'AI is far from reaching its theoretical capability': Anthropic launches new tool to warn us when jobs might lost to AI

Anthropic, the AI safety company behind the Claude models, has released a new research paper detailing a framework for collecting and analyzing real-world data on artificial intelligence's impact on the labor market. The paper, which explores metrics like theoretical versus observed AI penetration across job types, aims to move beyond speculative warnings and provide empirical evidence for researchers and policymakers. Anthropic emphasizes the tool is designed not just as a job loss alarm but to help identify where workers need upskilling support, with early data suggesting AI is currently augmenting human workers rather than causing large-scale displacement.

The research identifies management, business, finance, computer/math, legal, arts/media, and office administration as the most theoretically exposed occupations. However, a key finding is that observed AI coverage in these fields remains significantly lower than its theoretical potential. Anthropic suggests future research should track how graduates navigate evolving hiring trends, as initial data shows hiring slowing for entry-level roles amid AI uncertainty. The company positions this as foundational work for proactive workforce planning, arguing more data and context are needed to understand AI's true labor market effects beyond hype cycles.