Superset. No philosopher I've ever encountered believes that TMs do more than brain; always less. Non-determinstic state machines are more general than deterministic state machines, if I recollect.

Turing machines can also solve non-deterministic problems by having external feedbacks, just like humans

@JD AI is designed to solve non-deterministic problems and yet is a subset of TMs.

Well, then you're externalizing intelligence. Turing machines in the hands of poets can write poetry.

Generalization is a function of what class of problems do they address. A deterministic machine hooked up to a non-deterministic apparatus is no longer a deterministic machine any more than a person with a computer is just a person.

Actually all AIs work with external feedback, just like human.

Let us continue this discussion in chat.

Surely you know that computers produce only pseudo-random sequences, which are entirely deterministic. Yes?

@user4894 I know. But that doesn't change anything , the sequence is deterministic but the numbers are randomly distributed. Could you explain why you think we can't have non-deterministic outcomes from a deterministic sequence?

@user123456789 "Could you explain why you think we can't have non-deterministic outcomes from a deterministic sequence?" Because they're deterministic! If the sequence is deterministic, it's not non-deterministic. If by non-deterministic you mean "appearing statistically random, even if it's not actually random," then yes, that's what a pseudo-random sequence is. But it's deterministic, not non-deterministic. The two terms are mutually exclusive.

@user4894 I don't understand why you disliked my answer. Let me give you a counter example. some processes are irreversible for example multiplication. If I say 18 you can't determine whether I multiplied 2 by 9 or 6 by 3. While everything is deterministic by having the outcome of a multiplication you can't guess what numbers were multiplied. Anyway what do you mean by non-deterministic?

@user123456789 You wrote, "Every computer is able to produce enough random numbers by an algorithm called 'random generator'." I pointed out that if it's produced by an algorithm, it's not random by definition. It's deterministic. It may be pseudo-random, but it's not random. A pseudo-random process is deterministic, NOT random. Since that misunderstanding underlies your response, your answer undermines itself. Computers don't do anything randomly. See if this helps. en.wikipedia.org/wiki/Pseudorandomness

@user4894 I deliberately wrote "enough" because I knew that If we surpass the total number of the sequence then it will repeat the sequence (so deterministic). When we use pseudo-random numbers we first make sure not to let that happen. Just like having 18 but not having 6 and 3. So, I didn't deserve a downvote.

@user123456789 I didn't downvote you. You need to understand what pseudorandom means. An algorithm can never produce a random sequence, only a pseudorandom one. You keep doubling down on your misunderstanding. Instead, you need to learn the difference between random and pseudorandom. Repeating your misunderstanding doesn't help. In any event someone else downvoted you, not me.

Let us continue this discussion in chat.

@user123456789: There's a graduate school level proof that a nondeterministic Turing machine is a Turing machine. The complexity classes are the same.

@user123456789 You cannot have a non-deterministic outcome from a deterministic sequence. Full Stop. However, there is an odd line of questioning as to how long you must observe the system to distinguish the non-deterministic system from a high quality pseudorandom deterministic one. Cryptographic PRNGs typically depend on "proofs" that their output is indistinguishable from random, at least for a time period of output far longer than a human life, or even all human lives put together!

@CortAmmon "You cannot have a non-deterministic outcome from a deterministic sequence. Full Stop." Ok. What's the point? I'm not saying that the outcome of the machine is non-deterministic. I say assuming there exist a nondeterministic problem aka if nature or brain were non-deterministic , a TM can find its possible outputs. There is a wrong assumption by all of you who downvoted this answer that brain can deduce problems without externally considering the nature or AI works independent of external feedbacks. Those feedbacks are non-deterministic not the machine algorithm.

@CortAmmon Based on these downvotes I would like to edit my answer, I need the most philosophical definition of non-deterministic so that I rewrite the answer compatible with that. Could you help me defining non-deterministic with an example?