Overview

• Yale’s Budget Lab reports that 33 months after ChatGPT’s release, aggregate employment patterns remain stable and occupational shifts track close to pre‑AI eras.
• Exposure metrics and observed usage correlate weakly, and occupation-level automation or augmentation measures show no link to changes in employment or unemployment.
• Information, Financial Activities, and Professional and Business Services show larger mix changes, yet trends largely predate ChatGPT and are not directly attributable to generative AI.
• Actual generative AI use is concentrated in Computer and Mathematical roles, with Arts/Design/Media overrepresented, underscoring why exposure scores alone are poor guides.
• Analysts see only nascent, mixed signals for early-career workers; separate Stanford/ADP research found a 13% relative decline in employment for young workers in AI-exposed fields, while commentary warns a recession could hasten automation-driven cuts.