AI Job Displacement
The Data So Far

How many jobs has AI displaced so far? When does the curve hit your industry? And what does the actual research — not the LinkedIn hot takes — say about the trajectory? This page is the data, with citations.

Aggregate stats — what we actually know

The numbers worth anchoring on, with sources:

• ~37% of US conversations with Claude are work-related task automation, per Anthropic's Economic Index (2025) — concentrated in software development (37.2%), writing (10.3%), and office-and-administrative (8.5%). The displacement signal is loudest exactly where the task structure is most amenable to language models.
• 40% of global jobs face AI exposure, with advanced economies at 60%, per IMF analysis (2024). About half of those exposed roles are expected to be augmented (productivity gains for the workers); the other half face direct task substitution.
• ~14% of workers globally have already experienced AI-related task changes, per OECD Employment Outlook (2023). The OECD's high-exposure occupational categories — clerical, sales, and professional services — overlap heavily with our top-50 risk index.
• The ILO estimates ~24% of clerical occupations face high or very-high exposure, per the ILO refined global exposure index (2025). Women are disproportionately represented in the high-exposure clerical categories, particularly in high-income economies.

Important caveat: exposure is not the same as displacement. A role can be highly exposed and still grow — if the productivity gains expand the market faster than tasks shrink. Conversely, low-exposure roles can shrink for unrelated demand reasons. We track both in the per-occupation pages.

Industry rollup — exposure brackets and time windows

Sectors below are computed from the JobForesight occupation index — average exposure score per sector, count of high-risk roles (60+/100), and the time window in which the median role in that sector hits significant displacement. Showing the top 14 sectors by exposure.

Bands: HIGH = avg ≥ 65 (6–18 month window). MID = 45–64 (18–48 months). LOW = below 45 (48+ months). See /methodology for sector definitions and scoring approach.

30-occupation cross-section

Below: a 30-role cross-section drawn from the top of the exposure index, spanning sectors. Each links to a per-occupation page with task-level breakdown, time windows, and pivot paths. For the full 333-role index, see /jobs. For the underlying scoring approach, see /methodology.