As opposed to what?

added sarcasm note. it was referring to the parent.

Agree, LLMs cost millions to train and we only have a few, who knows if the next LLM will be an Einstein

Couldn't an effective theory/mean-field theory have almost any emergent behavior depending on the underlying configuration space/system? Kind of like how one can build a domain-specific language out of an expressive enough underlying language it's implemented in

Maybe, but getting fractional charges out of a system with charges that are integer multiples of the electron charge is no mean feat, and I want to understand the math behind this in detail. Someone knows, but I haven't found any clear explanations.

In theory yes, but unfortunately AI hasn't been invented yet

I don't know, shouldn't the AI then be trapped at evaluating all possible AI implementations? And since it will face the halting problem, it won't discriminate the very best one, though it will probably be able to return the best one given a capped amount of resources that is reachable through exhaustion in its space. It won't necessarily be better than what can be provided by human beings given an equivalent amount of resources.

On the flip side I always enjoyed seeing all these ideas get successfully applied in condensed matter such as https://en.wikipedia.org/wiki/Topological_insulator

While there certainly are ideas from string theory that have filtered down to CMP, topological insulators are not one of them.

That's probably true and my association of the two is probably subjective since at the time I was learning about Chern-Simons and saw it used in both contexts https://www.math.purdue.edu/~ebkaufma/RMP.pdf

Chern-Simons theory is a string theory.

I think the response to this is then to just study condensed matter then.