Key issues
Artificial intelligence (AI) has the potential to provide
several benefits, risks and challenges for the future Australian workforce.
These include:
- greater
productivity and quality of work for individual workers and workplaces
- higher
productivity across the economy more generally (but the gains may be
restricted if the development and use of AI is predominantly concentrated
within a few large corporations)
- likelihood of
unequal impacts across occupations, with possible future job loss and
creation, and changes in tasks within occupations
- widening economic
inequality if workers in some occupations benefit more than others.
Introduction
The Australian labour market has changed significantly over
the past century. New occupations have emerged, disappeared and changed,
particularly in response to technological developments. For example, the
introduction of computers has required new skills within existing and newly
created occupations. Greater mechanisation, robot technology and
computerisation has seen the decline or disappearance of occupations such as switchboard
operators, cashiers, factory workers, and data-entry clerks.
The emergence and diffusion of artificial intelligence (AI) across
the Australian economy is the latest technological phenomenon to challenge the
way we work and interact. Its impact raises several questions about what
Australia’s future labour market could look like. Will AI impact some jobs and
occupations more than others? Will it create new occupations and new jobs? And
will it benefit some workers more than others, further accentuating income and
wealth inequality?
What is AI?
There is no universally accepted definition of AI, but it is
commonly understood to be machine-based systems (AI systems) that autonomously
and adaptively make decisions or generate textual, visual or audio content
based on user inputs. In the workplace, AI systems assist in routine work such
as scheduling, resource management, document drafting and filing, data analysis
and decision-making.
Potential benefits
The
OECD has predicted that the introduction of AI should deliver greater
productivity growth across various industries and deliver new types of jobs. However, measuring
the extent of productivity gains from previous technological development has
proven difficult. Specifically, there is a lack of consensus regarding
productivity dividends from digitisation (such as increased personal computer
and digital platform use). Accordingly, the magnitude of AI’s potential
productivity benefits is difficult to quantify. Some
economists predict substantial gains, while others are less optimistic.
Oxford
Economics found the uptake of AI could spur new waves of innovation, lifting
the rate of productivity growth over the long run. It estimates that by 2032,
the US, as a leading generator of AI technology, could experience GDP growth of
1.8%–4%, depending on how quickly AI is adopted. Conversely, as more of a user (rather
than generator) of AI technology, it is estimated Australia will reap around
0.17% higher GDP by the mid-2030s.
Employee
surveys undertaken for the OECD have highlighted AI’s economic and non-economic
benefits. These include improved job performance, greater enjoyment and
positive physical and mental health outcomes resulting from replacing or
reducing menial and tedious tasks. Employees in managerial and professional
occupations in industries such as finance and insurance were more likely to
report greater satisfaction from working with AI and these occupations were
more likely to be filled by men. Employers
cite improving productivity, lowering costs and improving worker health and
safety as prime motivators for implementing AI.
Labour
economist David Autor has stated that AI has the potential to benefit
workers more broadly than other information and communications technology (ICT)
breakthroughs, which have disproportionately benefited professionals and
managers. Autor postulates that, with appropriate training, AI could enable a
larger share of the workforce to perform ‘higher-stakes decision-making tasks’
currently performed by elite experts, such as doctors, lawyers, software
engineers and college professors. This could help restore the ‘middle-skill,
middle-class heart of the U.S. labour market’ most heavily impacted by job loss
through automation and globalisation (known as job
polarisation). However, this hypothesis is not based on empirical research and
differs from the OECD’s findings that it will be highly qualified professionals
in more AI-exposed occupations who will most likely be impacted.
Potential risks associated with AI
The
OECD has also highlighted several risks associated with AI’s introduction
and diffusion. These include the dangers of market concentration and potential
widening of inequality and poverty within and between countries.
If AI is controlled by a few large firms through market
concentration, there is a danger that the benefits of AI may not translate into
higher productivity across the broader economy or deliver lower prices for
consumers. This concentration could also limit the adoption of AI by smaller
and medium-sized firms and organisations that lack the necessary resources to
adopt it. From a geopolitical perspective, countries such as the US and China with
more substantial AI investment, may reap more benefits than other less-resourced
countries.
AI could also worsen income and wealth inequality through
accelerating job loss already experienced due to automation. The more highly
educated and qualified workers achieving AI-generated productivity gains could
receive higher wage increases, increasing the potential for income and wealth inequality.
A
2023 report by McKinsey found generative AI could undertake tasks that
currently account for 30% of hours worked in the US economy. Within this model,
low-paid workers (such as clerical, retail and administrative staff) were around
14 times more likely to lose their job due to AI than higher paid workers. The
report notes that the transitioning of workers to new occupations and different
skill sets due to job loss resulting from AI may be difficult and would
disproportionately impact women.
In January 2025, the House
of Representatives Committee Inquiry into the Digital Transformation of
Workplaces released its Future
of work report, which recommended Australian governments develop
campaigns to inform people about how AI works, outline its associated opportunities
and challenges, and encourage training and upskilling in industries most
exposed to AI.
The Australian
Council of Trade Unions recognises the benefits of adopting AI in the
modern workforce, but has also raised concerns. These primarily include the
potential erosion of worker’s rights, conditions, and job security in the
absence of adequate regulation and meaningful consultation.
Industry sectors most likely to be affected
The OECD
has identified AI’s potential impact on various industry sectors,
including:
- transport (such
as self-driving or autonomous vehicles)
- agriculture (to optimise
harvesting timing and monitor conditions needed to maximise crop yields)
- financial
services (to facilitate assessing credit worthiness of potential customers and
risk of default); financial technology lending (to enable consumers to apply
for and obtain loans online); reducing regulatory compliance costs (by enabling
better fraud detection); and facilitating algorithmic market trading
- marketing and
advertising (to enable customer preferences for consumer goods to be better
matched)
- science
(convergence of AI and robotics facilitates faster scientific discoveries, reduced
costs of experimentation and improved knowledge-sharing)
- health (earlier
detection of health conditions, delivery of preventative services, optimal
clinical decision-making and new treatment and medication discovery).
Occupations and demographic groups most/least likely
affected by AI
The OECD
analysed occupational and demographic groups across 22 OECD countries
(excluding Australia) to understand AI exposure levels. The average exposure
ratios shown in Table 1 measure how closely an occupation’s required technical skills
corresponded or overlapped with AI’s technical capabilities. The higher the
ratio score, the more exposed occupations were to AI.
Managers and professional occupations recorded the highest
AI exposure scores. A strong positive relationship existed between AI exposure
and workers’ educational attainment. However, this does not predict whether
such exposure will have a positive or negative effect. AI may simply change the
nature of tasks within occupations without significant job loss or impact on
wages.
The OECD found that from 2012–22 employment growth was
strongest in occupations with higher exposure to AI. It also saw no evidence of
widening wage inequality within occupations resulting from greater exposure to
AI. The OECD found the relationship between AI exposure and gender was much
weaker, with males and females facing similar occupational exposure to AI.
Table 1 Occupational exposure to AI (%) (average
for 22 OECD countries), 2022
Occupation categories |
Average AI exposure |
Tertiary educated |
Male |
30–54 years |
Five most exposed occupations |
|
|
|
|
Science & engineering professionals |
0.84 |
87 |
69 |
67 |
Chief executives |
0.85 |
72 |
68 |
62 |
Managers |
0.86 |
76 |
59 |
73 |
Business professionals |
0.87 |
82 |
45 |
69 |
IT technology professionals |
0.88 |
79 |
81 |
70 |
Five least exposed occupations |
|
|
|
|
Cleaners, helpers |
0.25 |
9 |
18 |
56 |
Agriculture forestry & fishery labourers |
0.34 |
8 |
65 |
46 |
Food preparation assistants |
0.39 |
7 |
31 |
47 |
Labourers |
0.42 |
8 |
72 |
54 |
Refuse workers & other elementary workers |
0.43 |
10 |
72 |
49 |
All occupations |
0.65 |
37 |
57 |
60 |
Source: Marguerita Lane, Who
Will be the Workers Most Affected by AI?: A
Closer Look at the Impact of AI on Women, Low-skilled Workers and Other Groups,
OECD Artificial Intelligence Papers, Working paper no. 26, (OECD Publishing,
October 2024).
ABS Census of Population and
Housing data for 2021 and Australian
and New Zealand Standard Classification of Occupations categories allow
similar analysis for Australia (see Table 2). The data shows people employed in
occupations most exposed to AI (such as science,
engineering and ICT professionals) were more likely to be male and aged
30–54 years. However, males also accounted for larger shares of occupations that
were least exposed to AI such as farm, forestry and garden workers and other
labouring occupations.
Younger people aged 15–29 years accounted for much higher
shares of less-skilled occupations least exposed to AI. However, many young
people may be engaged in these occupations relatively briefly. This is
primarily due to them combining study with part-time jobs in occupations such
as cleaning, food preparation or labouring.
Table 2 Demographic profile (%) of occupations
with most and least exposure to AI, Australia, 2021
Occupation categories |
Tertiary educated |
Male |
15–29 years |
30–54 years |
55 years plus |
Five most exposed occupations |
|
|
|
|
|
Design, engineering, science & transport professionals |
87 |
67 |
23 |
62 |
16 |
Chief executives, general managers & legislators |
69 |
71 |
3 |
65 |
32 |
Other managers |
56 |
59 |
11 |
65 |
24 |
Business, human resource & marketing professionals |
80 |
48 |
19 |
65 |
16 |
ICT professionals |
83 |
79 |
16 |
73 |
11 |
Four least exposed occupations |
|
|
|
|
|
Cleaners & laundry workers |
22 |
40 |
22 |
49 |
30 |
Farm, forestry & garden workers |
15 |
72 |
34 |
43 |
23 |
Food preparation assistants |
15 |
52 |
66 |
23 |
11 |
Other labourers |
15 |
77 |
33 |
14 |
20 |
All occupations |
47 |
52 |
25 |
55 |
20 |
Source: Australian Bureau of Statistics, Census of Population
and Housing, 2021 (TableBuilder)
Assuming greater exposure to AI
contributes to higher wage growth, men’s higher representation in such
occupations may exacerbate measures of inequality, such as the gender
wage gap. However, if AI can be diffused into industries with higher female
representation, such as education and training and healthcare and social
assistance, this may narrow gender inequities.
Prevalence of workers with AI skills
OECD analysis of
LinkedIn members has gauged the prevalence of workers with AI skills by
gender and country from 2015 to 2023. Calculating the ratio between a country’s
proportion of workers with AI skills relative to the average for all OECD
countries canvassed, results in an AI skills penetration factor (see Figure 1).
Australia recorded an AI skills penetration factor ratio of
0.98, just below the OECD average (of 1.0). This ranked Australia 15th out of
36 OECD countries in 2023, an improvement from 19th in 2022, but slightly worse
than 14th in 2021.
Only India and the US recorded AI penetration ratios for
women above the OECD average of 1.0 (at 1.65 and 1.23 respectively). Australian
women recorded an AI penetration ratio of 0.37 which ranked them 18th out of 30
OECD countries (where estimates were available). Australian men had a reported
AI skills penetration ratio of 0.92, which ranked them 20th for males out of
the 30 OECD countries surveyed.
Australia’s relatively poor AI penetration rate, compared
with other OECD countries, indicates there may be scope for greater investment
in AI-related skills and training for Australian workers in industries more
likely to use AI.
Figure 1 AI skills penetration factor – average for 2015 to
2023
Note: A country needed at least 100,000 LinkedIn members to
be included in the survey. Due to its narrow focus on members of LinkedIn (who
are mainly professionals), it excludes the rate of AI skills penetration for
less-skilled workers.
Source: OECD.AI (2025), ‘Cross-country
AI skills penetration’.
Investment in AI in Australia and across the globe
Venture capital (VC) investment expenditure in AI can be
used to monitor growth in capital provided to start-up companies and small
businesses with long-term growth potential. OECD data shows VC investment in AI
in Australia increased substantially from A$100 million in 2017 to just under A$2.2
billion in 2021 (see Figure 2). Investment subsequently more than halved to A$906
million in 2023, but the reasons for this decline are not specified.
Figure 2 Venture capital investment in AI, Australia, 2012 to
2023 (A$ million)
Note: OECD data in USD has been converted to AUD using the
average exchange rate for each year.
Source: OECD.AI Policy Observatory, Live
data sourced from Prequin; Reserve Bank of Australia, Historical
Data, Exchange Rates.
The rise and decline in AI VC
investment in Australia aligns with trends also observed in both the US and
China, which are world leaders in AI investment. Our World Data shows global corporate investment in AI in real terms increased dramatically from US$17
billion in 2013 to US$337.4 billion in 2021, before almost halving to US$168
billion in 2023. Global private investment in AI stood at US$85 billion in 2023,
with the US contributing 70% (US$60 billion).
S&P
Global suggests that the cost of developing AI models and higher interest
rates has constrained investment in AI in recent years. As a result, larger
corporations with greater financial resources (such as Google, Amazon and Microsoft)
have dominated recent AI investment activity.
The Stanford
HIA 2024 AI Index also highlighted the significant decline in global
investment in AI in 2023, with a major reason being a fall in the level of mergers
and acquisitions. The index showed a substantial shift in global private equity
investment to AI infrastructure, research and governance in 2023 compared with
the previous year, and contracting investment in medical and health care
services.
Conclusion
International research reveals several
potential benefits and risks to the Australian economy in adopting AI. While higher
educated professionals and managers are more likely to be exposed to AI, the
overall impact of such exposure on employment and wages remains uncertain.
Aggregate productivity and efficiency benefits
from the development and diffusion of AI may be reduced if its development and
use is restricted to a few large corporations. Given these challenges, Parliament
may want to monitor this area and consider how to ensure AI’s benefits are being
distributed widely and equitably, and that workers have the skills and training
necessary to engage in this new technological landscape.
Further reading
- Francesco Filippucci, Peter Gal, Cecilia Jona-Lasinio, Alvaro
Leandro and Giuseppe Nicoletti, The Impact of Artificial Intelligence on Productivity, Distribution
and Growth: Key Mechanisms, Initial Evidence and Policy Challenges, OECD Artificial Intelligence Papers, Working paper no. 15, (OECD
Publishing, April 2024).
- Marguerita
Lane, Who
Will be the Workers Most Affected by AI?: A
Closer Look at the Impact of AI on Women, Low-skilled Workers and Other
Groups, OECD Artificial Intelligence Papers, Working paper no. 26,
(OECD Publishing, October 2024).
- David Autor, ‘AI Could Actually Help Rebuild the Middle Class’, NOĒMA, (12 February 2024).
- House of Representatives Standing Committee on Employment,
Education and Training, Inquiry into the Digital
Transformation of Workplaces, The Future of Work (Canberra:
House of Representatives, January 2025).