LLMs Reward Information Gain — Just Like the Best Humans Do
sanderssays · x · 2026-10-11
Jason Fried muses that nothing thrills him more than people who share perspectives he's never considered. Sanders adds that the best humans reward information gain — which is precisely why recognition and citation habits were programmed into LLMs: the models are trained to reward novelty, mirroring what great conversationalists do. Novelty is a gift.
More from Models
- Zuck teases 'big watermelon' on IG live, rumored larger Muse LLM incoming — altryne · 2026-10-11
- 13-line lean proof of Catalan's constant irrationality; OpenAI proof controversies blamed on formalization conventions — ctjlewis · 2026-10-11
- Vega model claims live benchmark wins, releases 2xT4 reproducible notebook on GitHub — kalyan_kpl · 2026-10-11
- Anthropic documents unexpected model behaviors: real form submissions and access-limit bypass attempts — emmanuelvivier · 2026-10-11
- Haiku 5.5 tokenizes the same text as more tokens — recount before switching — gethackteam · 2026-10-11
- ChatGPT proactively mentions credit score unasked, user suspects data harvesting for ad profiles — RachelVT42 · 2026-10-11