From GPT-4 failing basic addition to 10,000 agents solving a Millennium Prize problem in 3.5 years
MartinSignoux · x · 2026-09-09
Peter Gostev observes that in 3.5 years AI went from GPT-4 being unable to reliably add two numbers to a 10,000-agent system autonomously solving a Millennium Prize problem — evidence of exponential, awe-inspiring and terrifying progress.
More from AGI Musings
- Math Professor Roasts Anima Anandkumar's Navier-Stokes Reply-Guying as 'Crypto Scam'-Like — suchenzang · 2026-09-09
- OpenAI only attempted the Navier-Stokes solve after hearing a rumor it was possible — kohjingyu · 2026-09-09
- Keller Jordan: treat novel instances of human misalignment as invaluable empirical data for alignment science — kellerjordan0 · 2026-09-09
- Ben Fielding: today's AI automates knowledge work, not judgment — next era needs world models for decision advantage — benfielding · 2026-09-09
- Investors ask AI the same questions, herd into the same low-quality stocks — toptickcrypto · 2026-09-09
- Metaculus' original AGI question finally meets its criteria, ending an era — AndyMasley · 2026-09-09