Bandwagon

Luskin at EvoNews points to a new paper by establishment biologists who acknowledge that the standard theory of evolution is flawed. They propose, without actually doing it, that AI might help to sort out the question.

Unfortunately they miss the basic problem with LLMs. In current parlance, the term AI means systems like ChatGPT.

If we take the older broader meaning of artificial intelligence, covering any big software system that analyzes data and predicts results, then it’s a tautology. A data analysis system can help to analyze data, in the same way that an elevator can help to lift an object.

LLMs totally fail in the one area needed for this particular question. LLMs don’t see connections and linkages and causations at all. ChatGPT could sort out which existing articles contain more skepticism about each specific point, and we could then examine those articles without wasting time on the irrelevant stuff. The rest would still be up to human thinking.

In science, human thinking is driven by tenure and paychecks, not by facts. If you want a less biased analysis you’ll have to reshape the granting agencies and accreditation boards FIRST.

So this proposal is just hype, just a way of jumping onto a bandwagon without knowing what tune it’s playing.