Is AI poised to crush the job market like a giant robot hand crushing a cubicle worker?
Anthropic's recent report projected that large language models (LLMs) could theoretically handle 80% of job tasks across numerous professions-from Arts & Media to Management.

However, these "theoretical capabilities" stem from a 2023 study that relied heavily on speculative guesses rather than concrete evidence. Researchers used O*NET’s Detailed Work Activity reports and asked non-experts to estimate how AI might improve productivity.

Their methodology admitted significant subjectivity, with unclear logic in aggregating tasks and discrepancies in labeling.
Critically, the blue "theoretical capability" line assumes future LLM-powered software, not current models. Even the researchers warned that forecasting AI's real-world impact remains difficult.

While Anthropic found no clear job market shifts yet, it cautions that AI effects may lag behind observable trends-similar to past technological disruptions.

The key takeaway: Anthropic’s projections are based on outdated speculation, not verified performance.