3 ms·
Not an example of complex mathematical reasoning, but aren't AlphaZero and its cousins evidence that ML can independently rediscover principles that humans have
by ummonk 4y ago
Not an example of complex mathematical reasoning, but aren't AlphaZero and its cousins evidence that ML can independently rediscover principles that humans have found, as well as discover new principles of its own? For example, LC0 plays for advantages that humans hadn't considered before in Chess.
- gpderetta 4y agoVery good point. Apparently human players stared studying AIs playing go and Chess to learn new techniques.
- omnicognate 4y agoI'm not an expert in those algorithms, so... maybe? If so, maybe someone will successfully apply those ideas to the challenge I've described. I'd love to see it happen.