Economists argue AI's fast-moving economy needs noisy evidence now, not decades-perfect identification
JMateosGarcia · x · 2026-09-21
Alex Olegimas reflects on reading empirical AI economics papers: top journals prize clean causal identification (instruments, parallel trends), work that is vital but often takes years or decades to get right. AI will eventually get such research too, but given how fast the field moves, we also need immediate signals — papers where researchers did their best methodologically while acknowledging the field is moving too fast to wait. The argument: AI economics must pursue both rigorous-but-late and noisy-but-timely evidence in parallel.
More from AGI Musings
- Thomas Dietterich: human supervision must tighten for non-routine agent steps in volatile settings — tdietterich · 2026-09-21
- Dietterich vs Ambrosio: why isn't anyone working on continual learning? — tdietterich · 2026-09-21
- NVIDIA exec Ambrosio: human oversight may not even be necessary for AI safety — bdambrosio · 2026-09-21
- Why fear AI: a long read on Anthropic resignations and the 10% extinction-risk consensus — Jsevillamol · 2026-09-21
- Decades of automation research: autonomous systems always need human oversight — JessicaHullman · 2026-09-21
- The Era of Great Cope: Why Adapting to Superhuman AI Will Hurt, Then Pass Quickly — Dr_Singularity · 2026-09-21