Viral AI job-exposure dashboards misread the research: exposure ≠ displacement
soumitrashukla9 · x · 2026-09-25
CMC scholars Alex Imas and Soumitra Shukla clarify what the widely cited "AI exposure" research actually says.
- The 2023 paper "GPTs are GPTs" found 80% of U.S. workers could have at least 10% of tasks affected by LLMs, and 19% may see half or more impacted.
- But the authors were explicit: exposure means AI can be involved in the job (cutting task time by 50%+), not that the job will be automated away.
- The term has nonetheless been widely misread as displacement risk; Andrej Karpathy's weekend vibe-coded dashboard of occupational exposure went viral, feeding narratives of rapid AI-driven job loss.
- The article re-examines the economics of how AI automation actually affects jobs.
More from AGI Musings
- Ezra Klein proposes banning AI from writing code, stunning the AI community — soleio · 2026-09-25
- ICLR debate: bull-spotting reviewers opting out skews the reviewer pool — sarahwiegreffe · 2026-09-25
- ICLR reviewer pool debate: when experts opt out, bad papers gain an edge — BlancheMinerva · 2026-09-25
- Hugging Face CEO: AI risk comes from secret frontier labs, open-source is the fix — ivan_bezdomny · 2026-09-25
- Anti-AI sentiment turns to street violence, scientist calls for honest public framing — josephdviviano · 2026-09-25
- Jensen Huang flat-out refuses AI doomers, onlookers cheer: 'More of us should do this' — basedjensen · 2026-09-25