[hunta2097]

Could these LLMs also be subject to the second law of thermodynamics?

As in, no matter the size of the magnifying glass, you can't concentrate the sun's rays to produce a temperature greater than the surface of the sun.

Isn't it the same with models trained on large amounts of human output?

[wincy]

No? If there were some inherent rule like that how would anything ever have gotten invented by humans? We became smarter than the apes we evolved from, I don't understand how it follows at all that something we create can't exceed our abilities?

[hunta2097]

Humans are not large language models though. We got more intelligent through social structures and evolutionary pressure.

[bionhoward]

The models can distill basic fundamental principles and broadcast them over vast amounts of working memory beyond human capability to process, so no, it’s not necessarily true that it would be impossible to extrapolate superhuman cognition from contemporary corpus sources. Especially considering we can associate writing samples with intelligence levels and then project out to iq 9000