With these new abilities, we can have shared prosperity to a degree that seems unimaginable today; in the future, everyone’s lives can be better than anyone’s life is now. Prosperity alone doesn’t necessarily make people happy – there are plenty of miserable rich people – but it would meaningfully improve the lives of people around the world.
With every new language model that makes headlines, an age-old debate flares up again: Will technology close the gap between rich and poor, or does it exacerbate inequality? These two visions have been colliding since the steam engines of the 19th century. Today, with AI assistants long since taking over tasks that were once the preserve of academic professionals, the coordinate system of work, capital, and knowledge is changing more rapidly than ever before. But does “rapid” automatically mean “fairer”? The crucial question is therefore: Is AI shrinking the social divide—or is it merely enlarging the pie while the pieces remain unevenly distributed?
Democratizing Impulses – AI as a Tool for Equalization
A large-scale field study conducted at a US call center provides spectacular evidence that AI can level the playing field: a simple language model increased workforce productivity by an average of 14 percent, with the greatest gains seen among less experienced, lower-paid employees. The experienced top performers hardly improved at all – for the first time, a digital tool accelerated those who would otherwise require extensive training and narrowed the performance gap at the lower end of the scale.
Overall, access to the new tool increased worker productivity by about 14% on average, with the productivity gains concentrated among lower-skill, newer workers. For the most skilled workers, the efficiency gains from the tool were close to zero.
By using the tool, “newer workers move down the experience curve faster,” Li explained, noting that such generative AI tools learn from the behavior of better workers and disseminate those best practices to lower-skilled workers.
Macroeconomically, there are also growing indications that AI does not automatically favor high-tech elites. An OECD analysis of 19 industrialized countries has so far found no clear link between high AI exposure in a profession and growing wage gaps—on the contrary, inequality actually declined slightly within individual occupations.
Where training data is freely available globally and software only incurs copying costs, a genuine “equalizer effect” could emerge: distributed knowledge beats concentrated capital.
However, all of this must always be viewed against the backdrop that this is a snapshot in time. The analysis does not assume that AI can or will lead to widespread job losses. This assumption must be examined separately.
Concentration Tendencies – When AI Reinforces Capital
But there are two sides to every coin. A recent working paper by the International Monetary Fund warns that the same algorithms that give simple routine work a performance upgrade also increase the return on investment in highly capital-intensive industries. High-earning experts whose tasks are complementary to AI could become more productive and thus even more expensive, while owners of large data centers could tap into additional returns.
Using household microdata and a calibrated task-based model, we show these narratives reflect different channels through which AI affects the economy. Unlike previous waves of automation that increased both wage and wealth inequality, AI could reduce wage inequality through the displacement of high-income workers.
However, two factors may counter this effect: these workers’ tasks appear highly complementary with AI, potentially increasing their productivity, and they are better positioned to benefit from higher capital returns.
Economist Daron Acemoglu also puts hopes for widespread prosperity into perspective: his model calculations predict that AI will only lead to GDP growth of 1.1 to 1.6 percent by 2034 – an increase, but not a tsunami.
If overall cake growth remains moderate and capital benefits proportionally more, there is a risk of a wider wealth gap despite overall economic gains.
This also raises the question of the concentration of computing power. Computer chips are the new gold. Ownership of chips not only enables the training of new, better models, but more importantly, their inference, i.e., their use and application. In this respect, the question of democratizing control over chips is another issue that has not yet been discussed.
Labor Markets in Transition – More Jobs, Different Jobs
The World Economic Forum's Future of Jobs Report 2025 predicts a net gain of 78 million jobs by 2030, but also warns that 92 million roles will disappear. Capacity building and retraining will therefore become the bottleneck for participation.
Accordingly, the tide lifts all boats—but only those that have sufficient time and resources to set their sails in time and thus is an equilizer.
Power To Shape the Future Instead of Fate – Politics and Business at the Helm
“Historically, technology has mostly complemented rather than replaced,” emphasizes Erik Brynjolfsson in a conversation with OpenAI. Those who design AI to empower people are investing in broad prosperity; those who implement it as a replacement are cultivating exclusion.
Whether it's equalizers or a flood of performance, ultimately it's not the code alone that decides, but the institutional framework: tax incentives for continuing education, open-source models, fair data access, and progressive taxation of capital income determine whose hands the additional value flows into.
In this respect, it is not clear how this trend will develop.
Bill Gates recently pointed out that in the next decade, we will only have to work two days a week because AI will make us more productive across the board. This is a promise that John Maynard Keynes saw as certain in the 1930s, but it was never fulfilled. What is certain, however, is that productivity will increase significantly. The question that remains, however, is this: will AI, in its qualitative otherness, lead to the improvement (augmentation) or replacement of humans? Will AI and robotics be better than humans at wage labor everywhere? Depending on the answer to this question, new problems and possible solutions will arise. Is an unconditional basic income necessary? Do we need a robot tax? Do we need a computer tax? These are all questions that can be explored in a further article. Far removed from the question of job displacement, it should first be noted that the studies are inconclusive.
Conclusion
The data available so far paints a dialectical picture: AI has the potential to close productivity gaps and support wages at the lower end of the scale, while at the same time generating super profits for capital-rich players. It is neither a leveller per se nor automatically does it exacerbate inequality. Rather, it resembles a powerful stream whose channels are still being dug: where regulation, education, and open infrastructures guide it, it can irrigate fields that have been lying fallow until now. Where these guard rails are missing, it undermines the dams of social balance.
For decision-makers, this means that it is not the technology itself that is destiny, but the political will to distribute its fruits inclusively. Whether we will be talking about an egalitarian AI era in ten years' time or a new “Gold Coast” for data oligopolies depends on how consistently we adapt education systems, anchor data justice, and enable participation today. The open question is therefore no longer whether AI will transform our economy, but who will be at the helm of this transformation.
—
Ready for more content from Kim Isenberg? Subscribe to Forward Future
