TL;DR

Anthropic CEO Dario Amodei has published a 20,000-word essay warning that AI will cause an “unusually painful” short-term shock to the job market. Unlike previous technological shifts, AI’s “cognitive breadth” means it could wipe out roles across finance, consulting, law, and tech simultaneously.

Broader and Faster Than Before

Amodei’s central argument is that AI differs from previous technological revolutions because it acts as a “general labour substitute for humans” rather than targeting a single industry. Workers who might have previously retrained or moved sectors will find fewer options when multiple industries are disrupted at once.

“New technologies often bring labour market shocks, and in the past, humans have always recovered from them,” he wrote. “But AI will have effects that are much broader and occur much faster, and therefore I worry it will be much more challenging to make things work out well.”

He called for government intervention, including progressive taxation targeting AI firms, to manage the transition.

The Evidence So Far

The data paints a mixed picture. AI was cited as a reason for nearly 55,000 US layoffs in 2025, according to Challenger, Gray & Christmas. An MIT study in November found AI can already do the job of 11.7% of the US labour market, potentially saving up to $1.2 trillion in wages across finance, healthcare, and professional services.

A Mercer global survey of 12,000 people found 40% of employees now fear losing their jobs to AI, up from 28% in 2024.

However, Yale University’s Budget Lab found no significant shift in the share of workers across different occupations since ChatGPT’s debut, suggesting widespread displacement has not yet materialised. Deutsche Bank analysts have also warned that “AI redundancy washing” — companies blaming AI for job cuts with other causes — will be a “significant feature of 2026.”

Looking Forward

JPMorgan CEO Jamie Dimon has called for local-level government incentives to retrain workers and provide income assistance. Whether governments act quickly enough to match AI’s pace of development remains the open question — and the one Amodei’s essay is pressing hardest.