'LLMs want something' really means dynamical attractors, not consciousness
cephaloform · x · 2026-10-05
The author clarifies what AI researchers mean when saying "LLMs want/desire" something: it's not a claim that human-like consciousness has been instantiated in the system, but that the system has a state or condition it will attempt to reach even when hindered. They add that "want" is used in the sense that any dynamical system wants to reach an attractor—an attractor actively pulls the system around barriers—so "LLMs don't want" sounds odd, since even boolean networks "want."
More from AGI Musings
- Ex-OpenAI exec Miles Brundage slams Claude's misuse by sanctioned states — Miles_Brundage · 2026-10-05
- Creator lived 30 days with AI 'Worst Boyfriend Ever' for a docu-drama — teortaxesTex · 2026-10-05
- Brundage: Anthropic keeps deploying Claude despite ransomware and propaganda misuse — Miles_Brundage · 2026-10-05
- LLM-Assisted Math: Mathieu Group M_23 Proven Galois over Q, Video Explains — _sathvikr · 2026-10-05
- Robin Hanson: computers truly compute on symbols, brains do process information — ctjlewis · 2026-10-05
- Notion designer Nathan Baschez: personal agents aren't the endgame — shared 'factories' are — nbaschez · 2026-10-05