...these machines do not directly anticipate the world. These machines anticipate how we would talk about our anticipation of the world.(Vervaeke 2025).
Gravity attenuates with the inverse-square of the distance. Newton discovered the relation and postulated the inverse-square law. Einstein’s field theory explained it. The inverse-square law falls out of the geometry of space-time. It could not have been any other way. There is a difference between describing a phenomenon and explaining it.
The progress of scientific understanding: phenomenon, over-fitting, fitting, explanation.
NASA uses Newton’s laws because they are simpler to compute and get the job done. LLM providers are aiming at the functional equivalence of human intelligence, and that is fine for their pragmatic purposes. It is not true human-level intelligence, however. We will get the functional equivalent of human-level intelligence before we get human-level intelligence.
Phenomena. Over-fitting. Fitting. Explanation.
The first three are descriptions, ever more refined, crisp, formalized ones. When we postulate a relationship that accurately describes all the instances of a phenomenon (say, the inverse-square law), then we have a standard, a fitness test, against which to test a hypothesis.
Postulates and descriptions do not always seem as such. A purported explanation can really be a sophisticated description. You read an article and the argument is not persuasive – a gut feel, an intuition – hard to describe.