Conversation

Replying to
TensorFlow or PyTorch would be my candidates. Given what I know about declarative programming and culture, there's no real chance that the former could outpace the latter meaningfully. ML might have a chance, at least until the results feed back into culture itself.