and how it is used, and upholds the rights
of users and societal interests.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
As shown on the top half of Figure 3, when it comes to AI Figure 3. Perception of the ability, integrity
systems in general, most Australians disagree (32%) or are and humanity of AI systems
ambivalent (35%) about whether these systems operate
with integrity and humanity. In contrast, most Australians AI in general
either agree (56%) or are ambivalent (29%) about the ability
Ability
of AI systems to produce reliable, accurate output. % Disagree
15 29 56 % Neutral
For specific applications of AI (bottom of Figure 3), the
Integrity & Humanity % Agree
difference between ability and integrity/humanity is less
pronounced, with a majority of Australians either agreeing 32 35 33
or feeling ambivalent about the ability of AI to produce
accurate output (79%) and operate with integrity and Specific applications of AI systems
humanity (71%). This most likely reflects the fact that
Ability
specific applications of AI are more tangible and enable
people to evaluate trustworthiness better. In contrast, for AI 21 35 44
systems in general, people are more likely to be influenced Integrity & Humanity
by stereotypes about AI systems.
29 29 42
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
‘How much do you support or oppose the development and use of AI?’
44%
% AI
% Specific AI systems
34%
31%
26%
17%
13%
11% 11%
6% 7%
Strongly oppose Somewhat oppose Neither support nor oppose Somewhat support Strongly support
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
University Education 73
No University Education 60
CS Experience 78
No CS Experience 61
% Accept, approve or embrace
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Who do
Australians trust
to develop and
regulate AI?
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Australians are most The majority of Australians (51 – 55%) It is noteworthy that around a third
have high or complete confidence in of Australians report no or little
confident in Australian Australian universities and research confidence in government (federal
research and defence institutions and the Australian and state governments), technology
organisations to develop Defence Forces to develop and companies and commercial
and use, and regulate use AI in the best interests of the organisations. The lack of confidence in
public. In comparison, about a third technology companies and commercial
and govern AI, and least of Australians have high or complete organisations is striking given that the
confident in commercial confidence in technology companies, majority of the population’s experience
organisations. and a little over a quarter in federal and of AI is with applications developed
state governments. Australians have and used by such organisations8.
the least confidence in commercial A solution may be for commercial
organisations (see Figure 8). and technology AI companies and
government to collaborate with more
trusted entities, such as universities
and research institutions.
'How much confidence do you have in each of the following entities % Don’t know or missing
to develop and use AI in the best interests of the public?' % No or low confidence
% Moderate confidence
% High or complete confidence
Technology companies 2 31 33 34
My State Government 3 33 37 27
Commercial organisations 2 37 37 24
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Australians show a similar pattern regarding confidence in entities to regulate and govern AI in the
best interest of the public (Figure 9). Yet Australians generally show less confidence in institutions to
regulate and govern AI than to develop and use it.
Australians report higher confidence in Australian security and defence agencies and Australian university
and research institutions to regulate and govern AI than any other entity.
However, a majority also express moderate or high confidence in governments and government-funded
bodies, regulators, international organisations and scientific organisations to regulate and govern AI.
Conversely, a little over two in five Australians (43 – 45%) have no or low confidence in technology
companies or commercial organisations to regulate and govern AI.
'How much confidence do you have in each of the following % Don’t know or missing
to regulate and govern AI in the best interests of the public?' % No or low confidence
% Moderate confidence
% High or complete confidence
My State Government 2 35 35 28
AI partnership 7 28 38 27
International organisations 5 34 36 25
Technology companies 2 43 30 25
Commercial organisations 3 45 33 19
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
76% % Disagree
% Agree
67%
57%
44% 44%
34% 35%
28%
24%
13% 11%
7%
Younger people are more confident Figure 11. Age influences confidence in entities to
develop and use AI
in tech companies to develop and use AI
Younger respondents are more confident in technology % High or complete confidence
companies and independent research organisations to
develop and use AI in the best interests of the public
Gen Z & Millennials (18-39)
(see Figure 11).
43 42
Gen X (40-55)
28 28
Boomers & Greatest (56-89)
29 28
% Technology companies
% Independent research orgnisation
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
What
expectations
do Australians
have about
AI regulation?
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
66%
60% 59%
42%
4%
The Government Co-regulation Existing regulators Industry Regulation is not needed
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Current regulations are insufficient given Relatedly, the large majority (81%)
believe the impact of AI is uncertain and
the uncertainty around AI unpredictable (see Figure 14). There are
Australians generally disagree (43 – 47%) or are ambivalent no substantive differences in either of
(19 – 21%) that the current safeguards around AI (rules, these measures between AI in general
regulations and laws) are sufficient to make the use of AI safe and specific applications.
or protect them from problems (see Figure 13). Similarly, the
majority either disagree or are ambivalent that the government
adequately regulates AI.
Figure 14. Perceptions of AI
uncertainty
5 44 19 32 81%
Agree
7 44 21 28
6 45 21 28
8 47 20 25
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Figure 15. Perceived balance of risks and benefits of AI and specific AI systems
‘Overall, which best represents your view on the benefits and risks of AI systems?’
33%
31% 32%
% AI
% Specific AI systems
25%
20%
16%
14%
12%
10%
7%
Risks are much Risks are slightly Benefits and risks Benefits are slightly Benefits are much
greater than benefits greater than benefits are about equal greater than risks greater than risks
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
What principles
are important
for Australians
to trust AI
systems?
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
A proliferation of reports and guidance documents on the development and deployment of trustworthy
AI have been produced over the past few years11, with public sector institutions and private organisations
producing policy documents outlining key practices and principles.
One goal of this survey was to determine what practices and principles are important for Australians to trust
in AI. To answer this question, we asked about the importance of 40 practices associated with the eight
principles for trustworthy AI shown below. These principles were adapted primarily from the 2019 European
Commission Principles for Trustworthy AI12, as well as the Australian AI ethical principles13. Specifically,
we asked how important each of these practices are for people to trust in AI systems.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Results indicate that every one of Figure 17. Principles for Trustworthy AI
the practices underlying these eight
principles are highly important for
'How important are the following [...] for you to trust AI systems?'
the public’s trust in AI systems (see
Figure 17). This provides clear public
endorsement for these principles Technical Robustness
91
& Safety 2 7
and practices and a blueprint for
developing and using AI in a way Digital Privacy, Security
that supports trust. & Governance 2 8 90
Human Agency
& Oversight 2 9 89
Transparency
& Explainability 3 11 86
Fairness, Inclusion
& Non-discrimination 3 11 86
Accountability
& Contestability
2 13 85
AI Literacy 3 13 84
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
How do
Australians
feel about AI
at work?
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Figure 19. Comfort with the use of AI at work Australians are comfortable with AI use at
work when it is not focused on them
‘How comfortable are you with AI being used in the
following ways at work?’ As demonstrated in Figure 19, most Australians are either
highly or moderately comfortable with AI use
Security Monitoring in monitoring organisational security, and task automation
Monitoring organisations' and augmentation, such as supporting marketing
digital and physical security 21 27 52
and assisting with queries. However, they are considerably
less comfortable with AI use when it is focused
Task Automation and Augmentation
on themselves as employees – such as to monitor
Automating data pattern
detection and interpretation 24 31 45 employees, evaluate performance and support recruitment
Automating and selection decisions.
physical tasks 26 29 45
Automating administrative On average, people who report using AI in their work are
processes 29 33 38
more likely to feel comfortable with its use across various
Automating marketing
activites 30 35 35 functions than those who do not report using AI in their
Assisting with queries
work. This most likely reflects their greater familiarity with
35 30 35
the use of AI at work.
Employee-focused Activity
Supporting recruitment
and selection 46 29 25
Evaluating employee
performance 50 29 21
Monitoring employees 57 24 19
% Low comfort
% Moderate comfort
% High comfort
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Younger people Figure 20. Age differences in comfort with employee-focused AI at work
are more comfortable
with employee-focused Supporting recruitment
62
49
AI at work and selection
48
Younger people are more likely to be
comfortable with employee-focused 61
AI activities at work than older Evaluating employees 45
respondents are. For instance, over 40
half of Generation Z and Millennials are
at least moderately comfortable with 51
the use of AI to monitor employees, Monitoring employees 36
compared to just a third of older 37
respondents (see Figure 20). % Comfortable
% Gen Z and Millennials (18-39) % Gen X (40-55) % Boomers and Greatest (56-89)
Figure 21. Perceived impact of AI Figure 22. Employee expectations should their work be automated
on jobs
'In general, to what extent do you
agree that AItowill
'In general, whatcreate more
extent jobs
do you
than it will eliminate?'
agree that AI will create more jobs To be given advanced notice 3 4 93
than it will eliminate?'
To be provided with
opportunities to retrain
3 5 92
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
How do
Australians
view key AI
challenges?
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
'How important is it for companies and governments to carefully manage this challenge?'
Cyber Attack
2 97
Critical AI Failures
3 96
Data Privacy
3 96
HR Bias
4 95
Disease Misdiagnosis
4 95
Autonomous Weapons
4 94
Technological Unemployment
4 94
Autonomous Vehicles
5 93
Surveillance
4 93
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
‘In the next 10 years, how likely do you think it is that this challenge will impact large numbers of the people in Australia?'
Surveillance
13 18 69
Data Privacy
13 18 69
Cyber Attack
14 19 67
Disease Misdiagnosis
19 25 56
HR Bias
20 24 56
Technological Unemployment
29 20 51
Autonomous Vehicles
30 20 50
Critical AI Failures
30 25 45
Autonomous Weapons
50 15 35
Unlikely = 'Very unlikely (<5% chance)' ,'Unlikely (5-20% chance)' or 'Somewhat unlikely (20-40% chance)'
Equally likely as unlikely = 40-60% chance
Likely = 'Somewhat likely (60-80% chance)', 'Likely (80-95% chance)' or 'Very likely (>95% chance)'
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
How well do
Australians
understand AI?
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
'In the past 12 months, have you 'To what extent do you...
heard, read or seen anything a) feel you know a lot about AI?
about AI?' b) feel informed about how AI is used?
c) think you understand when AI is
being used?'
9%
9%
Don’t
Don’tknow
know 12%
High
27%
40%
40% 51%
51% Moderate
No
No
Yes
Yes
74%
Agree 61%
Low
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Australians report a low understanding Figure 27. Use of AI technologies and understanding
of these technologies use AI
of when AI is used
Given the low understanding of AI, it is not surprising that 'For each technology below, please indicate if you have
Australians are often unaware that AI is used in common used it and if it uses AI?'
everyday technologies. When asked if the common
technologies shown in Figure 27 use AI, less than 50%
correctly answered yes. That is, people could not correctly
identify if the technology used AI better than Accomodation
73
a chance guess. Sharing Apps 43
Australian sample. 19
Driverless Cars
We further probed people’s awareness of the emerging and Trucks
10
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
41
Gen Z & Millennials (18-39)
42
74
No Computer Science Experience
42
% Yes
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Men 49 33 18
Women 72 22 6
Gen X (40-55) 64 26 10
University Education 51 31 18
No University Education 68 24 8
CS Experience 32 39 29
No CS Experience 71 22 7
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Trust is central to the acceptance of AI systems and is influenced by four key drivers.
This model lays out a pathway to building trust and acceptance of AI.
Figure 30. A model of the key drivers of trust and acceptance of AI systems
.03
AI Uncertainty -.20
Generation .06
The belief that current The extent to which The extent to which
regulations, laws people trust AI systems people accept and
and safeguards are and perceive them to approve of AI.
sufficient to protect be trustworthy.
people and ensure AI
use is safe. This is the
strongest predictor of Other factors also had a
trust in AI systems. small impact on trust and
acceptance.
eople are slightly more
P
The belief that the trusting of AI systems in
societal impact of general than of specific
AI is unpredictable applications of AI.
and there are many
Gen Z and Millennials
unknowns about AI.
are more accepting of AI
than older generations.
The belief that AI will
create more jobs than
it will eliminate.
he extent to which
T
people feel they
understand AI, know
when AI is used in
common applications,
and have used common
AI applications.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
A key insight from the survey is that – The public clearly expect AI – Our findings further reveal
the Australian public is generally systems will be monitored and that while most Australians
ambivalent in their trust towards evaluated on an ongoing basis. are comfortable with AI use
AI systems. Given trust is a central Organisations should undertake for the purposes of protecting
factor influencing the acceptance and regular in-house and independent organisational security and task
adoption of AI24, this ambivalence ethics reviews of their AI systems automation and augmentation,
in trust is likely to impair societal to ensure they operate according they are less comfortable with
uptake and the ability of Australia to to these principles and practices. AI use for employee-focused
realise the societal and economic – Our survey revealed that most activities, such as evaluating and
benefits of AI, if left unaddressed. The people believe organisations monitoring performance, and
following insights lay out a roadmap for innovate with AI for commercial recruitment and selection.
enhancing public trust in AI. reasons (e.g. cost saving or profit – Taken together, these findings
Live up to Australians’ maximisation) rather than to highlight that organisations
expectations of trustworthy AI benefit society more broadly. This looking to accelerate the use and
imbalance is most pronounced uptake of AI need to build trust
– Our findings reveal that the public for commercial organisations, with customers, employees and
have very clear expectations of the followed by government and then the public more broadly – it is
principles and practices they expect non-profit organisations. This not enough to focus on only one
AI systems to uphold in order to be highlights the opportunity for all stakeholder group.
trusted. They expect organisations organisations to better use AI – Organisations also need to
to maintain high standards of AI systems for the benefit of citizens, consider that different cohorts in
systems in terms of: customers and employees, as well the workplace and community
– performance and accuracy as better demonstrate how their have different views about AI,
– data privacy, security use of AI supports societal health with younger people and the
and governance and wellbeing. university educated being more
– transparency and explainability – In the event their jobs are trusting and accepting of AI.
– accountability automated, employees clearly A one-size-fits-all approach is
expect to be given fair notice therefore unlikely to work.
– risk and impact mitigation
and provided with opportunities
– fairness to retrain or be redeployed.
– human oversight. Many Australians believe AI will
eliminate more jobs than it creates
These principles and practices reflect
making this a real threat. Living
those identified in numerous recent
up to employees’ expectations
government reports on trustworthy,
in the event of automation will
ethical AI25, and our findings provide
require strategic long-range
clear public endorsement for them, as
workforce planning and retraining
well as underscoring their importance
opportunities that are available to
for public trust.
employees of all ages.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Appendix
AI Systems in general versus Specific AI use cases.
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
Endnotes
1 International Data Corporation. (2019, September 14 Zhang, B., & Dafoe, A. (2019). Artificial 24 AI HLEG. (2019). Ethics Guidelines for
4). Worldwide Spending on Artificial Intelligence intelligence: American attitudes and trends. Trustworthy AI. European Commission.
Systems Will Be Nearly $98 Billion in 2023, Retrieved from SSRN 3312874. Retrieved from https://ec.europa.eu/
According to New IDC Spending Guide. Retrieved
from www.idc.com 15 Eurobarometer. (2017). Attitudes towards the Futureye. (2019). Humanising AI: Creating Social
impact of digitisation and automation on daily Licence for Artificial Intelligence to Flourish.
2 AI HLEG. (2019). Ethics Guidelines for Trustworthy life (Report no. 460). Retrieved from https:// Retrieved from https://futureye.com/
AI. European Commission. Retrieved from https:// ec.europa.eu/
ec.europa.eu/ Siau, K., & Wang, W. (2018). Building trust in
Edelman AI. (2019). 2019 Edelman AI Survey. artificial intelligence, machine learning, and
Walsh, T., Levy, N., Bell, G., Elliott, A., Maclaurin, Retrieved from https://www.edelman.com/ robotics. Cutter Business Technology Journal,
J., Mareels, I.M.Y., Wood, F.M., (2019) The effective 31(2), 47-53.
and ethical development of artificial intelligence: 16 Zhang, B., & Dafoe, A. (2019). Artificial
An opportunity to improve our wellbeing. Report intelligence: American attitudes and trends. Walsh, T., Levy, N., Bell, G., Elliott, A., Maclaurin,
for the Australian Council of Learned Academies, Retrieved from SSRN 3312874; Descriptions J., Mareels, I.M.Y., Wood, F.M., (2019) The
www.acola.org of the AI challenges were adapted from this effective and ethical development of artificial
US survey. intelligence: An opportunity to improve our
3 https://www.chiefscientist.gov.au/news-and- wellbeing. Report for the Australian Council of
media/ai-can-we-get-it-right-please 17 Zhang, B., & Dafoe, A. (2019). Artificial Learned Academies, www.acola.org.
intelligence: American attitudes and trends.
https://www.weforum.org/agenda/2018/05/alan- Retrieved from SSRN 3312874. 25 AI HLEG. (2019). Ethics Guidelines for
finkel-turing-certificate-ai-trust-robot/ Trustworthy AI. European Commission. Retrieved
18 Eurobarometer. (2017). Attitudes towards the from https://ec.europa.eu/
4 Makridakis, S. (2017). The forthcoming Artificial impact of digitisation and automation on daily
Intelligence (AI) revolution: Its impact on society life (Report no. 460). Retrieved from https:// Australia’s AI Ethics Principles Retrieved
and firms. Futures, 90, 46-60. ec.europa.eu/ from https://www.industry.gov.au/data-and-
publications/building-australias-artificial-
5 Schwab, K. (2016). The Fourth Industrial 19 Zhang, B., & Dafoe, A. (2019). Artificial intelligence-capability/ai-ethics-framework/
Revolution. World Economic Forum. intelligence: American attitudes and trends. ai-ethics-principles
Retrieved from SSRN 3312874.
6 OECD (2019), Artificial Intelligence in Society, The Assessment List for Trustworthy AI
OECD Publishing, Paris. https://doi.org/10.1787/ 20 Zhang, B., & Dafoe, A. (2019). Artificial (2020). European Commission. Retrieved from
eedfee77-en intelligence: American attitudes and trends. https://ec.europa.eu/futurium/en/ai-alliance-
Retrieved from SSRN 3312874. consultation/guidelines/2
7 Biddle, N., Edwards, B., Gray, M., Hiscox, M.,
McEachern, S., & Sollis, K. Data trust and privacy 21 Recent research undertaken by Essential For a review see Jobin, A., Ienca, M., & Vayena,
in the COVID-19 period. Retrieved from https:// Research on behalf of the Australian Human Rights E. (2019). The global landscape of AI ethics
csrm.cass.anu.edu.au/ Commission examined Australians’ awareness guidelines. Nature Machine Intelligence, 1(9),
of the federal government’s use of automated 389-399.
8 In line with prior surveys, we separate out decision-making. The findings of that survey also
technology companies (e.g.Google, Facebook, indicates greater awareness in men, the university- Fjeld, J., Achten, N., Hilligoss, H., Nagy, A.,
Apple) from other commercial organisations using educated and higher income earners. & Srikumar, M. (2020). Principled Artificial
AI, such as banks and retailers. Intelligence: Mapping Consensus in Ethical and
22 Current safeguards only influence AI acceptance Rights-Based Approaches to Principles for AI.
9 Eurobarometer. (2017). Attitudes towards the impact through trust, meaning that when current
of digitisation and automation on daily life (Report no. Berkman Klein Center Research Publication No.
safeguards are perceived as sufficient, trust in 2020-1. doi.org/10.2139/ssrn.3518482
460). Retrieved from https://ec.europa.eu/ AI systems increases, which in turn promotes
10 Zhang, B., & Dafoe, A. (2019). Artificial AI acceptance. AI uncertainty, the job impact of 26 An example is Finland’s Elements of AI course
intelligence: American attitudes and trends. AI and familiarity with AI influence acceptance run by the University of Helsinki (see https://
Retrieved from SSRN 3312874. directly, as well as through their impact on trust. www.elementsofai.com/)
11 Jobin, A., Ienca, M., & Vayena, E. (2019). The 23 University education was also included in the
global landscape of AI ethics guidelines. Nature model. However, as it did not have a significant
Machine Intelligence, 1(9), 389-399. impact on either trust or acceptance, it is not
displayed for ease of presentation.
12 AI HLEG. (2019). Ethics Guidelines for
Trustworthy AI. European Commission. Retrieved
from https://ec.europa.eu/
13 Australia’s AI Ethics Principles. Retrieved
from https://www.industry.gov.au/data-and-
publications/building-australias-artificial-
intelligence-capability/ai-ethics-framework/
ai-ethics-principles
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International Limited, a private English company
limited by guarantee. All rights reserved. The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.
KPMG
James Mabbott Ali Akbari Rita Fentener van Vlissingen
National Leader, KPMG Innovate Artificial Intelligence Associate Director, Human
KPMG Australia Capability Lead Rights & Social Impact
T: +61 2 9335 8527 KPMG Australia KPMG Australia
E: jmabbott@kpmg.com.au T: +61 2 9335 7740 T: +61 2 9346 6366
E: aakbari@kpmg.com.au E: ritafentener@kpmg.com.au
Richard Boele
Global Leader, Business Rossana Bianchi
& Human Rights Services Associate Director, Strategy,
KPMG Australia Growth & Digital
T: +61 2 9346 585 KPMG Australia
E: rboele@kpmg.com.au T: +61 2 9335 7036
E: rbianchi2@kpmg.com.au
KPMG.com.au
The information contained in this document is of a general nature and is not intended to address the objectives, financial situation or needs of any particular
individual or entity. It is provided for information purposes only and does not constitute, nor should it be regarded in any manner whatsoever, as advice and is
not intended to influence a person in making a decision, including, if applicable, in relation to any financial product or an interest in a financial product. Although
we endeavour to provide accurate and timely information, there can be no guarantee that such information is accurate as of the date it is received or that it will
continue to be accurate in the future. No one should act on such information without appropriate professional advice after a thorough examination of the particular
situation.
To the extent permissible by law, KPMG and its associated entities shall not be liable for any errors, omissions, defects or misrepresentations in the information or
for any loss or damage suffered by persons who use or rely on such information (including for reasons of negligence, negligent misstatement or otherwise).
©2020 KPMG, an Australian partnership and a member firm of the KPMG global organisation of independent member firms affiliated with KPMG International
Limited, a private English company limited by guarantee. All rights reserved.
The KPMG name and logo are trademarks used under license by the independent member firms of the KPMG global organisation.