honestly they aren't really 'programming' the base models of those llms either. they have a pretty bland architecture and everything else emerges from there fro

honestly they aren't really 'programming' the base models of those llms either. they have a pretty bland architecture and everything else emerges from there from its training data (as much text as they could find). that's separate from the rlhf they do afterwards in 'finishing school' that makes it say things like 'as an ai language model...'. > And anyway, nobody so far has explained how neurons firing leads to consciousness. that's cool but idk how that relates to your point, unless you are proposing the neurons are magic like they are antennas for the will of god or like they really work by quantum effects that cause consciousness and are beyond the reach of our science and technology.