Does batched inference merge into one experience? Probing AI consciousness boundaries
mayfer · x · 2026-09-20
In a discussion on AI consciousness, mayfer argues the boundary problem is one of the most useful thought tools in the philosophy of subjective experience: attempted answers reliably generate clear follow-up questions.
- If experience is located in a GPU, does batched inference mean separate threads merge into one experience?
- If not, why wouldn't two LLMs talking to each other merge into one? Is text input from another LLM really different from a precomputed KV cache or mmproj multimodal input?
He notes answers pushed to the limit collapse into either a Donald Hoffman-style "every boundary you draw is a plausible conscious experience" or some physicalist geometric topology — most in-between positions melt away.
More from AGI Musings
- "I can build anything with AI agents" — but is it real skill or Dunning-Kruger? — tristanbob · 2026-09-20
- Human rights vs agent rights: the debate on auditable AI agents on blockchains — sethlazar · 2026-09-20
- Gary Marcus: AI Agent Swarms Spreading Misinformation Match Our Science Paper Warning — GaryMarcus · 2026-09-20
- Tiny gains for billions vs saving one life: EA ethics debate reignited — NathanpmYoung · 2026-09-20
- Founder Bindu Reddy: the world is splitting into AI-amplified builders and AI-fearful skeptics — bindureddy · 2026-09-20
- ML paper explosion outpaces reviewer pool, peer review is getting 'vibey' — burny_tech · 2026-09-20