We kinda sorta understand some empirical stuff, the tech is being scaled, but the theory of what's going on underneath is still resisting understanding. People who claim to have a grand theory are usually wrong, but sometimes someone finds a theory that does predict the empirical stuff. Most are mostly useful for doing more empirical stuff.
My main issue with this metaphor is that not a single chemical known to any human in 1890 was able to execute complex plans like writing code, autonomously proving complex theorems or hacking into HuggingFace. I would compare it to studying in 1944 whether a nuclear explosion can destroy the atmosphere, or with skeptics' derision of psychology as a Ptolemaic model with no attempt to find a theory.