5 ms·

...say we build/grow an AI system that passes the Turing test. We talk to it, it comes across as plausibly human. How do we know we've created something with c

...say we build/grow an AI system that passes the Turing test. We talk to it, it comes across as plausibly human. How do we know we've created something with consciousness, and not just a very sophisticated program? The philosophy you disparage already has a term for this: the "philosophical zombie". For all intents and purposes, they appear human, but they have no internal experience whatsoever. All you'll have done is shunt the problem downstream. Also, you're wrong about early vision. That's the best studied part of the brain, and in fact, researchers have applied ML techniques to fMRI data and extracted out the images being shown.

mannykannot 9y ago
The jury is still out on whether the concept of p-zombies contributes to the understanding of consciousness, or is a misleading dead-end.
UnquietTinkerer 9y ago
If we can't quantify "internal experience" then what leads philosophers to believe it exists? What evidence is there that conciousness is distinguished by something other than raw complexity?
- roenxi 9y ago
  Adding a few sentences to this point; the mechanism for this "AI system that passes the Turing test" likely to involve neural networks. So we'd be dealing with a situation where we have something that is behaving in a human like manner using mechanisms that are (to a first approximation) how humans do it. Claiming that that program isn't conscious is close to defending the idea that the sun orbits the earth - the argument is 'even though we have accurate measurement everything and a well verified model, the implications are discomforting; hence the model must be wrong'. There are people who won't handle the idea that consciousness is rooted in physical processes that we already have a handle on.
  - marcoperaza 9y ago
    Consciousness as it’s being discussed is not intelligence or even the ability to reason about one’s own existence. Is both less and more than those things, really just different altogether. It is subjective experience itself, the inner world somehow projected for you by your mind. It is the experience of seeing and hearing, the feeling of an emotion.
    - stcredzero 9y ago
      If we made a machine with the capacity for simulating intelligence, the ability to reason about one's own existence, and furthermore, gave it the ability to simulate a belief in its experience of seeing and hearing and feeling emotion and a survival instinct, would it be ethical to destroy such a machine? At that point, wouldn't it protest its own sentience, consciousness, and desire to live as convincingly as a human?
      - marcoperaza 9y ago
        That’s really a separate question. We really can’t be sure that other people are conscious at all either. Consciousness may have behavioral consequences, but it’s not clear that it is necessary for those behaviors. Perhaps those behaviors can come about some other way. Our ability to talk about consciousness doesn’t prove much either, because it turns out we can’t explain the idea in words. Trying to explain consciousness comes down to statements like: it’s the difference between seeing and !!!seeing!!! We can only communicate its true nature by way of alluding to the other’s experience of the same, not by direct explanation. I don’t think it’s helpful to being morality into it. Depending on your moral views, there might be good reason to treat things that we think are conscious as if they are conscious. That’s arguably what we do with other people. It doesn’t speak to the underlying questions though.
        c22 9y ago
        If, as you surmise, we're discussing a concept which cannot be articulated and which has no moral consequence then what underlying questions remain and what use should we find in answering them? It strikes me as similar to logical paradoxes--a fuzzy ambiguity more likely reflective of limitations in our languages and our particular models of reasoning than of any underlying physical truth.
- marcoperaza 9y ago
  What leads them to think it exists? Having the internal experience themselves of course.
- discoursism 9y ago
  It will not be provable, the same way you cannot prove your consciousness to me. All the philosophizing in the world will never change that fact. The most we'll know is that the artificial intelligence appears to be conscious, and shows the same or similar physical appearance of consciousness that we ourselves do.
- CuriouslyC 9y ago
  Given the fact that the only reason you are aware of the world around you is through internal experience (i.e. consciousness), the reverse question is actually more salient: how can you prove that the "real world" really exists? How do you know that it isn't entirely imagined?
joe_the_user 9y ago
How do we know we've created something with consciousness, and not just a very sophisticated program? -- How do we know other humans aren't "very sophisticated programs"? How do we care? Which is to say the whole "omg they look and act the same but are so, sooo different" of philosophical zombies talk and similar, uh, rot, is kind of leveraging the person's intuitive attachment the idea of a soul, a magic thing with no measurable properties that makes us oh-so-different than a crude material object. Also, you're wrong about early vision. That's the best studied part of the brain, and in fact, researchers have applied ML techniques to fMRI data and extracted out the images being shown. If you read the GP in context, it doesn't say neurologists haven't studied consciousness, it says people have no self-awareness of the preprocessing that goes into creating vision.
- aninhumer 9y ago
  >How do we know other humans aren't "very sophisticated programs"? We don't. Most people assume that other people have consciousness, with various justifications, but no one can know that. >a magic thing with no measurable properties that makes us oh-so-different than a crude material object. But it does make a difference morally speaking. If we consider a Utilitarian model of morality, then what is good or bad is that which causes the experiences of happiness or suffering. But if a philosophical zombie does not experience pleasure or suffering, then there is no moral consequence to harming them, any more than there is moral consequence to watching someone die in a movie, or to splitting a rock in half. And this becomes even more worrying when we start to think about AI. If an AI can develop consciousness as a result of our actions, then perhaps it can also experience suffering as a result of our actions. Of course, since we cannot identify p-zombies, possibly by definition, one can argue that there's not much point thinking about it. But, well, you could say that about a lot of philosophy. Personally, I think it's valuable to think about these things as an intellectual exercise, even if we can't really conclude anything from it.
- KingMob 9y ago
  > a magic thing with no measurable properties that makes us oh-so-different than a crude material object No, qualia/awareness is not magical, it's prima facie undeniable, we are aware of things, and we have an experience of the world. In fact, consciousness is more fundamental than everything else we know about the world, because all else we know is known by awareness. That scientific paper you read? You were aware of it. Bacteria under the microscope? You had an experience of seeing. It's the very definition of the hard problem. How does biology lead to awareness? We have strong evidence they're related (sleep, scotomas, mental disease, drugs, etc etc etc), but we have yet to pinpoint the connection. As for the philosophical zombie being "rot", consciousness neuroscientists have already debated if it were actually possible, and if so, detectable. We already know about a phenomenon called blindsight for vision, where you can perform visual tasks but have no experience of seeing; how do we know there's not some frontal lesion that could wipe out or reduces your awareness, but leave your other faculties mostly intact?

mannykannot 9y ago

The jury is still out on whether the concept of p-zombies contributes to the understanding of consciousness, or is a misleading dead-end.

UnquietTinkerer 9y ago

If we can't quantify "internal experience" then what leads philosophers to believe it exists? What evidence is there that conciousness is distinguished by something other than raw complexity?

roenxi 9y ago
Adding a few sentences to this point; the mechanism for this "AI system that passes the Turing test" likely to involve neural networks. So we'd be dealing with a situation where we have something that is behaving in a human like manner using mechanisms that are (to a first approximation) how humans do it. Claiming that that program isn't conscious is close to defending the idea that the sun orbits the earth - the argument is 'even though we have accurate measurement everything and a well verified model, the implications are discomforting; hence the model must be wrong'. There are people who won't handle the idea that consciousness is rooted in physical processes that we already have a handle on.
- marcoperaza 9y ago
  Consciousness as it’s being discussed is not intelligence or even the ability to reason about one’s own existence. Is both less and more than those things, really just different altogether. It is subjective experience itself, the inner world somehow projected for you by your mind. It is the experience of seeing and hearing, the feeling of an emotion.
  - stcredzero 9y ago
    If we made a machine with the capacity for simulating intelligence, the ability to reason about one's own existence, and furthermore, gave it the ability to simulate a belief in its experience of seeing and hearing and feeling emotion and a survival instinct, would it be ethical to destroy such a machine? At that point, wouldn't it protest its own sentience, consciousness, and desire to live as convincingly as a human?
    - marcoperaza 9y ago
      That’s really a separate question. We really can’t be sure that other people are conscious at all either. Consciousness may have behavioral consequences, but it’s not clear that it is necessary for those behaviors. Perhaps those behaviors can come about some other way. Our ability to talk about consciousness doesn’t prove much either, because it turns out we can’t explain the idea in words. Trying to explain consciousness comes down to statements like: it’s the difference between seeing and !!!seeing!!! We can only communicate its true nature by way of alluding to the other’s experience of the same, not by direct explanation. I don’t think it’s helpful to being morality into it. Depending on your moral views, there might be good reason to treat things that we think are conscious as if they are conscious. That’s arguably what we do with other people. It doesn’t speak to the underlying questions though.
      - c22 9y ago
        If, as you surmise, we're discussing a concept which cannot be articulated and which has no moral consequence then what underlying questions remain and what use should we find in answering them? It strikes me as similar to logical paradoxes--a fuzzy ambiguity more likely reflective of limitations in our languages and our particular models of reasoning than of any underlying physical truth.
marcoperaza 9y ago
What leads them to think it exists? Having the internal experience themselves of course.
discoursism 9y ago
It will not be provable, the same way you cannot prove your consciousness to me. All the philosophizing in the world will never change that fact. The most we'll know is that the artificial intelligence appears to be conscious, and shows the same or similar physical appearance of consciousness that we ourselves do.
CuriouslyC 9y ago
Given the fact that the only reason you are aware of the world around you is through internal experience (i.e. consciousness), the reverse question is actually more salient: how can you prove that the "real world" really exists? How do you know that it isn't entirely imagined?

joe_the_user 9y ago

How do we know we've created something with consciousness, and not just a very sophisticated program? -- How do we know other humans aren't "very sophisticated programs"? How do we care? Which is to say the whole "omg they look and act the same but are so, sooo different" of philosophical zombies talk and similar, uh, rot, is kind of leveraging the person's intuitive attachment the idea of a soul, a magic thing with no measurable properties that makes us oh-so-different than a crude material object. Also, you're wrong about early vision. That's the best studied part of the brain, and in fact, researchers have applied ML techniques to fMRI data and extracted out the images being shown. If you read the GP in context, it doesn't say neurologists haven't studied consciousness, it says people have no self-awareness of the preprocessing that goes into creating vision.

aninhumer 9y ago
>How do we know other humans aren't "very sophisticated programs"? We don't. Most people assume that other people have consciousness, with various justifications, but no one can know that. >a magic thing with no measurable properties that makes us oh-so-different than a crude material object. But it does make a difference morally speaking. If we consider a Utilitarian model of morality, then what is good or bad is that which causes the experiences of happiness or suffering. But if a philosophical zombie does not experience pleasure or suffering, then there is no moral consequence to harming them, any more than there is moral consequence to watching someone die in a movie, or to splitting a rock in half. And this becomes even more worrying when we start to think about AI. If an AI can develop consciousness as a result of our actions, then perhaps it can also experience suffering as a result of our actions. Of course, since we cannot identify p-zombies, possibly by definition, one can argue that there's not much point thinking about it. But, well, you could say that about a lot of philosophy. Personally, I think it's valuable to think about these things as an intellectual exercise, even if we can't really conclude anything from it.
KingMob 9y ago
> a magic thing with no measurable properties that makes us oh-so-different than a crude material object No, qualia/awareness is not magical, it's prima facie undeniable, we are aware of things, and we have an experience of the world. In fact, consciousness is more fundamental than everything else we know about the world, because all else we know is known by awareness. That scientific paper you read? You were aware of it. Bacteria under the microscope? You had an experience of seeing. It's the very definition of the hard problem. How does biology lead to awareness? We have strong evidence they're related (sleep, scotomas, mental disease, drugs, etc etc etc), but we have yet to pinpoint the connection. As for the philosophical zombie being "rot", consciousness neuroscientists have already debated if it were actually possible, and if so, detectable. We already know about a phenomenon called blindsight for vision, where you can perform visual tasks but have no experience of seeing; how do we know there's not some frontal lesion that could wipe out or reduces your awareness, but leave your other faculties mostly intact?