New Testable AGI Definition Puts GPT-4 at 27% and GPT-5 at 58% of the Way
davidpattersonx · x · 2026-09-04
To ground the vague, moving goalpost of "AGI," @hendrycks and collaborators propose a comprehensive, testable definition and use it to quantify progress: GPT-4 (2023) scores 27% of the way to AGI, while GPT-5 (2025) reaches 58%. The thread details how the definition and measurement work. David Patterson shared it, asking how close GPT-6 will be under this standard.
More from AGI Musings
- "We already have AGI in software development": a narrow take on Astra — haider1 · 2026-09-04
- Yoav Goldberg: Every popular benchmark will be brute-forced away — so how do we measure real progress? — yoavgo · 2026-09-04
- Guillaume Verdon: real AGI test is building GTA 7 in a month, not a decade — beffjezos · 2026-09-04
- Jack Rae endorses thread: human expertise is safety infrastructure in the AI era — jachiam0 · 2026-09-04
- Blogger finds reader clicks and Google AI Overview citations barely overlap — shashib · 2026-09-04
- AI Operator: The Real Work Is in Constructing Training Data, Not the Dataset Itself — joecole · 2026-09-04