Google DeepMind Expert: No Evidence of AI Job Collapse, but AI Panic Could Trigger Copycat Layoffs

What Alex Imas Told the Dwarkesh Podcast About AI and White-Collar Jobs

Google DeepMind's director of AGI economics, Alex Imas, says the data he has seen do not yet show the white-collar employment collapse that many people fear from artificial intelligence. Imas, who is also a professor of economics at the University of Chicago, was asked on the Dwarkesh Podcast whether he had seen evidence that office and professional jobs are being devastated. His answer was blunt: no.

Imas is not arguing that AI will never reshape work. His point is that current evidence does not support the claim that large AI-driven job losses are already happening. He singled out software engineering, often described as one of the occupations most exposed to AI, and said he has seen little visible movement there so far. Instead, he argues, AI can raise productivity by automating parts of a job and letting workers concentrate on the tasks machines cannot do.

The more urgent warning is behavioral. Imas described a hypothetical but plausible dynamic in which companies cut staff simply to avoid being seen as slow to adopt AI. If layoffs become a way to signal that a business is AI-adapted, he argued, firms might reduce headcount even when the cuts do not improve performance. A Google DeepMind spokesperson stressed that Imas was speaking in a personal capacity and that the scenario was hypothetical.

The comments land amid pressure on executives to show investors and employees that their companies are responding to AI. Some firms, including Block and Snap, have cited AI among the reasons for job cuts. At the same time, AI leaders such as Anthropic CEO Dario Amodei have warned that AI could eventually eliminate many entry-level white-collar jobs. Imas says that may be a future risk, but it is not what the current data show.

Where the Real Risk Lies: AI-Panic Layoffs as a Corporate Signal

Google DeepMind's positioning, and the line Imas is walking

Imas holds a role at Google DeepMind, but the company moved quickly to distance its official position from the layoff-scenario warning. The spokesperson repeated CEO Demis Hassabis's more optimistic line that AI can lift productivity and create jobs. That split matters: DeepMind's commercial message is that AI is a productivity tool, while Imas's concern is that corporate psychology around AI could become self-fulfilling.

The AI-panic mechanism is about signalling, not measured productivity

The risk Imas describes is not that AI does too much work. It is that executives cut jobs to look modern. In that scenario, layoffs are a reputational play. The danger is that firms announce AI-linked reductions without evidence that automation has actually changed output, and may end up worse off than before the cuts. That distinction separates real automation-driven restructuring from AI-themed headcount reduction.

What the current absence of evidence does and does not prove

Imas's observation that software engineering shows little visible displacement is notable because programmers are repeatedly predicted to be among the first affected. But absence of evidence now is not evidence that AI will not displace roles later. The more careful reading is that the immediate white-collar collapse some commentators describe has not yet shown up in the data Imas reviewed, while productivity gains in existing jobs are more plausible in the short term.

The divide with Anthropic's warning

Anthropic CEO Dario Amodei has publicly warned that AI may take over many entry-level white-collar jobs, a much darker near-term view than Imas's reading. The disagreement is partly about evidence: Imas says current data do not show collapse, while Amodei is describing a coming displacement. For companies and workers, that split means AI job forecasts remain genuinely uncertain.

What Boards, Executives and White-Collar Workers Should Demand Before AI-Linked Cuts

For boards, executives and white-collar workers, the test is whether AI-linked staffing decisions are tied to measured automation rather than appearance.

• Boards and investors: Ask for the specific automated workload or productivity metric behind any AI-related layoff announcement. Imas's warning is that some firms may cut for optics, not because automation has actually reduced the work.
• Executive teams: Before linking headcount reductions to AI, compare the promised cost savings with the productivity loss and morale damage of copycat cuts. Imas's scenario suggests performance can worsen after a cut made mainly to signal AI adoption.
• White-collar workers in software and professional roles: Use the current absence of visible displacement as a reason to focus on the tasks AI does not automate, not as proof that no change is coming. Imas's 9-of-10-tasks example frames the short-term shift as task-level productivity rather than immediate job elimination.
• Media and analysts covering layoffs: Distinguish between companies citing AI as context and companies showing measured automation. Block and Snap are examples where the two are easily blurred.