Extrapolating OpenAI's plots: agents may eat 25% of compute within 9-12 months
joshua_clymer · x · 2026-09-08
Joshua Clymer naively extrapolates OpenAI's published plots, with corrections in both directions:
Faster case
- OpenAI is on track to spend >25% of its compute on agent inference within 9-12 months
- In 18 months agents might autonomously complete AI R&D tasks taking a human work-year (50% success threshold)
- The extrapolation assumes no acceleration, and tasks required "0 human interventions" — real effective horizons may be underestimated
Slower case
- Doubling times at 75% success rate look much longer; that trend suggests 10 years to a 75%-reliable human-work-year agent
- AI R&D task distributions are skewed toward tasks agents are good at; value-weighted numbers would be less impressive
Conclusion: near-term progress remains genuinely ambiguous; quantitative trends are consistent with fully automated AI R&D anywhere between 2027-2030.
More from AGI Musings
- Dev: Too Many Founders Are Building Pre-AGI Companies Instead of Post-AGI Ones — willcb · 2026-09-08
- Who's more cooked? Non-technical workers, not coders, argues this AI take — brandon_galang · 2026-09-08
- Anthropic cofounder Jack Clark publishes a cryptic blog about the future — Wonderful_Buffalo_32 · 2026-09-08
- CS theory pays off daily in the AI era, dev says degree value is rising — willcb · 2026-09-08
- Sriram Krishnan: open source tools may win the AGI era thanks to training data and hackability — firstadopter · 2026-09-08
- Sentry CEO calls out AI industry: pushing browser-use while preaching safety — zeeg · 2026-09-08