Why LLM advice feels flat now: maybe you've built a theory of mind for the model
StewartalsopIII · x · 2026-09-02
The author argues LLM guidance has gotten noticeably flat and lays out four possible explanations:
- Opus 4.5 in chat was a different architecture, abandoned because it was too costly to scale;
- Faulty memory;
- LLMs were always bad at guidance, and the novelty of fluent conversation masked it;
- The author has developed a theory of mind of how LLMs typically respond — mimetic transfer, like how you learn to model a lover, teacher, or enemy after enough time — so the magic has worn off.
They favor #4: humans can accurately model the non-technical parts of an LLM's "mind," but never the technical parts, which is where the real value lies.
Related event: Users debate why LLM advice feels increasingly mediocre(2 posts)→
More from Models
- Scale AI Releases Muse Voice: SOTA Streaming Speech-to-Text Model — alexandr_wang · 2026-09-02
- Gemini Adds Agentic Video Understanding, Cuts Token Usage by 88% — GoogleDeepMind · 2026-09-02
- Fable 5.1 spotted in Claude support docs, release appears imminent — kimmonismus · 2026-09-02
- Meta's Muse Voice Transcribe Balances Speed and Accuracy with Adaptive Delay — AIatMeta · 2026-09-02
- Fable and Sol Ultra models accused of subtle hallucinations — StewartalsopIII · 2026-09-02
- Feed LLMs a table of pure noise and they'll confidently invent 'sensor data' — No-Plant-5234 · 2026-09-02