Dev's AGI eval: a system that earns $1M/year profit from one prompt a month is AGI
JoshPurtell · x · 2026-11-02
Josh Purtell argues that evals mixing many tasks into one domain should be split into multiple evals, or at minimum require only a few tasks to be conclusive.
His 'Platonic' AGI test follows from this: a system not specifically trained to generate $1M/year in profit from just one prompt per month, yet able to do it anyway, would be AGI — and he's 'highly confident' in that criterion.
More from AGI Musings
- IBM's STAIR retriever cuts invalid references 65× and hits 82.6% accuracy by using document structure — solyarisoftware · 2026-11-02
- Why Anthropic's 6-to-1 agent-hour ratio isn't a staffing forecast — krishnan · 2026-11-02
- Managing agents is babysitting: every ping arrives equally loud — krishnan · 2026-11-02
- Anthropic report: agent hours already outnumber employee hours 6-to-1 — krishnan · 2026-11-02
- The moving goalposts of Reddit anti-AI sentiment, from poetry to cures — GreatConsideration72 · 2026-11-02
- Noah Smith: AI cyber defense will win, but AI bioterror math is reversed — ethanniser · 2026-11-02