What the AI revolution will certainly destroy is the illusion of a working society, one that has been maintained with great difficulty since the IT revolution of the 1980s.
The widespread use of artificial intelligence (AI) systems in late-capitalist working society – apart from a few niches in the IT industry – is still in its infancy, and the consequences of this upheaval are only beginning to emerge. Consequently, this is the era of major, often contradictory studies, which the academic community now publishes on a weekly basis. Whether the labor market is facing an apocalypse or a boom – virtually every opinion seems to be supported by research or expert statements.
In late 2025, the Massachusetts Institute of Technology (MIT) published a study indicating that approximately 11.7% of wage earners in the U.S. could already be replaced by AI systems, which would correspond to an annual payroll of around $1.2 trillion. In early 2024, the institute’s journal, MIT Technology Review, still reached the optimistic conclusion that job losses caused by agent systems and large language models could be offset by new jobs expected to emerge as part of the AI revolution. The startup Anthropic also has something for everyone: In early March, the company published a report that found no evidence that unemployment was rising due to AI, while at the same time, the business magazine Fortune, citing research from the same AI firm, warned of an impending “recession for office workers.”
A report from Stanford University identifies young people in the U.S. as being particularly at risk from these upheavals, as there are already 13% fewer entry-level jobs in AI-prone industries. Anthropic, on the other hand, concludes that older, female, well-educated, and highly paid workers are most likely to be replaced by AI. Other contributions to the discussion pointed to the potential for automation among self-employed individuals who offer online content, ranging from flashy influencers to OnlyFans porn stars. There are also empirical studies, for example from the Brookings Institution think tank, which simply conclude that it is not yet possible to draw clear conclusions about the effects of AI.
So nothing is known for certain. Nevertheless, it is possible to identify some of the contours of the coming upheavals in late-capitalist working life that are evident in virtually all studies. According to these studies, the AI revolution could trigger a surge in automation in the service sector similar to that brought about by the IT revolution of the 1980s and 1990s in industrial production. The advanced economies at the core of the world system are considered service-based societies precisely because the share of industrial workers has fallen sharply in the wake of digitalization. However, the extent of the impending job losses is disputed, as is the question of whether newly created jobs in the AI sector can mitigate them. Within the service sector, office work, customer service, creative jobs in the media industry, legal and financial services, and especially the IT sector are considered particularly at risk. The United Kingdom is already experiencing job losses due to the dominance of the financial industry, as a Morgan Stanley report concluded in early 2026, noting that AI-driven productivity gains of 11.5% had ultimately led to an 8% net loss of jobs.
The profession of programmer, in particular, appears to be at risk of job losses, as AI systems are already driving enormous short-term productivity gains in this field (even if this could lead to code instability in the long run). Estimated project hours in the industry are already shrinking from days to hours. Industry critics like Ed Zitron like to point out – and rightly so – that essential services for business customers, in particular, cannot be fully automated, but they don’t need to be. The labor market, as the arbiter of general competition, will likely ensure that rising productivity manifests itself in declining employment and compensation for IT workers, since AI tools simultaneously reduce the demands placed on the workforce.
In fact, the IT industry – which has enjoyed decades of success and remains one of the few remaining bastions of middle-class employment – is being hit by massive waves of layoffs: Oracle, Amazon, Meta, Epic – the mass layoffs, often numbering in the tens of thousands, are frequently justified this year by citing investments in AI. According to the Swedish industry publication Tech Insider, the number of layoffs in the IT sector worldwide has skyrocketed from approximately 150,000 in 2024 to around 245,000 last year, with around 264,000 projected for 2026. The justification of layoffs with investments in AI has risen sharply: from 5% in 2024 to 8% in 2025 and to around 20% in the first quarter of 2026. Tech Insider estimates that this share could rise to as much as 30% in 2026.
However, caution is warranted when looking at such figures. Oracle, for example, plans to cut 30,000 jobs, but this is not being justified by AI-driven productivity gains, but rather by cost-cutting measures aimed at raising sufficient capital for the costly construction of AI data centers. The Harvard Business Review, citing a broad survey of managers conducted in December 2025, reported that most AI-driven waves of layoffs were motivated not by actual productivity gains, but by the “anticipation of the impact of AI.”
In other words: Capital’s leadership groups are speculating on productivity gains before they can be realized. This doesn’t always work: A corporate survey conducted in late 2025 by the market research firm Forrester Research reached the sobering conclusion that 55% of decision-makers regret their AI-related layoffs. Respondents assume that either the productivity gains failed to materialize or that well-paid workers are being replaced by low-wage workers. Another study by MIT found that the use of generative AI in business applications proved unprofitable in 95% of all cases.
If AI is ever to achieve a widespread breakthrough in the economy, it will likely only happen during the socioeconomic upheavals resulting from the expected bursting of the AI bubble. Only those market participants who utilize this technology most efficiently are likely to survive the cutthroat competition of the crisis. But the question arises as to whether a new stable regime of accumulation can even be established, since AI merely drives the already well-advanced crisis process to extremes, a process fueled precisely by the internal contradiction of capital. In a sense, the elimination of artificial jobs by artificial intelligence is currently looming.
Labor is the substance of capital; in this respect, capitalism is indeed a “labor society,” and the internal contradiction of capital lies in the fact that it seeks to rid itself of this very substance through competition-driven rationalization. The first major IT revolution already made this internal contradiction acute through the decline of the industrial workforce, yet its full-scale societal impact was delayed by the financialization of capitalism in the neoliberal era of globalization, as debt – which was rising faster than economic output on a global scale – was transformed into credit-financed growth through speculative bubbles. The service sector, which now faces AI-driven upheavals, is thus artificial insofar as it could only develop within the framework of the neoliberal deficit economy. AI technology merely destroys the illusion of a working society that the neoliberal deficit economy laboriously attempted to maintain amid ever-increasing frictions in response to the IT revolution.
Without a broad employment base in industrial production, there can be no stable labor society. But there is no going back to an industrial society; the wheel of time cannot be turned back. Instead of succumbing to the general fetishization of labor – which is already recognizable in its fascistic traits, particularly in Germany – the remaining emancipatory forces must move toward a radical critique of labor that consciously and openly seeks to overcome capitalist society, which is in a state of open decay.
Originally published in Jungle World 15 in 04/2026, and then expanded and posted on konicz.info on 04/17/2026.
EXIT! in English 