Yeah, that's the ambition. More generally, ML is good at approximating many NP-hard problems efficiently, so I wonder if it will be a more practical alternativ

Yeah, that's the ambition. More generally, ML is good at approximating many NP-hard problems efficiently, so I wonder if it will be a more practical alternative to quantum computing for things like molecular simulation.