Nate Silver: Give an LLM a 10,000-line program to debug — it's way more than a 'stochastic parrot'
aran_nayebi · x · 2026-09-25
Responding to a critic who claimed LLMs merely create the illusion of human-like thinking, Nate Silver argues the 'stochastic parrot' framing fails: hand Astra or Fable a 10,000-line program and ask it to debug, understand, and revise it. He admits the models aren't perfect and remains bullish on human expertise, but says their capabilities clearly exceed what the parrot label implies.
More from AGI Musings
- It's Unfolding Just as Paul Drew It: Easy vs. Hard-to-Measure Goals in AI Alignment — RichardMCNgo · 2026-09-25
- Blogger calls for futurism in schools to counter anti-datacenter TikTok propaganda — mickeyxfriedman · 2026-09-25
- Carl Feynman: chain of thought since 2022 is reasoning, critics should try modern LLMs — aran_nayebi · 2026-09-25
- AI now beats humans at some TPU design tasks, but is still seen as just a tool — burny_tech · 2026-09-25
- Hyundai to deploy 25,000 Atlas robots as Boston Dynamics opens factory training centre — VraserX · 2026-09-25
- CMU professor redesigns course assessments after AI agents ace every assignment — AustinZHenley · 2026-09-25