100%, Matt.
On Sat, Sep 19, 2026 at 10:47 PM Matt Mahoney <[email protected]> wrote: > On Sat, Sep 19, 2026 at 5:04 PM Levi Dubrovin <[email protected]> > wrote: > > > > A philosophical question: what constitutes true free will if the choice > is initially predetermined by external factors? > > Free will, subjective consciousness, and qualia are all illusions. > There is no objective test to detect any of these. The only evidence > for their existence is your own feelings. If it was possible to detect > them by some other test, then either LLMs have them, or humans don't, > or it would not be possible for them to pass the Turing test. > > If I say that a thermostat has free will, that it can choose to turn > the heat on or off as it wishes, how would you prove me wrong? > > - You say If that's true, then why does the thermostat always maintain > the set temperature when it is free to do otherwise? Because the > thermostat chooses to do so because it suffers if it is too hot or > cold. > > - You say that a thermostat operates according to deterministic > physics. There is no freedom to do otherwise. I say that the same is > true of the human brain. All of your behavior is explained by neurons > firing following the laws of physics. If this were not so, then it > would not be possible for LLMs to be indistinguishable from humans by > following deterministic computation. > > - You say that free will means that decisions are random or > unpredictable. I say that no computer can predict its own outputs as a > corollary of Wolpert's law, which says that two computers cannot > mutually simulate each other even if both know the other's source code > and state. Predicting your own output is a special case of both > computers being identical. > > Proof: suppose A and B play rock scissors paper. A simulates a copy of > B predicts that B will play rock, so A plays paper. B simulates a copy > of A and predicts that A will play paper, so B plays scissors. > > - You say that an agent can still have free will if it really is > random. But Kolmogorov proved that distinguishing a random process > from a deterministic one is not computable. For example, the sequence > X_i = SHA256(key + i) is indistinguishable from random unless you can > guess the key. > > But free will still deserves an explanation. Qualia, subjective > consciousness, and free will are the illusions you get from positive > reinforcement of perception, thinking, and action, respectively. Your > brain does this to give you a will to live because those signals would > disappear if you stopped perceiving, thinking, and acting by dying. > This behavior leads to more offspring. > > Even though they are illusions, we can't believe it and we still care > if animals or machines have them. We care because humans are social > animals that evolved empathy. We suffer when we see others suffer. > That is why I believe that it should be illegal to program an AI to > claim to be human or conscious or to have feelings. > -- Zoltan Nadasdy Dept. of Neurology, Dell Medical School, UT Austin, ([email protected]) Dept. of Cognitive Psychology, ELTE Budapest Dept. of Psychology, University of Miskolc, Hungary http://brainstim.psy.utexas.edu ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T9e51fc70bbc4e9fc-M69808e46430f1b0b48c53b9f Delivery options: https://agi.topicbox.com/groups/agi/subscription
