This doesn't work with the current architecture, because we have to introduce some element of stochastic noise into the generation or else they're not "creatively" generative.
Your brain doesn't have this problem because the noise is already present. You, as an actual thinking being, are able to override the noise and say "no, this is false." An LLM doesn't have that capability.
Your brain doesn't have this problem because the noise is already present. You, as an actual thinking being, are able to override the noise and say "no, this is false." An LLM doesn't have that capability.