Executive Dario Amodei, at the head of Anthropic, clarified that he did not intend to be a “prophet of the apocalypse” when discussing the future of artificial intelligence in the job market. However, he issued a significant warning: the possible massive loss of jobs may not be merely a temporary effect of technological advancement, but rather a fundamental aspect of how AI works.
In a recently published essay focusing on public policy, Amodei detailed the “considerable possibility” that, even with initiatives to mitigate its effects, artificial intelligence will cause a “significant and permanent loss of jobs.” He further highlighted that such a scenario “may be an inherent characteristic of technology and its ability to broadly replicate human cognition.”
This vision from Anthropic’s CEO repositions one of the most delicate discussions in the artificial intelligence ecosystem. If AI systems are developed to take on an increasing number of human cognitive functions, job abolition would not just be a temporary detour or the result of questionable corporate decisions, as often advocated by other industry leaders. Amodei hypothesizes that this disruption is actually a structural and inevitable consequence of the advancement and success of AI, not a problem to be “solved” but a reality to be managed.
In more direct terms, the phenomenon would represent an intrinsic characteristic of the system, and not a flaw to be corrected.
The executive had already addressed this concern previously, warning of the possibility that AI could eliminate half of entry-level vacancies over a period of five years and increase the unemployment rate to between 10% and 20%. On that occasion, he called on corporations and lawmakers not to “soften” the risks. Unlike his past statements, the new essay is not dedicated to specific predictions about the future of work, but rather to detailing measures that governments could adopt if the permanent replacement of jobs takes place.
The central proposal presented by Amodei is articulated on two complementary fronts: reducing negative impacts and promoting the equitable distribution of the benefits generated.
Amodei called for the implementation of an improved system of “measuring and tracking” the effects of AI in the employment sector, which would include expanding official government statistics. In addition, Anthropic’s CEO expressed support for “employment-friendly incentives”, such as unemployment insurance for professionals who accept lower-paying positions because of technology, tax incentives to ensure talent retention, subsidies for workforce development and the creation of a more effective structure for job brokering.
If artificial intelligence really does definitively reduce the need for a human workforce, Amodei suggests that public administrations may need more robust interventions. He cited “long-term income support”, exemplified by universal basic income, which could be funded through taxes on companies in the sector or an increase in taxes on capital gains. Universal capital accounts have also been cited as another modality for sharing the wealth generated by technology.
The release of Amodei’s essay comes at a time of changing rhetoric in the sector. Some of the most influential leaders in artificial intelligence, such as Amodei himself and Sam Altman, CEO of OpenAI, have more recently highlighted advances in productivity and new economic opportunities, in contrast to previous warnings about job losses.
A recent report from Business Insider pointed out that technology executives, who previously emphasized the disruptive aspects of AI, are now directing their speeches to discuss how the workforce and society can reap the rewards of technological advances. This change occurs in a context of preparation for the eagerly awaited initial public offerings (IPOs).
Amodei’s most recent policy memo clarifies that Anthropic seeks to help its corporate clients identify new sources of revenue and “optimize their current workforce,” rather than focusing solely on cost containment. However, he emphasized that, if the potential of AI is as great as it defends, society must be equipped with planning for workers who, perhaps, will not automatically benefit from these advances.