AI displacement lag

Item 5 cites new findings that challenge the assumption automation exposure causes wage losses, directly echoing the idea that early AI/automation adoption isn’t yet producing measurable job displacement in exposed occupations.

The EIG analysis highlighted shows unemployment and labor exits are not higher in AI-exposed jobs, and switching away from exposed roles has decreased.

TaxCalcBench finds frontier LLMs aren’t robust enough to be drop‑in accountant replacements, reinforcing evidence that highly exposed professions haven’t yet seen large‑scale displacement despite hype.

The piece presents counterevidence to the 'lag' thesis by citing a CEO cutting ~20% of staff due to AI, a law partner saying AI is doing 1st–3rd year associate work, and McKinsey deploying thousands of AI agents to handle junior tasks—implying near-term displacement pressure in entry-level white-collar roles.

The referenced 'big study of 100k software devs' reporting 15–20% productivity gains from AI tools without evidence of mass job loss supports the pattern that early AI adoption boosts output before measurable displacement.

By highlighting early unemployment spikes among tech/finance/computer-science graduates and citing forecasts of 10–20% unemployment, the article challenges the prevailing claim that AI adoption hasn’t yet produced measurable job losses in exposed occupations.