Anthropic Launches New Tool To Monitor Jobs AI Could Replace

Anthropic is developing an early-warning system to monitor the risk of an AI-driven downturn in employment. In a new research paper, economists Maxim Massenkoff and Peter McCrory said the artificial intelligence company has created a fresh measure of AI displacement risk called “observed exposure.”

The index combines theoretical capabilities of large language models (LLMs) with real-world usage data, placing greater weight on automated tasks rather than those where AI merely assists human workers.

The researchers said the goal is to identify potential disruption early, before large-scale effects appear in the labour market.

“This approach won't capture every channel through which AI could reshape the labour market, but by laying this groundwork now, before meaningful effects have emerged, we hope future findings will more reliably identify economic disruption than post-hoc analyses,” Massenkoff and McCrory wrote.

Evidence Of Impact “Limited” So Far

The report stated that artificial intelligence has yet to reach its full theoretical capabilities, with real-world coverage still representing only a fraction of what might eventually be possible, according to Anthropic.

According to the study, occupations with higher observed exposure are projected by the US Bureau of Labour Statistics to experience slower growth through 2034. Workers in the most exposed professions are also more likely to be older, female, more educated and higher paid. Despite those risks, the company said there is currently little evidence of widespread disruption.

Anthropic said its detection tool shows “limited evidence that AI has affected employment to date.” The researchers also found no systematic rise in unemployment among highly exposed workers since late 2022, although there are early signs that hiring of younger workers may have slowed in exposed occupations.

How The Exposure Measure Works

According to a report in Axios, the company's methodology examines three key factors: the tasks involved in a specific occupation, estimates of which of those tasks can be performed by large language models, and which of those tasks are already being carried out by AI today. Jobs are reportedly considered more exposed when their core tasks can be automated by AI and when those tasks are already being automated in practice.

The research identified several occupations with high task coverage by AI systems. These include computer programmers with 75 percent task coverage, customer service representatives at 70.1 percent, data entry keyers at 67.1 percent, medical record specialists at 66.7 percent, and market research analysts and marketing specialists at 64.8 percent.

Many Occupations Still Below Exposure Threshold

Anthropic's economists estimate that around 30 percent of occupations do not meet the minimum threshold to be considered exposed in the company's index.

“This group includes, for example, Cooks, Motorcycle Mechanics, Lifeguards, Bartenders, Dishwashers, and Dressing Room Attendants,” the company wrote.

The economists added that the index could become more valuable as AI adoption grows and the economic impact becomes harder to interpret. They said the measure will be most useful when the effects of AI disruption are “ambiguous, and could help identify the most vulnerable jobs before displacement is visible.”

Meanwhile, Anthropic chief executive Dario Amodei has warned in the past about the broader economic implications of AI technology. In January, he outlined his concerns in a 19,000-word essay titled “The adolescence of technology,” stating that AI could cause “unusually painful” disruption to the job market.