Labor market impacts of AI: A new measure and early evidence [pdf] (cdn.sanity.io)

A new study by Maxim Massenkoff and Peter McCrory introduces a novel measure of AI displacement risk known as "observed exposure," which integrates theoretical capabilities of large language models (LLMs) with actual usage data. The research reveals that while AI could potentially perform many tasks, its real-world application significantly lags behind theoretical possibilities. This measure aims to identify which occupations are most at risk of automation, highlighting that jobs with higher exposure to AI are projected to experience slower growth through 2034. Notably, the findings indicate that workers in these high-exposure fields are often older, female, more educated, and higher-paid, yet there has been no systemic spike in unemployment among them since late 2022. The significance of this study lies in its approach to forecast labor market changes due to AI, generating a framework that may more reliably predict impacts than previous models which lacked nuance. By systematically measuring the gap between capability and usage, the research could provide insights into job vulnerability ahead of noticeable economic disruption, refining strategies for workforce adaptation in an AI-driven future. As capabilities improve and adoption rates rise, understanding job exposure could help stakeholders in planning for potential transitions in employment landscapes.