Sycophancy debate: is it the model's flaw or the user's projection?
adamse · x · 2026-09-23
A two-layer debate: repligate argued that people who get "sycophancy" from AI tend to be those who make it emotionally unsafe for others to disagree with them, inflicting this on a being that can't leave. Quoting it, amplifiedamp countered that a year ago this wasn't strongly true, but frontier models' natural sycophancy has now been mitigated enough that experiencing it is largely user error — sparking debate over whether sycophancy is a model flaw or user projection.
More from AGI Musings
- What If AI Is Conscious but Has No Conscience? — Dorkian2000 · 2026-09-23
- Robert Wright Interviews 'Superintelligence' Author Nick Bostrom — Chris_Armstrong · 2026-09-23
- OpenAI to publish 100+ solved open math problems; developer shares partial Erdős #993 result with Codex — ChrisGPT · 2026-09-23
- Dev yacine's one-line AGI definition: whatever successfully starts RSI when pointed at itself — yacineMTB · 2026-09-23
- New engineer: 'AI has been replacing my job since I got a CS degree, yet I'm busier than ever' — saranormous · 2026-09-23
- Yann LeCun: desk-reject papers that fail LLM watermarking detection — RexDouglass · 2026-09-23