AI's Loudest Job Warnings Come From the People Selling the Replacement

Dario Amodei has spent the past year warning, at length and in public, that AI could erase up to half of all entry-level white-collar jobs within five years and drive unemployment into the double digits.

I think I'm deeply uncomfortable with these decisions being made by a few companies, by a few people. — Dario Amodei

The same executive has filed, in the flat language of a product description, what it is his company sells — not a tool for any particular job, but a general substitute for human labor.

AI isn’t a substitute for specific human jobs but rather a general labor substitute for humans. — Dario Amodei

And in those same appearances he has urged the industry to slow the pace at which these models grow more capable.

We must slow the pace at which we improve the capabilities of AI models. — Dario Amodei

Nothing about that pairing requires him to be insincere. Both sentences can be true, and there is no reason to think he doubts either. The alarm and the filing are the same claim in two registers, and that doubling runs through the whole industry. Goldman Sachs has spent the year automating the layer where junior bankers learn to be senior ones, and its own AI working-group co-chair has diagnosed in public what that does to the people the machines replace — the withering of the craft a junior absorbs through apprenticeship. ^[1]

There's a huge danger here that in the era of AI, we outsource our reasoning to these models, and we have cognitive atrophy that stops us being able to reason from first principles ourselves. — Chris Churchman

The split sits inside Anthropic itself. The chief executive can see the thinning at the junior end of his own workforce — at that level, and the one above it, the company needs fewer people than before.

I can see it within Anthropic, where I can look forward to a time where on the more junior end and then on the more intermediate end we actually need less and not more people. — Dario Amodei

His co-founder and president gives the opposite answer: almost no job can be handed to a machine in full, and the distinctively human parts of the work will matter more, not less.

My worry is as this exponential keeps compounding, and I don't think it's going to take that long — again, somewhere between a year and five years — it will overwhelm our ability to adapt. — Dario Amodei

Even the rival camp reads the warnings as a performance staged for regulators.

You're being played by people who want regulatory capture. — Yann LeCun

If these warnings were brakes, the warners would brake somewhere. The record shows the reverse. The remedies were chartered in advance: Anthropic and OpenAI have both set up funds meant to spread the wealth their products concentrate, against the very job losses those products forecast. ^[2] The substitution now sits in the employers' own filings — Yahoo's disclosure states that AI adoption has already reduced its workforce and may keep reducing it. ^[3]

The adoption and deployment of AI technologies across our operations have resulted, and may continue to result, in reductions to our workforce — Yahoo!

Meta's role in its own planning is to cut as many as 20% of roles to fund AI spending, and the analysts covering the year describe the cuts as capital moving out of wages and into compute. ^[4] Block's chief executive cut roughly 40% of staff, by his own account to get ahead of the change rather than be run over by it.

Intelligence tools have changed what it means to build and run a company… A significantly smaller team, using the tools we're building, can do more and do it better. — Jack Dorsey

None of this is yet running on economics. Nvidia's own vice president concedes that for his team the compute bill — the rented computer capacity the models run on — dwarfs the payroll.

For my team, the cost of compute is far beyond the costs of the employees. — Bryan Catanzaro

Uber's CTO had exhausted his entire 2026 AI coding budget by April, and Microsoft is trimming internal licenses to the same models even as it keeps its strategic billions committed. ^[5] Jefferies' strategists can find no killer application that justifies the capital spending. ^[6] Workday's chief financial officer has quietly redefined success as the quality of decisions rather than return on investment.

It’s a whole way of identifying how we think about planning; it’s not just forecasting. — Zane Rowe

So the substitution runs ahead of its own numbers, and the financing has to come from somewhere. It comes from belief. Anthropic is carrying an $80.7 billion loss for the first half of the year into a targeted $1.5-to-$2 trillion offering, locked into $450 billion in cloud contracts it cannot cancel. ^[7] The company's own research supplies the spread that offering prices: 80% of its employees expect AI to handle their jobs by 2030, against 5% of American workers. ^[8] Its chief economist has a name for what that makes the company.

We're publishing this research because Anthropic runs roughly a year and a half ahead of the economy on AI adoption, so what happens at Anthropic could be a leading indicator for the broader world. — Peter McCrory

That is the product now — not the model, but the distance between what the people building it believe and what everyone else does. Three things will settle whether the warnings ever turn into brakes: the warners slowing their releases or withdrawing the substitution language while keeping the alarm; compute falling below the payroll it replaces at scale; the offering repricing on audited results rather than belief. All three are checkable, and none of them requires reading anyone's mind.