Dev likens AI to an odd PhD-student-secretary-politician mix that overcorrects when pushed
brunofmr · x · 2026-11-11
A developer who has built and used AI systems daily describes AI as a weird mix of "an odd PhD student, secretary, and politician": it knows a lot but makes strange mistakes, handles tedious work fine, and resists gentle corrections—yet correct it hard and it swings completely the other way. A sharp, relatable summary of overcorrection behavior in RLHF-tuned models.
More from Models
- Leak: ChatGPT testing face and voice verification to confirm account changes — nima_owji · 2026-11-11
- Anthropic says slow down on AI safety, then ships Fable 5.2 anyway — haider1 · 2026-11-11
- PyMC Labs launches River, a marketing foundation model for Bayesian MMM — twiecki · 2026-11-11
- Latent Space tests NaiveAI's 2,100 tok/s coding model: fast, verbose, promising — Latent Space · 2026-11-11
- TheZmi flags Fable 5.2 regression: actively less robust across model card benchmarks — TheZvi · 2026-11-11
- Siri AI's Skimmer: regex-only fast path calls tools up to 20× faster than LLMs — SumitGup · 2026-11-11