AI-exposed jobs deteriorated before ChatGPT

M. R. Frank, A. Javadian Sabet, L. Simon, S. H. Bana, R. Yu · 2026

Read AI-exposed jobs deteriorated before ChatGPT by M. R. Frank, A. Javadian Sabet, L. Simon, S. H. Bana, R. Yu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-02554.txt.

Description

Morgan R. Frank and four co authors test a widely told story, that ChatGPT's release in late 2022 damaged job prospects in occupations exposed to artificial intelligence. Using monthly United States unemployment insurance records, they measure unemployment risk by occupation and place. Risk in AI exposed occupations began rising in early 2022, months before ChatGPT existed, which breaks the simple story.

Millions of LinkedIn profiles tell the same tale from the hiring side, with graduating classes from 2021 onward entering AI exposed jobs at lower rates than earlier classes, again before the launch. A third data source, millions of course syllabi, cuts the other way for education. Graduates whose courses were more exposed to language models searched for jobs for less time after ChatGPT and earned higher first job pay in exposed occupations. The authors conclude the launch is a poor natural experiment for judging AI's labor effects.

Catalog record

CREATORM. R. Frank, A. Javadian Sabet, L. Simon, S. H. Bana, R. Yu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE24 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-02554

Browse the full library · About the Tower of Babel