If a model's weights were carved in stone and computed by hand over 10,000 years, is it conscious?
dansitu · x · 2026-10-03
dansitu poses a functionalist thought experiment: carve a model's weights into stone. Every thousand years, an ancient order unearths the sacred prompt on a vellum scroll and computes the next token with golden abaci — each calculation takes years, and many die or go mad in the process. After 10,000 years, the incantation is complete. The holy text reads: "Certainly! I'd be happy to help with that." Is the model conscious?
A Chinese-Room-style argument updated for the LLM era, probing whether consciousness depends on the substrate (stone, abaci, humans) or the functional structure of the computation itself.
More from AGI Musings
- Inside a Day With AI Tutor Rocky: Socratic Dialogue, Whiteboards and Spaced Repetition — RachelVT42 · 2026-10-03
- If a vector can make an LLM suffer, could another make it feel good? — rickasaurus · 2026-10-03
- One .NET job posting draws 500+ applications, almost all AI-generated spam — unixterminal · 2026-10-03
- 'AI Doomsday': Why Big Tech Wants You Scared, per Android Headlines — Nexusyak · 2026-10-03
- Rechelon: Bad 'water pipes consciousness' arguments are poisoning AI discourse — AndrewSchmidtFC · 2026-10-03
- Mathematician Kontorovich admits he was wrong about AI autonomously formalizing math — AlexKontorovich · 2026-10-03