Will AI ever = conciousness or sentience?

[Angrboda]

This question has two, more or less equally important halves. The question as posed is largely incoherent, postulating capabilities of human minds that are so poorly defined as to compete with contemporary theology for emptiness, and much of the capability that is vaguely alluded to hasn't, in the manner described, even been shown to be possessed by human brains or minds.

So the first half of answering the OP is clearing away all the bullshit, folk psychology, unsubstantiated metaphysical kruft and new agey whackadoo that infects the question as posed. Do you have an actual entity with known properties that you are asking to be duplicated, or are you asking whether God can create a rock that he cannot lift? I think your OP leans strongly toward the latter.

Having just this past month put the last major brush strokes on my own theory of mind and brain, I have my own conclusions about such questions, unfortunately much of that thinking is still on the shelf marked "proprietary." In philosophy of mind discussions, the types of properties of mind which the OP and others allude to, I often refer to simply as "special sauce," ala the McDonald's Big Mac, whose "special sauce" was considered an essential contributor to the sandwich's unique appeal. Consciousness and similar effects of minds or brains is often put in the role of the special sauce, that we have a thinking machine, but we also have this something extra which makes the operation of the human mind categorically different from other machines (and incidentally, different from all or at least some other animals, which poses innumerable difficulties given the nature of biological evolution). I've yet to see a good, workable definition of either consciousness or this special sauce in general, but if the OP will provide it, I may oblige with a more substantive response.

For my part, my belief and my theory is that there is no special sauce. Once you remove the questionable assertion that this special sauce exists with the magical properties frequently attributed to it (such as the degree of self-referentiality referred to above), the business of explaining the human mind becomes much more tractable (and in my view, more realistic). So, in a nutshell, yes, we might be able to recreate a mind like that which humans possess. The only obstacles being the practical and political ones which face any technological project. The brain, excluding extra-brain contributions to mind, consists of 100 billion neurons (1) (in addition to other cells and biological materials; the role of glial cells is coming to be appreciated to be much greater than previously surmised, and a human without a body and functioning endocrine system would hardly be human). And we still understand its operation largely piecemeal and by inference. We do not yet have a Darwin of brain science who has proposed a plausible unifying theory. (A saying in neuroscience is that, "neurons which fire together, wire together; neurons that fire apart, wire apart"; this rather stochastic feature of brain development likely underlies the topological layout of things like tactile sensation in the cortical tissues, which mirrors that of the physical tissue in its layout; a hand neuron will fire more often, more closely in time to a forearm neuron than it will a toe neuron, and thus ends up being wired more closely to it; this principle likely has dramatic implications for the nature and function of our brain systems, but as yet, it's difficult to extend it beyond a few isolated modalities.)

So what we lack is twofold. A proper understanding of the question. And the actual answer to the question.

There is a hidden question here that bears voicing. Even if we could create a mind which operates in the same manner as the human mind does, would we want to be limited to that pattern? The preference matching algorithm at Amazon.com is an example of modern machine intelligence. It performs its function supremely well, yet beyond basic principles of operation, we have no clue as to the specifics of how it does what it does. They created the basic form of its intelligence, and turned it loose to grow and learn and perform. What is lacking in the machine mind of the Amazon.com preference machine that you would want to add to it, from the capability of the human mind, speaking specifically of those things that you suspect the human mind uniquely capable, or that you believe is not duplicable?

In the interest of clarity, I will simply state that I do not believe in either free will or consciousness as popularly described. I know whateverist and I differ on the former, so there may be some aspects of the question here which may be fundamentally irreconcilable. And I suggest we don't go down that route: there are free will discussions aplenty. It might be useful though, to simply shelve that question, and address the question on account of what remains.

(1)

[The Grand Nudger]

Wondering when you'd show up Apo......welcome to the maze.

[Whateverist]

I want extra special sauce on my consciousness. =)

I'm monitoring quiz takers today so I have a little furtive attention for this now. Looking forward to absorbing the full impact of Apo's blow to my position later. (Agreed about the free will debate by the way, lets not go there. For me it is too much silly.)

[Rayaan]

Why the deference to "the same way that we do"? Is our route to consciousness or self awareness the only route? Even so, suppose our theoretical AI machine was built using organic materials...now how different are we?

The difference, to me, is that machines can be self-aware only in a mechanical type of fashion because their behaviors and their decisions are purely determined by a set of rules (which they have been programmed with). Consciousness, however, is something that I think of as a combination of many different things which includes things like attention, feelings, intentions, imaginations, and even free will to a certain extent. I don't think it's possible for AI machines to have those attributes.

Also, as far as I know, AI machines are intelligent only in the sense of being able to solve problems (by searching), but that is not "thinking."

How exactly do conscious beings 'work'? We understand lots about how a human body works and we've mapped the brain to find those places where a tweek will create a twitch or a severence can create a particular sort of dysfunction. Even so, I am not impressed that we are very close at all to understanding how conscious beings work.

Good question. But, yeah, I agree with you. We do not understand how exactly conscious beings work. There are many different theories on how consciousness works and there are even entire books on this topic. Consciousness is not something that I have a good grasp of, but it is still interesting to me, and I have some ideas what it may be.

How does that mean they will not be self aware or have consciousness? Just because it might be different does not mean that it won't be comparable.

See my response to Rhythm.

Currently I'd agree, but if a computer program was to become every bit as complex as DNA and the human mind what's stopping it from becoming as 'self referential' as us?

I think that the main issue, again, is not being able to identify what exactly gives rise to self-awareness/consciousness. We also don't know if it depends on the complexity of our DNA specifically. As for the complexity of the human mind, I think it is still debatable whether or not computer programs can actually become anything close to the human mind.

A fly isn't very self-referential is it? But that's because it's not at a similar complexity or development as a human is. Neither is a toaster. My view is that if a machine were ever created with relatively the same complexity as us, then there's no reason for it not to experience consciousness in a comparable way to what we do, if it were designed to do so.

Well, I don't think that a similar complexity as ours is the only requirement to produce consciousness in a machine. A particular AI machine may be built in the future which is more complex than the human brain, but that doesn't necessarily mean that it can possess consciousness as well.

As I said before, in my opinion, consciousness also embodies our faculty of attention, reasoning, feelings, intentions, imaginations, and even free will, among other things. It's not necessarily something that just sits in our brains all by itself. It's possible that the existence of our consciousness is dependent on all or at least some of those things - simultaneously.

[The Grand Nudger]

The difference, to me, is that machines can be self-aware only in a mechanical type of fashion because their behaviors and their decisions are purely determined by a set of rules (which they have been programmed with).

We'd have to show that ours aren't...not just that we're ignorant of those rules (if they exist) for this to make a sound difference between the two. Are you ready to do so?

Consciousness, however, is something that I think of as a combination of many different things which includes things like attention, feelings, intentions, imaginations, and even free will to a certain extent. I don't think it's possible for AI machines to have those attributes.

It's complex, okay (ignoring that it may not be). But complex isn't a barrier for a machine. I get that you don't think it;s possible..but why don't you think it's possible?

Also, as far as I know, AI machines are intelligent only in the sense of being able to solve problems (by searching), but that is not "thinking."

As far as you're aware.....you're aware of existent AI machines? Further, you don;t think you do any "searching" in the the way you think? Less complicated, currently, sure...but I haven't anything here that substantiates a fundamental difference in what a machine could accomplish lain aside what we can accomplish (or a demonstration that we accomplish any of these things to begin with).

The requirement to do any of these things we might invoke -simultaneously- isn't any barrier to what a machine might be capable of doing. So we build a bunch of modules capable of producing whatever it is we care to invoke..then hook them up to an aggregate. Just how good do you think we are at doing a bunch of things simultaneously anyway?

[jonb]

So lets play with the idea.

Billionaire Mad scientist Dr Moreau, manufactures a sentient being and puts it to work making other sentient beings. The Robot turns round and says I am sentient, therefore I have a right to self determination, and appeals to us for help. Dr Moreau says it is my property I made it, it belongs to me and it lives on my island it and all that come after it are my slaves and have to do as I say and I have the right to do with all of them as I wish. Where do we stand?

[The Grand Nudger]

Unless we want to devalue our own sentience and appeals for help in a similar scenario...we'd probably want to lend a hand.

[jonb]

Unless we want to devalue our own sentience and appeals for help in a similar scenario...we'd probably want to lend a hand.

Yes that's what I thought, so much for Dr Moreau or should I say god's position.

[Whateverist]

@ bolded- I think I've been remiss in explaining the challenge. Like you, I wouldn't imagine AI to "think like we do", to experience things "as we do"....but I don't think that this would disqualify AI from being self aware, or intelligent, from being genuine. In the same way that other creatures may experience "consciousness" differently than we do, a machine may be capable of experiencing something (perhaps even by a similar means logic gates to nuerons) indistinguishable from consciousness, at which point it's difficult to see why we would withhold the term.

I'm okay with thinking happening in a different way than we do it.

I'm also okay with describing a program's ability to take into account its own effects in the execution of a task as a kind of "self awareness", though there is very little special sauce in such a meaning.

Try to distance yourself from the notion of a trick to begin with.

Check. Simulations are off the table.

@ Both. Why assume that consciousness is somehow measured by our own as a "thing", and similarly why ignore those similarities our consciousness (and the structure we feel is at least somehow involved) has with things that are not ourselves? Why not consider "human consciousness" a -type- of consciousness that leverages principles which can be, though aren't always, leveraged by other things? Why assume that ours is the real deal and not a "trick" to begin with - wouldn't it be better to establish this than simply declare it..especially if we're trying to compare two proposed models of how something might be "conscious"? It bears mention in this, that I'm not trying to devalue your experience of consciousness by likening it to a "trick"...if it were a "trick" it would be a very valuable one, on that I think we both agree- just trying to pry this idea of what is or isn't trickery away from what is or isn't human, or like us. Trying to make this something other than bare bias towards what you and I possess and call "consciousness"

(I could argue against my own choice of words in this post all day long by the way...just hoping to have conveyed the general theme)

I just wouldn't know how to begin to define consciousness apart from what I know of it from the inside. I have no idea how it works, though a beating heart and the availability of oxygen seem crucial .. albeit only to me and my kind. What are the hallmarks of consciousness? John Searle says things like it (consciousness) being to the brain what digestion is to the stomach. (Was that Apo I just heard wailing?) Like him (only much more so) I don't have a clue what consciousness is in and of itself.

I have no doubt that my dogs are conscious and I don't think they ever formulate a proposition or contemplate ones validity. The way we process information would seem to be closer to "reasoning" and "thinking". But the fact that my dogs move purposefully through the world means there are other ways to do it. So for sure, robots and programs can be made to move successfully through the world attending to complex tasks. I really want to say they would be assigned their tasks by us of course, but then you would want to know what assigns ours' .. and I have no wish to invoke another free will discussion.

Really I have no problem with robots having as robust an inner life as possible. I'm not a speciesist so why should I want to favor carbon based life over any other? I just can't seem to rap my head around how that would work, even though I also don't know how it works in us.

---

So lets play with the idea.

Billionaire Mad scientist Dr Moreau, manufactures a sentient being and puts it to work making other sentient beings. The Robot turns round and says I am sentient, therefore I have a right to self determination, and appeals to us for help. Dr Moreau says it is my property I made it, it belongs to me and it lives on my island it and all that come after it are my slaves and have to do as I say and I have the right to do with all of them as I wish. Where do we stand?

Hypothetically speaking .. would there be pleasure-bots with special skills?

---

This question has two, more or less equally important halves. The question as posed is largely incoherent, postulating capabilities of human minds that are so poorly defined as to compete with contemporary theology for emptiness

Pretty much what I was going for so far.

.., and much
So the first half of answering the OP is clearing away all the bullshit, folk psychology, unsubstantiated metaphysical kruft and new agey whackadoo that infects the question as posed.

Now hold on there. You're taking away some of my favorite sources!

Do you have an actual entity with known properties that you are asking to be duplicated, or are you asking whether God can create a rock that he cannot lift?

Yes and no. We are entities but that doesn't mean we know everything there is to know about how we work. Not sure how you get God from any of this though. Whatever we have going that we call consciousness is entirely down to earth as far as I'm concerned. No deity required.

In philosophy of mind discussions, the types of properties of mind which the OP and others allude to, I often refer to simply as "special sauce," ala the McDonald's Big Mac, whose "special sauce" was considered an essential contributor to the sandwich's unique appeal. Consciousness and similar effects of minds or brains is often put in the role of the special sauce, that we have a thinking machine, but we also have this something extra which makes the operation of the human mind categorically different from other machines (and incidentally, different from all or at least some other animals, which poses innumerable difficulties given the nature of biological evolution). I've yet to see a good, workable definition of either consciousness or this special sauce in general, but if the OP will provide it, I may oblige with a more substantive response.

No can do. I have no idea what it is either. I certainly can't generalize beyond what I know about it from direct experience, but then it is hard to isolate exactly what the it is, isn't it?

I certainly don't think our brand of consciousness is categorically different than that of any other animal, and the closer you get on the family tree the greater the similarity I would imagine. But I do think that whatever it is has something to do with the way it feels to exist as the kind of organism you happen to be. It has something to do with what matters to you and what that feels like. Being a language using animal with some powers of abstraction means I can shoot my mouth off about this all day even though I admit I don't know what the hell I'm talking about. (I'm pretty sure I'm at the right party as far as that's concerned.)

For my part, my belief and my theory is that there is no special sauce. Once you remove the questionable assertion that this special sauce exists with the magical properties frequently attributed to it (such as the degree of self-referentiality referred to above), the business of explaining the human mind becomes much more tractable (and in my view, more realistic). So, in a nutshell, yes, we might be able to recreate a mind like that which humans possess. The only obstacles being the practical and political ones which face any technological project. The brain, excluding extra-brain contributions to mind, consists of 100 billion neurons (1) (in addition to other cells and biological materials; the role of glial cells is coming to be appreciated to be much greater than previously surmised, and a human without a body and functioning endocrine system would hardly be human). And we still understand its operation largely piecemeal and by inference. We do not yet have a Darwin of brain science who has proposed a plausible unifying theory. (A saying in neuroscience is that, "neurons which fire together, wire together; neurons that fire apart, wire apart"; this rather stochastic feature of brain development likely underlies the topological layout of things like tactile sensation in the cortical tissues, which mirrors that of the physical tissue in its layout; a hand neuron will fire more often, more closely in time to a forearm neuron than it will a toe neuron, and thus ends up being wired more closely to it; this principle likely has dramatic implications for the nature and function of our brain systems, but as yet, it's difficult to extend it beyond a few isolated modalities.)

I don't have any reason to think we could not in principle understand the workings of biological systems to the degree necessary to create our own biological creation. Perhaps we could surpass ourselves? It wouldn't shock or offend me but then again there are other projects I'd prefer to attend to.

So what we lack is twofold. A proper understanding of the question. And the actual answer to the question.

That about sums it up alright.

The preference matching algorithm at Amazon.com is an example of modern machine intelligence. It performs its function supremely well, yet beyond basic principles of operation, we have no clue as to the specifics of how it does what it does. They created the basic form of its intelligence, and turned it loose to grow and learn and perform. What is lacking in the machine mind of the Amazon.com preference machine that you would want to add to it, from the capability of the human mind, speaking specifically of those things that you suspect the human mind uniquely capable, or that you believe is not duplicable?

Not a thing. That is exactly what I think of when I think of AI. It is a perfect example of disembodied, self-sufficient intelligence. My OP was precipitated by discussions elsewhere with folks who wanted to imagine AI as being sentient and conscious in the exact same way we think of ourselves as being (whatever the hell that actually may be). AI is interesting as a feat of human intelligence. It doesn't need to wonder what it all means or worry about its civil rights. It is like all of our intelligence with none of our neuroses.

[Tiberius]

It all depends on what consciousness actually is. If it is just the result of chemical reactions in the brain, there are no reasons why this same result could not be achieved artificially by a complex computer.

If consciousness is more than that, it may well be impossible.

I'm not a believer in "consciousness" or "free will". Both are most likely illusions. We have consciousness only in the form that we are aware of our actions after being forced into performing them.