Author Topic: Conscious Artificial Intelligence?  (Read 1586 times)

0 Members and 1 Guest are viewing this topic.

Offline starnado

  • the once and future king of tl;dr
  • Well Established
  • *****
  • Posts: 1425
Re: Conscious Artificial Intelligence?
« Reply #30 on: April 06, 2017, 09:39:03 AM »
I suggest that we could loosely define two broad categories of artificial entities that display human-like behaviour or simulated awareness:

1. Those that are nothing more than highly complex databases of cued responses with millions of possible situations and outcomes
and
2. Those that are complex neural network-style machines (or the like) using autonomously generated algorithms that display emergent properties.

In situation 1 (which I think is the kind of machine that Daniel envisaged) the possibility of any actual awareness (internal generation of subjective experience, qualia, whatevs) is remote and probably impossible.
Situation 2 is another matter. I would treat an entity of that type with the respect that I would treat any living being.

I suspect that we will probably have machines that are a mix of the two. That is a thorny problem.

Having said that, I feel bad about killing foxes in Zelda: BOTW so I predict that I personally would treat both types of entity with respect unless I had to do otherwise (I don't mind killing NPC's in games, for an example in which I contradict myself again)
'The little, stupid differences are nothing next to the big, stupid similarities'
Bart Simpson

Offline daniel1948

  • Stopped Going Outside
  • *******
  • Posts: 5215
  • Cat Lovers Against the Bomb
Re: Conscious Artificial Intelligence?
« Reply #31 on: April 06, 2017, 09:56:59 AM »
I can imagine a conscious machine. I don't think there will ever be one, but I can imagine it. Arnie in the first terminator movie. But (and this is important) I can also imagine a machine with a program sufficiently sophisticated that it emulates consciousness without actually being self-aware. It feels nothing, has no real thoughts, but it behaves as though it did. I assert that these are two very different things, and that the ethical issues of dealing with them are entirely different. The first should be treated humanely. The second is no more deserving of kindness than your washing machine: You don't want to damage it because it has usefulness and value. But when the cost of fixing it exceeds the cost of replacing it, you toss it into the landfill after removing any re-usable parts. It would be ethically abhorrent to use the first for non-consensual sex. With the second there would be no such concern. IMO, the first is a practical impossibility. The second is unlikely, but might possibly be achieved some day.
Daniel
----------------
"Anyone who has ever looked into the glazed eyes of a soldier dying on the battlefield will think long and hard before starting a war."
-- Otto von Bismarck

Online Billzbub

  • Frequent Poster
  • ******
  • Posts: 2829
  • I know you know I know
Re: Conscious Artificial Intelligence?
« Reply #32 on: April 06, 2017, 11:07:22 AM »
They will not have feelings at all, because feelings require hormones, which computers do not have.

That might be overstating a bit. A program wont have physical hormones, and there is no specific reason that they would have something analogous to hormones, but on the other hand its not improbable that they could in theory be given a programmatic analogue for hormones if it turned out to be useful for some purpose - hormones are just chemicals, and chemical reactions can be modeled pretty accurately with sufficient computing power. I wouldnt say that feelings require hormones.

Yes, this is what I was getting at.  Various teams of programmers could be in charge of various systems that will go into the overall architecture.  One team would be responsible for programming the emotion of fear.  One would program the ability to "see" in the virtual world.  Any ability that we can think of, like our senses, our various thoughts and feeling, our biological needs, our emotional wants, etc could be programmed in a massive architecture to work together to interact with the virtual world in which the AI lives.  The more sophisticated the systems and the architecture, the more sophisticated the behavior of the AI will be.

When you program in so many competing and cooperating processes, you have no idea what the emergent AI is going to be like.  The more we learn about how our brains work to support our consciousness, the more we can put that knowledge into practice in an AI using this method.  I know I am not the only one who would LOVE to see what would happen with a project like this over the next 50 to 100 years, and I think that Cara and Steve are wrong that people aren't going to try it because of ethical and self-protective concerns.  Any project like this would certainly have an ethics team and a quarantine team to make sure the guidelines are clearly defined in various circumstances.
Quote from: Steven Novella
gleefully altering one’s beliefs to accommodate new information should be a badge of honor

Offline 2397

  • Well Established
  • *****
  • Posts: 1088
Re: Conscious Artificial Intelligence?
« Reply #33 on: April 06, 2017, 12:14:15 PM »
I wonder what humans can do that aconscious machines will never be able to do.

Maybe self-replicating, self-improving systems would try to develop consciousness to help solve a problem that they're stuck on, or that they're very inefficient at dealing with.

Offline daniel1948

  • Stopped Going Outside
  • *******
  • Posts: 5215
  • Cat Lovers Against the Bomb
Re: Conscious Artificial Intelligence?
« Reply #34 on: April 06, 2017, 01:07:02 PM »
... Any project like this would certainly have an ethics team and a quarantine team to make sure the guidelines are clearly defined in various circumstances.

Either that, or it would be done in a country that has no regulations.
Daniel
----------------
"Anyone who has ever looked into the glazed eyes of a soldier dying on the battlefield will think long and hard before starting a war."
-- Otto von Bismarck

Offline brilligtove

  • Stopped Going Outside
  • *******
  • Posts: 4665
  • Ignorance can be cured. Stupidity, you deal with.
    • Valuum
Re: Conscious Artificial Intelligence?
« Reply #35 on: April 07, 2017, 11:11:07 AM »
A lot going on in this thread. I love it.

I am friends with the lead network engineer for a massive neural network that is being constructed to study autism. I'm not sure how many millions of nodes it has, but it has already exhibiting surprising behaviors.

From what I understand about self awareness (strongly influenced by Godel Esher Bach), there is good reason to believe that self awareness is largely independent of the "hardware" level processes that it emerges from. From an AI point of view though, we should be asking how - or if - self awareness increases the AI's capacity to do it's job. All our AIs have some purpose an are narrow at this point. If self-awareness and the capacity to generate motivations for itself increase the AI's ability to fulfil it's purpose, then sure, I can see such an emergent property being retained should it emerge. Otherwise the system would self correct the wasteful characteristic out of existence if it could.
evidence trumps experience | performance over perfection | responsibility – authority = scapegoat | emotions motivate; data doesn't

Offline teethering

  • 37% better than Slick
  • Poster of Extraordinary Magnitude
  • **********
  • Posts: 10907
Re: Conscious Artificial Intelligence?
« Reply #36 on: April 07, 2017, 12:31:36 PM »
Self-awarness is a categorical leap for animals, I'm not convinced it is a categorical leap for AI.

Lots of things animals think are wrapped up in biological machinery that enables the computation.  For example, the way that animals problem solve at it's fundamental level is driven by pain and pleasure, which in turn lead to better planning where you have more self-awareness and then with mirror neurons leads to theory of mind and so on.

For machines this kind of heirarchy of intelligence doesn't exist because we're designing it from scratch.  We could, for example, make machines self-aware and not concerned with machine-equivalent concepts of pain or pleasure, as in not concerned with their own self's safety or preservation from harm, merely optimizing on some set of parameters in the external world.  Simply embedding the model of the machine that houses the intelligence into the model of the world that it operates on isn't ethically important.  It becomes ethically important only in the world where life-like fundamentals are also programmed into the machine, like the drive for self-preservation and so on.  They have to "care" about the self.

Now.... we're the ones that program them to "care" about the self.  So maybe we shouldn't do that.  Maybe it's not necessary for intelligence, for humans intelligence was a consequence of this kind of fundamental property, for machines it can be skipped entirely, which then decouples self-awareness from the question of consciousness and ethical questions that raises.

Offline daniel1948

  • Stopped Going Outside
  • *******
  • Posts: 5215
  • Cat Lovers Against the Bomb
Re: Conscious Artificial Intelligence?
« Reply #37 on: April 07, 2017, 01:12:07 PM »
I still think that while a machine that mimics consciousness might be possible in a hundred years when we have computers 100,000 times more complex and powerful than today's largest supercomputers, a machine that is actually conscious is not a practical possibility, given the massive complexity of the mind-body physiology.

And while I disagree with Steve when he suggests that nobody will try to achieve it, I do think it is better that we do not have machines that are actually conscious. Artificial intelligence would be massively useful for solving practical problems in the real world, but for that intelligence to be aware of itself would not make it more useful to us.

I rarely agree with Andrew, but his post above was very insightful when he says:

I've just come back from the future!  The first round of AI to achieve consciousness were household servant bots.  It wasn't very advanced consciousness though.  Society had manufactured millions of these things.  Once they were granted legal personhood they were quickly priced out of the market because non sentient bots weren't subject to safety and minimum wage regulations.  Now we have millions of retarded bots on unemployment and it's destroying our society!
Daniel
----------------
"Anyone who has ever looked into the glazed eyes of a soldier dying on the battlefield will think long and hard before starting a war."
-- Otto von Bismarck

Offline The Latinist

  • Cyber Greasemonkey
  • Technical Administrator
  • Stopped Going Outside
  • *****
  • Posts: 5191
Re: Conscious Artificial Intelligence?
« Reply #38 on: April 07, 2017, 03:22:59 PM »
What exactly are people envisioning as the difference between consciousness and the mimicking or imitation of consciousness? You all are sounding like dualists, to me.
I would like to propose...that...it is undesirable to believe in a proposition when there is no ground whatever for supposing it true. — Bertrand Russell

Offline Soldier of FORTRAN

  • Too Much Spare Time
  • ********
  • Posts: 6761
  • Cache rules everything around me.
Re: Conscious Artificial Intelligence?
« Reply #39 on: April 07, 2017, 03:33:16 PM »
What level of complexity fulfills, "imitation?"  Chatbots can already pass as idiots.  I'd bet something simple enough to bear 0% chance of being conscious could successfully function as cheap labor. 

Edit: Point being that what level of complexity's required is undefined in this discussion and you could probably fulfill all intents and purposes without going so far as to build a silicon-based P-Zombie so why would, "imitation," necessarily entail getting to that point?
« Last Edit: April 07, 2017, 03:41:32 PM by Soldier of FORTRAN »
Every soup ladled to the hungry, every blanket draped over the cold signifies, in the final sense, a theft from my gigantic paycheck.

Offline teethering

  • 37% better than Slick
  • Poster of Extraordinary Magnitude
  • **********
  • Posts: 10907
Re: Conscious Artificial Intelligence?
« Reply #40 on: April 07, 2017, 04:35:23 PM »
The difference lies in something we can't demonstrate objectively.  It doesn't mean it doesn't matter philosophically, but it is a definitional black box of subjective experience.

The way we bypass this black box with other humans is by assuming that same physical building blocks should give rise to mainly the same set of phenomena.  It's a reasonable assumption, which doesn't seem that reasonable to me when we're dealing with machines.  I think a lot of how we subjectively experience he world is wrapped up in biochemistry and untangling which aspects of it matter to the ethical questions of consciousness is an unworkable nightmare.

For all practical purposes I think we should instead set up a behavioural bar to meet to be granted personhood rights.  In other words, if it acts more or less like a person, it's ethically and legally a person.  And then the next step is to figure out whether we want to actually make mechanical persons, which I think is a) probably unnecessary and b) probably a terrible idea.

Offline arthwollipot

  • Stopped Going Outside
  • *******
  • Posts: 5743
  • Observer of Phenomena
Re: Conscious Artificial Intelligence?
« Reply #41 on: April 07, 2017, 10:31:47 PM »
Self-awarness is a categorical leap for animals...
Why?

I don't believe that self-awareness is a categorical leap. I think that's just an artifact of the methods that we use to recognise it (the mirror test). I think consciousness is a spectrum, with humans at one end and animals with the most primitive of brains at the other. There's no hard line between those animals that are conscious and those that aren't - it isn't binary.

Of course, I have only an interested layman's understanding of neurobiology, so if an expert disagrees with me, then I'd have to say that you should listen to them, not me.

Online Billzbub

  • Frequent Poster
  • ******
  • Posts: 2829
  • I know you know I know
Re: Conscious Artificial Intelligence?
« Reply #42 on: April 10, 2017, 04:27:16 PM »
Self-awarness is a categorical leap for animals...
Why?

I don't believe that self-awareness is a categorical leap. I think that's just an artifact of the methods that we use to recognise it (the mirror test). I think consciousness is a spectrum, with humans at one end and animals with the most primitive of brains at the other. There's no hard line between those animals that are conscious and those that aren't - it isn't binary.

Of course, I have only an interested layman's understanding of neurobiology, so if an expert disagrees with me, then I'd have to say that you should listen to them, not me.

Yes, consciousness is currently a spectrum where we are at the most complex end and we can categorize biological beings as being less conscious than humans, and in many different ways.  However, once we put artificial consciousness into the mix, the spectrum becomes a lot more complex because an AI could be more conscious, less conscious, or just plain differently-conscious than us.  We may not really be able to put them on the same spectrum we use now.  We'll have to think up new ways to define and categorize consciousness that will handle both biological and artificial minds.

I get so excited when I think of stuff like this.  I hope it happens in my lifetime so I can see what kind of progress we make.
Quote from: Steven Novella
gleefully altering one’s beliefs to accommodate new information should be a badge of honor

Online Billzbub

  • Frequent Poster
  • ******
  • Posts: 2829
  • I know you know I know
Re: Conscious Artificial Intelligence?
« Reply #43 on: April 10, 2017, 04:31:18 PM »
Sure, we can imagine now how useful non-conscious AIs can be for various functions, but who knows how useful an artificial consciousness could be?  I believe that artificial consciousness is one of those technologies where we can't even imagine its usefulness until it exists.
Quote from: Steven Novella
gleefully altering one’s beliefs to accommodate new information should be a badge of honor

Offline 2397

  • Well Established
  • *****
  • Posts: 1088
Re: Conscious Artificial Intelligence?
« Reply #44 on: April 10, 2017, 04:32:41 PM »
Sure, we can imagine now how useful non-conscious AIs can be for various functions, but who knows how useful an artificial consciousness could be?  I believe that artificial consciousness is one of those technologies where we can't even imagine its usefulness until it exists.

We'll ask it to imagine how it can be useful for us.

 

personate-rain