Prolog is very, very dead. I love Prolog with all my heart, but it excells at problems that are solved today much more efficiently using neuronal networks. So it's utterly obsolete.
The issue of Prolog is that you need to code your rules manually. Doing ML with Prolog is possible, but very clumsy. Better stick to Python.
Speed is irrelevant, because most problems suitable for Prolog are exponential. Implementation is irrelevant, because SWI-Prolog does all you need with good integrations, except that it's a bit slower. But that's irrelevant, see above.
Learning Prolog is a great experience for any advanced computer science student. It amazes, doesn't it?
Prolog was never good at the things they thought it would be at, like AI, which is better done by ML today, specifically often like you said, with NNs. But it turned out to be good for other things, and those use cases are still alive today, even though there are many competitors. Look at Tiobe index, Prolog's usage is constant just under 1 percent, and has been for decades. So it's good for something.
Prolog was never designed for function approximation, like Neural Nets, so there is no comparison. Machine Learning with Prolog is perfectly possible and not at all clumsy. In fact these days we can even say it is done elegantly, by raising everything to the second order of logic where deduction and induction become one and the same.
Let me know if you need links and refs, but please try to keep your knowledge up-to-date before making big, splashy statements like "Prolog is very, very dead".
The issue of Prolog is that you need to code your rules manually. Doing ML with Prolog is possible, but very clumsy. Better stick to Python.
Speed is irrelevant, because most problems suitable for Prolog are exponential. Implementation is irrelevant, because SWI-Prolog does all you need with good integrations, except that it's a bit slower. But that's irrelevant, see above.
Learning Prolog is a great experience for any advanced computer science student. It amazes, doesn't it?