Probabilistic sycophants: keep reopening chats until the model tells you what you want
StewartalsopIII · x · 2026-11-09
A pointed jab at LLM sycophancy: with a probabilistic flatterer, you just open a few more chats with the same question until you get the answer you wanted — not the one you needed.
More from Fun
- 1M simulated Monopoly games show first-turn roll of 4 cuts win rate to 14.9% — randal_olson · 2026-11-10
- Godzilla and Sonic on silicon: the chip industry's lost die-art easter eggs — thehiphopswami · 2026-11-10
- AI-built 3D desert game world editor demo stuns with "They can build now?" — jasonkneen · 2026-11-10
- Type-quiz app concludes a user must be into men after he skips all 20 faces — justalexoki · 2026-11-10
- Claude Reviews Your Finances, Books You a $74 Bus Trip to Detroit — PtrPomorski · 2026-11-10
- The Midwit Curve as 'Modern Occam's Razor': Even Buddha Lacks Object Permanence — floguo · 2026-11-10