In any computational environment(digital,neural,analog or quantum mechanical) is it possible to write or construct a program which can be construed as free will?

  • 2
    Free will is a very slippery issue... According to some, there is no free will. – Mauro ALLEGRANZA Oct 3 '17 at 13:43
  • 2
    i upvoted because an answer seems relevant – user28660 Oct 3 '17 at 14:30
  • 1
    to questions about what free will is, i mean, could shed some insight into various questions about it, disappointed in the unhelpful downvote – user28660 Oct 3 '17 at 15:27
  • 2
    There are different notions of free will (see the link given by Mauro ALLEGRANZA) and there are different ideas of what qualifies as programming (e.g., do AI systems count as programs, or do you only consider explicitly programmed systems?) - as such, this question is not really answerable without more context. – user2953 Oct 3 '17 at 17:18
  • 2
    "Is it possible to construct a complex chemical reaction with free will?" Evidently, yes... so doing the same with a computer doesn't seem unreasonable. – Ask About Monica Oct 3 '17 at 20:11

No. First consider what we mean by free will.

One sense of free will is that with enough information it is possible to predict every choice the machine, let's call it Mike, makes. There are two major problems predicting what Mike will choose, chaos and quantum physics. If the universe were perfectly deterministic, then we could theoretically build a machine to predict Mike's choices. Chaos theory introduces the problem of initial conditions, that we have to have infinite information to predict Mike's choice. Quantum physics introduces Heisenberg's Uncertainty Principle, that it's not even possible to know everything about Mike's state at any point in time.

From this one might gather that free will is therefore true, but this is incorrect. Just because it is impossible to know the exact state of Mike at any point in time does not mean that there is not an exact state of Mike at any point in time. However you want to see mike, as a quantum system, a chaotic system, or a classical system, his choices are a product of events in that system. To claim free will is to say that somehow Mike is able to transcend that system.

That being said, Mike is a decision making machine, just like us humans. Because at some level the decision making process is opaque to all observers within the system, Mike is like a black box. Because of this fact, what is the practical difference between Mike and a machine that actually has free will?

Claiming free will exists is the same as claiming souls exist, because they are both outside the realm of physics. However, you could claim that Mike does have free will, or a soul, and no one would be able to disprove you. The only way to "build" a machine with true free will would be to do so outside of the boundaries of science.

  • 1
    "his choices are a product of events in that system". I think the issue here is: since the events in the system are random, how do we know that the events are a product of his choices? In other words, the randomness of quantum mechanics may be an artifact of free will. Example: we are free to observe which of two paths a particle may take-- quantum mechanics insures the particle takes that path. – user935 Oct 5 '17 at 13:44
  • 1
    @barrycarter Consider the various interpretations of quantum physics. In the Many Worlds interpretation every possibility does obtain, so free will makes no sense. Other interpretations such as De Broglie-Bohm are entirely deterministic. Quantum information theories claim that there is no objective state of the system, so I don't see how to connect free will, but maybe it's possible. In fact, there are hardly any interpretations that allow for free will. – AndyPBJ Oct 10 '17 at 12:28
  • @AndyPBJ Given our own free will and Conway-Kochen's free will theorem, one could claim that a quantum system has libertarian free will. arxiv.org/pdf/0807.3286.pdf A program using a quantum computer could have free will based on what quantum particles have. I haven't used these, but if the results of such quantum computers require that we have only a probability of getting a correct answer then I could see these programs as illustrating libertarian free will. – Frank Hubeny Dec 26 '17 at 15:14

The closest you could is to use a true quantum mechanical random number generator, and base the program's actions off that. If quantum mechanics is correct in saying the universe is inherently random, the program's actions could not be predicted in advance, and would thus constitute a form of "free will".

Of course, that doesn't mean the program (or machine running the program) has free will any more than a rock has free will, but, since you can't predict its actions, it can at least give the illusion of free will, even though it's running a fixed algorithm.

There are several methods to generate psuedo-random numbers and sites like https://www.random.org/ use atmospheric noise to generate high quality random numbers, but, if you accept that atmospheric noise is predictable (ie, it measures something "large" enough that quantum randomness "averages out"), then your program's actions could still theoretically be predictable.

By using a true quantum random number generator, your machine/program's actions would be unpredictable, even in theory.

NOTE: google suggests there are some sites which offer true quantum-level random numbers and that true quantum-level random number generators (hardware) are available. I did not explore either deeply.

  • 1
    But wouldn't it give the illusion of lack of control rather than free will? – user3017 Oct 4 '17 at 21:34
  • Since the OP mentioned programming, I was thinking free will as opposed to predictable actions (which most programs have, or at least should have). You're saying that simulating free will also requires the machine act in a consistent (thought not entirely predictable) way that appears to cater to its best interests? – user935 Oct 5 '17 at 13:39
  • I think I get your point—kind of like a Turing machine, such that an outsider can't tell the difference. – user3017 Oct 5 '17 at 18:27
  • @PédeLeão Yes, but you make a good point too. Free will isn't just about unpredictability. It's also about what an entity "wants". So, you'd have to simulate desire and tilt the probabilities in favor of obtaining that desire, so the actions wouldn't be completely random, but without going to the Pavlov's dogs extreme. – user935 Oct 5 '17 at 18:31
  • @barrycarter You could look at the output of a random number generator as environmental input. The program's response to it, like any other input, has to show free will. It would have to be able to generate a state that is not completely explained by anything coming before this state. One way to simulate the illusion of free will is to have a broken or buggy program, but that only stimulates free will. – Frank Hubeny Dec 26 '17 at 15:03

I think the answer is simple based on the wording;

"construct a program which can be construed as free will"

and the answer is yes. Whether or not free will actually exists is irrelevant considering that many people have construed our own existences as having free will. The chess playing computer has free will in the sense that it chooses what moves to make based on input.

Humans, which we'll say have free will for the sake of comparison, are also limited by knowledge/information but make similar decisions.

It is conceivable that as technology grows, and things like google's deep mind improve, we'll have programs that learn information and make decisions based on that information ultimately writing their own programming the way a human defines themselves and builds character from experience. But is it free will, it can be construed as such a thing, but if you break even our decisions down to reactions in our neural pathways and the deterministic behavior of atoms, there may be no other way to declare free will.

  • 1
    "The chess playing computer has free will in the sense that it chooses what moves to make based on input "... well, no. That's sort of the opposite of free will. The chess computer must make a specific move given any input. Unless you create one that makes random moves, though I'm guessing it wouldn't play very well (would make an interesting experiment though). – user935 Oct 5 '17 at 13:46
  • @barrycarter I agree with what you said about the alternative except that it cannot make "random moves" based on the input from a random number generator. That is still determinism. The program's answer has to be probabilistic and run on a quantum computer, assuming such provides probabilistic output. – Frank Hubeny Dec 26 '17 at 15:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.