Gary Marcus's 2-year-old o3 predictions hold up: gains in verifiable domains, little elsewhere
GaryMarcus · x · 2026-11-22
Gary Marcus revisits his three predictions about o3: people would initially be amazed, find it unreliable once they dive in, and see it work best in closed domains like math but less so in open-ended real-world reasoning. The retweet notes that two years on, models show significant improvement in verifiable domains but little in non-verifiable ones — he was largely right.
More from AGI Musings
- Thiel's math-people warning lands as Claude allegedly helps crack the 167-year Riemann hypothesis — rohanpaul_ai · 2026-11-22
- If AI smashes through every bottleneck, unchecked power could arrive in 6-18 months — NathanpmYoung · 2026-11-22
- Gary Marcus Fires Back: I Predicted in Dec 2024 That Math Would Fall First Due to Verifiability — GaryMarcus · 2026-11-22
- Jeremy Howard Endorses Warning: Showing You're Just an LLM Autocomplete Gets You Replaced — jeremyphoward · 2026-11-22
- Parent of math-prodigy son: AI means his passion may have no future — erikphoel · 2026-11-22
- Harder than any benchmark: lifting GDP growth by 0.5 points — QuintinPope5 · 2026-11-22