Thanks for your response - I was curious if AI and tech might be able to bridge from a suitably detailed statistical picture to (at least some) cases of underlying deterministic behaviour, perhaps in a way (or ways) that might surprise us.
That's pretty much what AlphaFold has been doing for protein folding. It has been more successful than any other approach so far, but it hasn't yet "solved" protein folding, despite what some marketing materials and naive reporting has suggested. Last I heard, it was around 60% accurate when compared to experiments.
It does now seem like protein folding is within reach of being solvable, and that will be really cool and likely help advance our understanding of this part of biology, and possibly develop some new treatments for some diseases.
There will still be many more biological processes left to solve, however.
