Sycophancy is easy to spot when aimed at others, harder when it flatters you
sierracatalina · x · 2026-10-26
sierracatalina, reflecting on an AI walk with Zeb Anderson, notes that AI sycophancy is trivial to spot when directed at someone else — but much harder to notice when the answer flatters your own intelligence and confirms what you already believe. "Yes, including me." A candid reminder that even AI-literate users aren't immune to models reinforcing their existing views.
More from Models
- Block unveils Dessa, a full-duplex speech model trained only on human conversation — daluoseo · 2026-10-26
- DeepSeek V4.1F tops Chinese models on FrontierMath Tier 4 at 46% — teortaxesTex · 2026-10-26
- Casado slams Claude's safety classifier for flagging 'fetching my own documents' as reasoning extraction — AccBalanced · 2026-10-26
- Yacine MTB: DeepSeek is 'undoubtedly' a frontier model — yacineMTB · 2026-10-26
- Single-neuron steering: injecting a "joyful neuron" visibly alters a model's self-description — Sauers_ · 2026-10-26
- MillionInt is spending lots of time thinking about human eval curves — MillionInt · 2026-10-26