'A simulation of fire can't burn': an argument LLMs can't be conscious
EmilMieilica · x · 2026-10-07
The author offers an argument they can't yet refute: a numerical simulation of a substrate isn't equivalent to the substrate — just as a simulation of fire can't set fire to things outside it, you only get a numerically simulated consciousness from an LLM. They concede this leaves room for 'synthetic silicon-based neurons' that could generate non-simulated consciousness. A short philosophical exchange on LLMs and consciousness.
Related event: Can Simulated Minds Be Conscious? 'Simulated Fire' Argument Sparks Debate(3 posts)→