AI digs through 4.4M archived posts to find badly aged takes like 'Google gave up on frontier models'
gabriberton · x · 2026-10-03
Robert Scoble had an AI search his archive of 4,445,587 X posts for badly aged hot takes.
- Standout example: in late August a user claimed "it's safe to say Google just gave up on building frontier models" — weeks later Google DeepMind introduced a new model.
- The AI transparently notes the search was broad, not exhaustive, and separates weak evidence from strong receipts.
- A fun demonstration of using agents for historical claim-checking, and a reminder of how often confident AI predictions flop.
More from AGI Musings
- Bryan Cantrill laments he may have to blog that a computer program is not a person — vivekhaldar · 2026-10-03
- Silicon Valley's AI paradox: PhDs with big-tech résumés jobless for a year or more — dotey · 2026-10-03
- Musk says growth will far exceed forecasts as energy remains the bottleneck — RachelVT42 · 2026-10-03
- LLMs Are Mathematically Equivalent to a Book, Argues Imbue Cofounder in Consciousness Debate — joshalbrecht · 2026-10-03
- Deleting the product: how sidebar bloat reveals dysfunctional team OKR culture — RachelVT42 · 2026-10-03
- How many corporate layers until humans no longer control AI-owned assets? — PierceLilholt · 2026-10-03