LinkedIn users are retroactively adding AI buzzwords while dropping remote and DEI terms

A working paper from the National Bureau of Economic Research, authored by Stanford economist Nicholas Bloom and colleagues at Stanford and Revelio Labs, analyzed 29.4 million U.S. LinkedIn profiles collected between August 2020 and January 2026. The study shows that nearly 20 % of accounts retroactively changed the title or description of a former role, with a sixfold increase in AI-related terms such as "AI," "GPT," "LLM" and "artificial intelligence" after ChatGPT’s release in late 2022.

At the same time, users became less likely to add remote-work language, and by the end of 2025 the addition and removal of such terms were roughly equal, coinciding with a wave of office-return mandates. Additions of diversity, equity, and inclusion terminology also fell sharply at the start of 2025, aligning with executive orders by President Donald Trump targeting DEI programs. The edits are most prevalent in technology and information sectors (31.6 % of workers) and tend to occur a median of four years after the job ended, suggesting they reflect perceived employer preferences rather than new experience. The authors caution that static snapshots of LinkedIn profiles may overstate the prevalence of AI skills in earlier years.