[Shrike]

Why would anyone want a sentient dishwasher anyhow?

[Grey Wolf]

You know, if we develop sentient AIs, we are going to have a nasty fight going on. Heck, even in Star Trek, they had problems with arguments over whether Data was sentient or property of Starfleet.

[Stealth]

and the chances of man inventing AI capable of making its own decisions/emotions are how few?

[Shrike]

Originally posted by Stealth
and the chances of man inventing AI capable of making its own decisions/emotions are how few?

Probably pretty good.  You could start with cultured neural cells or a processor system patterned off a brain.

[Stryke 9]

It'd make playing TMP more interesting, I'll say that for it. Though the damn thing beats me every time as it is...

[Liberator]

Sentience
1.  Has a concept of Self.  I.e. It recognizes an image in a mirror as a reflection of itself and not another being.

2. Has a rudimentary ability to Reason.  I.e. It can follow a path of logic and come up with an answer

3. Has a desire to maintain it's existence.


As to whether or not construction of a Sentient computer is ethical and moral.  A Sentient computer would still be a tool.  Again we go back to the Grandfather of this argument, Isaac Asimov.  Many of his books dealt with this very point.  The reason that such a thing would be constructed is to assist humans by placing an entity capable of real time Judgement calls.  An ordiary computer would look at a problem and present solutions.  A Sentient computer would examine the same problem and be able to carry out the solution itself without human interference.  An example is long-duration, solo or minimal crew space missions.  Granted this is a theoretical argument but as this whole thing is theoretical...  In a solo or minimal crew mission each Human crew member becomes a very valuable asset.  A sentient computer could be placed on board to fill the crew's need for companionship and keep the crew from having to go into dangerous situations.  Humans would still have to go to supplement the computer and keep it running.  Before you start saying that this places Humans into support roles, it gives the crew something to do on the voyage and thus allows them to earn their keep.

[CP5670]

Self awareness. I'm not talking about "emotions" or anything like that - those concepts would probably be pretty much abstract for it, even though it would understand them. I'm talking about an artificial intelligence that knows it exists and knows what it is - even though it is dedicated to performing any task the programmer designed it for.

That could be done even today, though. You could just program a computer to know that it exists. To really know something is to be able to expand upon it (i.e. deduce other stuff from it), so as long as that the fact that it is a computer is in its knowledge base that it uses to determine stuff, it would be quite sentient.

I think that a truly "sentient" computer can be described as one that can pass the famous Turing test for all five senses, at which point the human and the computer would have merged into one.

A true sentient AI would need hundreds, if not thousands, of paralell processors, like our own brains. The problem is, this is a much more complex system, and thus, slower. Not to mention it would be an engineering nightmare.

Actually, it would be far simpler, mathematically speaking, than the human brain is; one of the reasons the human brain is slower than the computer in certain ways is because of its more complex nature, but then it can also handle a wider range of tasks. And just as thousands of transistors are on an integrated circuit, perhaps one could also cram thousands of processors onto a little 1x1 mm square.

and the chances of man inventing AI capable of making its own decisions/emotions are how few?

Almost certain, given the full lifespan of human civilization.

[Kamikaze]

Well, who's definition of "sentient" are we going by here? If we decide the dictionary is the ultimate jurisdiction, I'd say sure...

1. Has a concept of Self. I.e. It recognizes an image in a mirror as a reflection of itself and not another being.

a program could be programmed to "see" itself in the mirror and deduce that it is itself. Just do a sort of match-image-up sort of thing or have it deduce that a reflection is an image of "itself" and the surrounding objects.

2. Has a rudimentary ability to Reason. I.e. It can follow a path of logic and come up with an answer

Computers can do this. What seems interesting is that humans tend to reason more poorly than humans, often taking options that would decrease their productivity/reproduction or others'. (assuming that productivity and reproduction are the most important factors for animals...)

3. Has a desire to maintain it's existence.

That could be programmed to.... (depending on what you call "desire", I'm assuming a programmed survival trati counts)

[Star Dragon]

Freespace: Beyond Redemption (or, when good AI's  go BAD! )

Hmm maybe I should change my storyline...

"Alpha-1 send a probe into the nebula"...

   "Um Control, we have a problem."

"What sort of problem Alpha-1?"

   "Kind of a mechanical one..."

"Can you be more specific?"

   "The probe says it doesn't wanna go in there cause it's afraid of the dark!"

(Oh yeah we need smarter AI's in the game, but NOT like that!)

[Nico]

if it's a nice AI yeah, if it's a mean one, no! no!
voila

[Fineus]

Originally posted by Grey Wolf 2009
Heck, even in Star Trek, they had problems with arguments over whether Data was sentient or property of Starfleet.

Heh, funny that - it's the episode we watched in my last Philosiphy course when we were discussing sentient AI in it... I guess it's part of the syllabus somewhere

[Kamikaze]

Originally posted by Star Dragon
Freespace: Beyond Redemption (or, when good AI's  go BAD! )

Hmm maybe I should change my storyline...

"Alpha-1 send a probe into the nebula"...

   "Um Control, we have a problem."

"What sort of problem Alpha-1?"

   "Kind of a mechanical one..."

"Can you be more specific?"

   "The probe says it doesn't wanna go in there cause it's afraid of the dark!"

(Oh yeah we need smarter AI's in the game, but NOT like that!)

If we do develop "emotional" bots I hope it's used only in places where it is practical...
No use in a dishwasher but maybe in an anime-chick-maid-bot

[Nico]

Originally posted by Kamikaze



If we do develop "emotional" bots I hope it's used only in places where it is practical...
No use in a dishwasher but maybe in an anime-chick-maid-bot

that's exactly what I was thinking let's make Mahoro

[Kamikaze]

Originally posted by venom2506


that's exactly what I was thinking let's make Mahoro

hehehe, but I prefer more grown bodies like CBD May

[Kazashi]

Originally posted by Kamikaze
Well, who's definition of "sentient" are we going by here? If we decide the dictionary is the ultimate jurisdiction, I'd say sure...

1. Has a concept of Self. I.e. It recognizes an image in a mirror as a reflection of itself and not another being.

a program could be programmed to "see" itself in the mirror and deduce that it is itself. Just do a sort of match-image-up sort of thing or have it deduce that a reflection is an image of "itself" and the surrounding objects.

2. Has a rudimentary ability to Reason. I.e. It can follow a path of logic and come up with an answer

Computers can do this. What seems interesting is that humans tend to reason more poorly than humans, often taking options that would decrease their productivity/reproduction or others'. (assuming that productivity and reproduction are the most important factors for animals...)

3. Has a desire to maintain it's existence.

That could be programmed to.... (depending on what you call "desire", I'm assuming a programmed survival trati counts)

I always thought that determining sentience of a machine relied on what the machine could learn itself. That is to say, if you start to program an AI but leave things such as self-awareness out of the program, it will be a smart machine, but not sentient. If the AI can then learn by itself to recognise itself as a seperate entity from the rest of the universe, then that would indicate sentience by the definitions laid out earlier.

ShadowWolf's comment strikes a chord within me. If you make a slave machine sentient, then place it in the position where it recognises that it's being held in slavery, what happens if it wants out? If a household appliance can do its job without needing to be self aware, then why add it? It might be fine to give a certain computer sentience because it deals with humans, such as handling welfare cases - assessing dole and disability payments, providing company to the homeless and lonely children, but does an android being sent into a potentially fatal pool of acid need to recognise that it might be in danger? If it recognises its situaiton it might want to get out of it, leaving the job unfinished.

[CP5670]

Wouldn't that definition of sentience leave out the human, however?

[Kamikaze]

Originally posted by Kazashi

I always thought that determining sentience of a machine relied on what the machine could learn itself. That is to say, if you start to program an AI but leave things such as self-awareness out of the program, it will be a smart machine, but not sentient. If the AI can then learn by itself to recognise itself as a seperate entity from the rest of the universe, then that would indicate sentience by the definitions laid out earlier.

self-awareness can most likely be programmed, and the fuzzy logic required to learn by itself is programmed into the AI...

ShadowWolf's comment strikes a chord within me. If you make a slave machine sentient, then place it in the position where it recognises that it's being held in slavery, what happens if it wants out? If a household appliance can do its job without needing to be self aware, then why add it? It might be fine to give a certain computer sentience because it deals with humans, such as handling welfare cases - assessing dole and disability payments, providing company to the homeless and lonely children, but does an android being sent into a potentially fatal pool of acid need to recognise that it might be in danger? If it recognises its situaiton it might want to get out of it, leaving the job unfinished.

yes, unpractical use of sentience is.... well, not too practical It'll make something that could have been productive without sentience, not as productive. Basically what I said earlier about CBD may and the dishwasher....

[CP5670]

Actually, I think that when we get a truly sentient computer (with some definition of sentience that includes the human but not the computer of today), that will be the last barrier between the two; the human and the computer will henceforth become completely indistinguishable, and so for all purposes would be the same thing.

[Grey Wolf]

Originally posted by Thunder

Heh, funny that - it's the episode we watched in my last Philosiphy course when we were discussing sentient AI in it... I guess it's part of the syllabus somewhere

You watch TV shows/videos made in the last thirty years that don't come from an educational crap company? I'm jealous.

[Styxx]

Originally posted by Shrike
Why would anyone want a sentient dishwasher anyhow?

So you can talk to it while it washes your dishes?