zlacker

>>nl+(OP)
I predict the way AI will be useful in science from the perspective of mathematics is by figuring out combinatorially complex solutions to problems that would otherwise not be interesting to (or far too complex to be comprehended by) humans. With such capabilities it could be imagined then that the AI will be useful for designing super materials, or doing fancy things with biology / medicine, and generally finding useful patterns in complex systems.

>>WhyOhW+Zo2
However, they at most would be the heuristic function of a search mechanism. A good heuristic, but heuristic at most. For search we need to identify when to abandon a path and which other entry point is promising. I'm not sure our current techniques are good for this kind of problems.