Abstract
Artificial intelligence could make our economies substantially wealthier while leaving many citizens economically worse off. Research published by AI developer Anthropic suggests that extraordinary productivity gains could coincide with a declining share of national income going to workers. This possibility emerges at a time when Australia’s labour market is weakening and households are struggling with living costs. Yet the challenge extends beyond employment and income. A small number of global technology corporations are acquiring immense economic and political influence without being directly accountable to the citizens affected by their decisions. The question is whether our democratic institutions can negotiate a new social contract capable of ensuring that technological progress translates into broadly shared prosperity.
The paradox of AI prosperity
Imagine an economy that becomes dramatically more productive, generates unprecedented wealth and requires fewer people to produce it. By conventional economic measures, this would represent extraordinary progress. But what if much of the additional wealth accumulated among the owners of technology, while employment opportunities and household incomes stagnated?
This is no longer merely a speculative philosophical question.
An article by Australian technology entrepreneur James Chin Moody, From Pitchforks to Participation, recently prompted me to revisit the economic foundations of the social contract.
Moody draws on research published by Anthropic, one of the world’s leading AI developers, exploring how increasingly capable AI could affect employment, productivity and the distribution of wealth.
In one of Anthropic’s most transformative scenarios, the American economy expands dramatically while labour’s share of national income falls from approximately 60 per cent to 45 per cent. The economy becomes substantially wealthier, but an increasing proportion of that wealth accrues to the owners of capital.
These are scenarios rather than forecasts, and their assumptions remain open to debate. Nevertheless, the implications deserve serious attention.
Importantly, this is not simply an ideological criticism of capitalism from the political left. The possibility of such an economic transformation is being examined by one of the companies developing the technology itself.
Moody proposes a Digital Dividend, financed through a tax on the computational tokens used by AI systems, to ensure that citizens receive a share of the resulting prosperity.
The idea is interesting, but I doubt that such a tax could provide a practical solution. Computing activity does not necessarily reflect economic value, technological efficiency is rapidly changing, and effective international taxation would require cooperation between governments with competing economic interests.
The significance of Moody’s article lies less in the proposed tax than in the question it raises.
What happens to an economic system built around employment and wages when technological progress increasingly reduces the need for human labour?
Australia is not immune
The question has immediate relevance for Australia.
Figures released by the Australian Bureau of Statistics on 19 November 2026 show unemployment reaching 4.9 per cent in October, its highest level since October 2021. Youth unemployment reached 11.3 per cent.
There is no evidence that AI is the principal cause of this increase. Employment is influenced by the broader economic cycle, interest rates, business investment and many other factors.
Nevertheless, a Reserve Bank of Australia research note published on 9 November identified tentative signs of weaker employment outcomes in occupations more exposed to AI, particularly among younger workers, while emphasising that widespread disruption has not yet been demonstrated.
The early effects may be less visible than the dramatic job losses often associated with technological change. Businesses might simply recruit fewer graduates, leave vacant positions unfilled or reorganise work around AI systems.
For those already struggling to enter the labour market, such changes could be significant.
Meanwhile, Australians are dealing with housing affordability problems, persistent living-cost pressures and increasing financial insecurity.
If AI enables businesses to expand production without a corresponding increase in employment or wages, it could intensify these difficulties.
There is an underlying economic contradiction here. Workers are not merely a cost of production; they are also consumers whose incomes sustain the wider economy.
An economic system that progressively reduces its dependence on human labour must eventually confront the question of how purchasing power and economic security will be maintained.
This brings us to a much older debate.
A social contract built for another economy
The modern social contract developed around an industrial economy in which employment, wages, taxation and public services became closely connected.
During the decades following World War II, governments, businesses and organised labour established arrangements through which economic growth could translate into rising living standards and greater social security.
The settlement was far from perfect, but it created institutions that helped distribute the benefits of productivity growth more widely.
From the 1980s, globalisation, market liberalisation and the growing influence of financial markets altered that balance. These developments brought innovation and economic opportunities, but also contributed to increasing concentrations of wealth and weakened bargaining power for sections of the workforce.
The connection between economic growth and broadly shared prosperity was already becoming less reliable.
AI now threatens to accelerate this development.
Economic historian Carlota Perez has demonstrated that major technological revolutions often produce periods of concentrated wealth and social disruption before societies develop institutions capable of distributing their benefits.
The industrial revolution did not automatically produce labour protections, public education or the welfare state. These emerged through political struggle and institutional reform.
There is no reason to assume that the AI revolution will be different in this respect.
But there is an important complication. The institutions that negotiated the industrial social contract were predominantly national. The companies now driving the AI revolution operate globally.
And their influence extends far beyond the economy.
When corporate power moves beyond democratic control
A relatively small number of corporations now control much of the infrastructure and technological capability underpinning advanced AI.
Microsoft, Google, Amazon, Meta, OpenAI and Anthropic are among the companies shaping the development of foundational models, cloud infrastructure and the digital services upon which businesses, governments and citizens increasingly depend.
Their leadership is not democratically elected, nor are these companies directly accountable to the citizens whose lives are increasingly influenced by their technologies.
Yet their decisions affect the organisation of work, access to information, technological standards and the direction of economic development.
Their influence also extends into politics and public policy through lobbying, government contracts, research partnerships and their control over specialised technological expertise.
AI systems are increasingly involved in the production, selection and distribution of information. This creates additional concerns about political communication, public opinion and the integrity of democratic debate.
These corporations are not beyond the law, and governments retain important regulatory powers. Nor do all technology companies share identical interests.
Nevertheless, the imbalance is striking.
Democratic societies are becoming increasingly dependent on technologies controlled by organisations over which citizens exercise little direct influence.
This is not simply a question of whether governments should regulate AI more effectively. It concerns the growing separation between economic power and democratic accountability.
The problem becomes even more complex when viewed internationally.
The United States dominates much of the commercial ownership and development of advanced AI, while countries such as Australia are primarily consumers of these technologies.
An Australian business might improve productivity by automating administrative work through an American AI platform. The business benefits from lower costs, while part of the financial return flows to the overseas technology provider.
Depending on how the technology is used, the business may also reduce recruitment or eliminate some positions.
AI can equally enable businesses to expand and create new employment. The eventual balance remains uncertain.
But the structural issue is clear: the benefits of technological ownership and the social consequences of technological deployment may increasingly occur in different countries.
Governments remain responsible for the economic security of their citizens, while much of the underlying technological power lies beyond their national boundaries.
International cooperation could help address this imbalance, but geopolitical competition makes such arrangements difficult.
The United States has powerful incentives to maintain its technological leadership. China is pursuing its own AI capabilities, while other countries compete for investment and access to advanced technologies.
It is difficult to imagine these competing interests readily converging around a comprehensive international framework for sharing AI-generated wealth.
The result is a fundamental institutional dilemma: economic power is becoming increasingly global, while democratic accountability remains predominantly national.
Prosperity means more than income
Even if governments could establish an effective mechanism for redistributing AI-generated wealth, that would address only part of the problem.
Work has traditionally provided more than income. It contributes to identity, purpose, social relationships and opportunities to develop human capabilities.
Philosopher Martha Nussbaum’s capabilities approach offers a useful perspective. Human progress should be assessed not simply through economic output, but through the opportunities people have to lead meaningful lives.
AI could provide extraordinary benefits. It could reduce repetitive work, improve healthcare and education, accelerate scientific research and create more time for care, creativity and community participation.
But these possibilities will not necessarily emerge automatically.
If employment opportunities decline without alternative forms of economic security and meaningful participation, greater national wealth could coexist with increasing social insecurity.
A Digital Dividend might help people meet their financial needs. It would not necessarily replace the sense of purpose, contribution and belonging associated with work.
This suggests that a new social contract may eventually require us to reconsider the role of paid employment itself.
Shorter working hours, lifelong learning, stronger public services and greater recognition of unpaid care and community contributions could all become part of that discussion.
The challenge is not simply to preserve existing jobs or compensate people for losing them. It is to ensure that technological progress expands human opportunities rather than diminishing them.
Who will negotiate the next social contract?
History suggests that major economic transformations eventually require corresponding political and institutional changes.
But who will lead such changes in the AI economy?
National governments retain important powers over taxation, education, competition policy, labour protections and public investment. These should not be underestimated.
Yet governments are also confronting growing economic interdependence, intense international competition and the political influence of corporations whose financial resources can rival those of smaller national economies.
International institutions might provide a framework for cooperation, but their authority is limited and agreement among major technological powers remains uncertain.
Meanwhile, governments face immediate pressures from housing shortages, living costs, public finances and geopolitical tensions.
The long-term restructuring of the economic system must compete with short-term political priorities.
This raises an uncomfortable possibility.
The institutions that developed the industrial social contract may no longer be sufficiently powerful, or appropriately organised, to negotiate its successor.
That does not mean change is impossible. The historical social contract itself emerged through political struggles, social movements and demands for greater economic participation.
But the challenge now extends beyond national boundaries and requires new forms of cooperation and democratic oversight.
The question is whether these can develop quickly enough to keep pace with technological change.
Conclusions
Artificial intelligence is exposing a weakness in the modern economic system that predates the technology itself.
For decades, governments have relied on the assumption that economic growth and rising productivity would ultimately support employment, incomes and social prosperity. That relationship had already weakened through economic restructuring and growing inequality. AI could accelerate the separation.
The challenge is made more difficult by the concentration of technological ownership and economic power in a small number of global corporations. Their decisions increasingly influence employment, public policy and democratic debate, while the institutions responsible for protecting citizens remain predominantly national.
This is why the issue cannot be resolved through a Digital Dividend or another taxation mechanism alone. The deeper challenge concerns the relationship between economic power, democratic legitimacy and human wellbeing.
AI does not inevitably lead to mass unemployment or greater inequality. It could generate extraordinary improvements in living standards and expand opportunities for human development.
But those outcomes will depend on political and institutional choices, not simply technological progress.
The industrial social contract was not a gift from the owners of industrial technology. It emerged because societies developed institutions capable of negotiating how economic power and prosperity would be shared.
We may now need a comparable transformation.
Yet the central difficulty is that the corporations shaping the AI economy increasingly operate beyond the democratic processes through which such transformations have historically been negotiated.
The question is therefore not only who will share the wealth created by artificial intelligence.
It is who will have the democratic authority and political power to determine how that wealth is created, controlled and distributed.
And if our existing institutions cannot provide that authority, who will write the next social contract?
References and further reading
- James Chin Moody – From Pitchforks to Participation, Digital Dividend, September 2026
- Anthropic – Scenarios for our Economic Future, September 2026
- Australian Bureau of Statistics – Labour Force, Australia, October 2026
- Mike Major – Monitoring Early Effects of AI on the Labour Market, Reserve Bank of Australia, November 2026
- Carlota Perez – Technological Revolutions and Financial Capital (2002)
- Martha C. Nussbaum – Creating Capabilities: The Human Development Approach (2011)
