Spinning a good narrative is critical to selling artificial intelligence these days. But the leaders of today’s biggest labs are giving us whiplash by changing their stories on employment. The most likely outcome is neither a job apocalypse nor productivity utopia, but something harder to measure: a quiet degradation of the quality of the jobs that remain.

Last year, Anthropic PBC Chief Executive Officer Dario Amodei warned AI would eliminate more than 50% of all entry-level white-collar jobs within five years. Now, he’s saying it’ll make everyone more productive. “If you automate 90% of the job, then everyone does the 10% of the job,” he told JPMorgan Chase & Co. CEO Jamie Dimon on stage at an Anthropic event last month. “And the 10% kind of expands to be 100% of what people do.”

OpenAI’s Sam Altman has also walked back his doom mongering. Having warned last year of a similar cleaving to the job market, he said in May that he’d been “pretty wrong” about AI’s economic impact.

Until now, technology leaders have rather odiously leaned into the idea of job destruction, publicly attributing layoffs at their own firms to AI displacement even though they’re more likely the result of previous over-hiring. But, as I argued at the start of this year, there’s been scant evidence of software models replacing workers on any large scale, with recent studies pointing to work-from-home policies as part of the problem for entry-level roles.

The change of perspective from Amodei and Altman is fortuitously timed. Both are prepping for initial public offerings as their companies near stratospheric private valuations of around $1 trillion each. Their target audience for pitching AI is no longer just businesses eager to harness tools that could slash labor costs and boost margins, but institutional and retail investors among a public that has become more skeptical about the tech, as well as pension funds that may be more wary of any regulatory backlash.

Lost in the binary debate about whether AI will destroy jobs or maximize productivity is the grey area that lies in between, about what might happen to the nature of work itself. The picture that some have painted can look unsettling. Zeb Evans, the CEO of software firm ClickUp, recently outlined a radical restructuring of his company around AI after cutting close to a quarter of his staff; humans would shift from producing work to managing thousands of AI agents, giving jobs to the bots and reviewing their output.

But how fulfilling is it to manage a swarm of AI agents? In her book We Are Not Machines: The Fight for the Future of Work, Sarah O’Connor argues that we shouldn’t ignore the sometimes mind-numbing nature of a new breed of jobs in which people babysit algorithms that perform the tasks they used to do. The Financial Times journalist profiles TV subtitle translators who now edit AI output, truck drivers who oversee self-driving 18 wheelers, and Amazon workers who have gone from walking across vast warehouses to find stuff to being given the products by robots and guided on which shelf or box to put them in.

While this all saves time and effort, it has a hollowing-out effect on the work itself, making it more intense, lonely and less creative. O’Connor points out that none of this is new. The early 20th century engineer Frederick Winslow Taylor famously broke factory work into tiny, standardized tasks to maximize productivity — successfully increasing efficiencies, but effectively relegating humans into components of a machine.

Many workers hated it. Resistance to Taylorism was so widespread that it triggered strikes across American factories in the early 1900s, with unions complaining that the new system was dehumanizing. It stripped employees of judgment and made their work psychologically grueling even as it increased factory output and enriched their owners.

O’Connor skirts around the question of how governments might address the issue, but she does point to a possible solution in the Netherlands. There, a nonprofit called Buurtzorg — Dutch for "neighborhood care" — has organized home nurses into small, self-managing teams of about a dozen, with virtually no middle management. Unlike a typical hospital ward or a public health system where nurses follow rigid protocols dictated from above — tracking minutes per patient, for instance — these teams decide for themselves how to allocate their time, drawing on their own expertise and the patient's informal support networks.

It is, as O'Connor puts it, the antithesis of Taylorism. Rather than breaking care into measurable micro-tasks, it trusts nurses to exercise their own judgment. The result is a decline in costs and happier patients and staff — a reminder that the problem with care work isn't always a shortage of people but how we choose to organize them.

The prognostications of Amodei and Altman on the future of work are starting to sound rosier when it comes to bare statistics, but they have yet to address how AI may make some forms of work much less meaningful by reducing human agency. That blind spot puts the onus on organizations to be more deliberate about the kind of tasks they hand over to AI. Amid their haste to avoid being left behind, they must resist the temptation to outsource tasks that make work fulfilling in the first place.

The tech CEOs may have stopped predicting the death of jobs, but no one is asking whether the jobs that survive will still be worth doing.

_____

This column reflects the personal views of the author and does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Parmy Olson is a Bloomberg Opinion columnist covering technology. A former reporter for the Wall Street Journal and Forbes, she is author of “Supremacy: AI, ChatGPT and the Race That Will Change the World.”

_____

©2026 Bloomberg L.P. Visit bloomberg.com/opinion. Distributed by Tribune Content Agency, LLC.