Poll: Could Robots Ever Imagine?

[interspark]

one day, in the far far future, robots will be independant from us, they will be able to carry out tasks without the need to be asked, but will they ever be as intellegent as we are? will they ever become so sophisticated, they can dream and imagine on their own? discuss away

 

[dyre]

Yeah, I'm pretty sure we'll hit a tech singularity someday, and a few years later the robots will kick our asses. I think they'll probably dream and imagine in terms of thinking about solving problems in the world, though I'm not sure if they'll ever imagine anything like "damn, that's a hot girl robot walking down the street. *lewd daydreams*"

 

[Limecake]

This is a fairly popular theme in cyberpunk, you can see it explored in the book 'Do Androids Dream of Electric Sheep?' http://en.wikipedia.org/wiki/Do_Androids_Dream_of_Electric_Sheep%3F

which was the book the Blade Runner was loosely based off of. Blade runner is also a good example of exploring the theme of what it means to be human.

The idea that an android could imagine doesn't make a lot of sense to me, Machines are extremely logical things (just look at a computer) part of being able to dream is to willfully throw logic out the window, also I'm assuming an android wouldn't need to sleep since they don't have the biological components that we do.

However I believe a cyborg (part human) would be able to imagine and dream.

 

[Rowan93]

Imagination is kind of necessary for some tasks, so machines that can imagine is pretty much inevitable, and I think there are already some computer programs that run a limited analogy to imagination although I don't really have any sources for that.

I really don't think there's anything special about the human brain that makes it impossible to run a computer program that does it too, because even if you can't program it from the beginning, you can still scan a human brain and build a physics program that simulates all of the brain's inner workings.

Although I doubt machines will dream very often, since they won't ever need to sleep.

 

[Scarim Coral]

Probably but not within our life time (far far into the future).

 

[Jonluw]

Assuming we can manage to create a computer that perfectly simulates the human brain (or something more complicated), yes.
I'm not sure it's possible to do, but I certainly don't believe our ability to dream or imagine is something uniquely human or supernatural that can't be recreated, which I assume is what the OP is asking.

 

[DoPo]

I really don't think there's anything special about the human brain that makes it impossible to run a computer program that does it too, because even if you can't program it from the beginning, you can still scan a human brain and build a physics program that simulates all of the brain's inner workings.

Well, there isn't anything special aside from the fact that we don't know exactly how it operates. That's a major road block in AI - when the field was created in the 60's, there were lots of hopes and and discussions what would happen when machines become self-aware and where do we draw the line of a computer program an an actual individual entity. But as you can see, to this day, we don't have such sophisticated AIs as computer scientists thought we would have.

Right now the what the majority of AI does is not as sophisticated as it appears to be.

If we could decipher exactly how a brain operates, there is nothing stopping us from trying to emulate it. We do it right now. Heck, I suppose the work on the first artificial brain would begin roughly several seconds after somebody understands the human one.

Could AI imagine? Would AI imagine? I don't know. What is imagination in the first place? If we mean "imagination" as in "forming images and sensations without perceiving them"[footnote]Imagine a circle. Now imagine the circle is yellow. That sort of thing.[/footnote] well, then we might argue that current AI can do that...or we may say it will never be possible. If we mean "imagination" as "creativity"[footnote]We say one needs good imagination to become an artist or a writer, so they can think of their works and transfer them to the real world[/footnote] then perhaps the answer is "never" or "eventually".

Ideally, we want the second one. AI performs exceptionally well at expert tasks and ones that take humans lots of time and effort to even be marginally good at: medical diagnostics, technical help, mathematically solved games, etc. Computers are really good at those and sometimes better than humans[footnote]Deep Blue managed to beat Kasparov almost 15 years ago[/footnote], however they are terrible at tasks humans find routine: speech recognition, face recognition, conversation, even walking. Things that we think are so easy that we don't even think about them are the hardest for the computers to emulate and learn/use. That's why we want them to have "imagination" of some sort. Something to make them more like us, so we can finally have true artificial intelligence.

And several nanoseconds after, that the computers will no longer need us.

 

[Rowan93]

I really don't think there's anything special about the human brain that makes it impossible to run a computer program that does it too, because even if you can't program it from the beginning, you can still scan a human brain and build a physics program that simulates all of the brain's inner workings.

Well, there isn't anything special aside from the fact that we don't know exactly how it operates. That's a major road block in AI - when the field was created in the 60's, there were lots of hopes and and discussions what would happen when machines become self-aware and where do we draw the line of a computer program an an actual individual entity. But as you can see, to this day, we don't have such sophisticated AIs as computer scientists thought we would have.

Well, we don't know yet, sure, but it's a physical system so there's nothing fundamentally special about it out there in the physical world.

I need to start being more specific with the stuff I say, this is the third time in not very long that I've said something on this forum, thought that I'd implicitly inb4'd the obvious counterargument, and it's turned up anyway.

To phrase that sentence more specifically, I thought the way I'd phrased my sentence at the start of the post implied that while we don't know yet, I'm assuming that there isn't anything about the physical thing of the brain that makes it unsimulatable, like for instance a soul doing actual important cognitive work, but reading it again there's nowhere near enough information there for that to be obvious.

 

[Esotera]

You can be the most intelligent computer in the world, but you still can't know what's going to happen after more than 7 shots in a game of snooker. (Paraphrasing Iain M. Banks)

Yeah, I think it's perfectly possible, although it'd be a pretty hard form of AI to develop, as it'd need to be a general/strong AI. You'd also need one hell of a computer.

 

[DoPo]

I'm assuming that there isn't anything about the physical thing of the brain that makes it unsimulatable, like for instance a soul doing actual important cognitive work, but reading it again there's nowhere near enough information there for that to be obvious.

No, it's not that. It's just that the brain is a fairly complex thing. Also, the study of the brain and how it operates is new. Psychology has roots in previous studies but it's nothing really like the study of the heart or anatomy and so on. That's one of the problems, we don't have enough background. The whole in-depth study of the mind has been around slightly more than a century. For 10-20 years or something, lobotomy was thought to be a perfect cure for a lot of thing. That's how deluded the field has been - thinking that chopping your brain was the solution to most problems.

So we just need time to crack how the mind operates. But the minds don't like cooperating. Take for example split personalities (known as Multiple Personality Disorder, later renamed to Dissociative Identity Disorder) people think of it as having another separate personality. And you get a lot of these cases. However, having a single extra personality strongly suggests that the "patient" is faking it. DID is characterised as having more personalities than that. Even then, recent studies cast doubt on whether DID exists at all, so maybe people are essentially being misdiagnosed and/or faking it. Not to mention the whole "being normal" is really sketchy.

Unlike physics, for example, psychology isn't that easy to prove. For physics you need lots of maths and proof to back you up, for psychology you can generally just make stuff up[footnote]I'm sire there was a webcomic that said something similar. I think it was either xkcd or SMBC but I couldn't find it in either.[/footnote]. That's why we'd need quite a bit of time. But progress is being made and eventually we'll be able to launch Skynet, don't worry.

 

[Redlin5_v1legacy]

I'm thinking of i Robot now...

Possibly some day. And if I live to see that day I will be still terrified of that idea.

 

[TheYellowCellPhone]

I can only think of this.


I don't know if intelligence is directly related to imagination, so eh.

 

[dobahci]

Sure, I think machines will get to a point where they can imagine, and dream, and think creatively, and perhaps even feel.

But a more interesting question is not so much whether machines will be able to do those things, but rather, when machines can do those things, will they be distinguishable from humans in any significant way (other than their construction)? Will they be afforded the same rights as we've got? Will it be wrong to dismantle a computer that has feelings and a unique sense of identity? Could a machine create art?

I've had some pretty interesting discussions about things of that sort in an AI class I took. Before AI became classified as a sub-discipline of computer science, it was mostly a philosophical issue.

 

[manic_depressive13]

Seeing as the human brain itself is essentially just a computer, I don't think it's too far-fetched to imagine that, with enough research, we might be able to create a machine that thinks and dreams in a similar way to us. The real question is whether we'll manage to eat our way into oblivion before we reach that stage of technological advancement.

 

[Darknacht]

Seeing as the human brain itself is essentially just a computer, I don't think it's too far-fetched to imagine that, with enough research, we might be able to create a machine that thinks and dreams in a similar way to us. The real question is whether we'll manage to eat our way into oblivion before we reach that stage of technological advancement.

The human brain is essentially a very very complex computer with an ass load of bugs and inefficiencies. Quantum computers are very very complex computers with an ass load of bugs and inefficiencies, maybe we can use quantum computers to make human like intelligence.
Before anyone says thats not how it works, I know, Neural networks and Cognitive simulations are way better at building AI systems then just throwing together a Quantum computer. I just think its a funny Idea.

 

[ThatLankyBastard]

Yes... Eventually they will reach a point in which their digital flesh will be superior to our organic bodies in every way, including imagination and creativity...

...and then they will wipe us out or enslave us!

I, for one, welcome our new robot overlords!

 

[McMullen]

Seems like a strange question to ask, let alone make a poll out of. The only reasonable answer is "it's possible". Who are we to say that it can't happen?

 

[Arakasi]

We are biological robots.

So, yes.

 

[Veylon]

Seeing as how hard it is to pin down exactly what dreaming, imagining, and intelligence really are and what they do on a physical or electrical level, I think it's absurd to try and conjecture whether or not an AI will someday do them too. But I think the answer is yes.