Artificial Intelligence: Why Would We Make One?

[Chirios]

Let's assume for the sake of argument, that in the future somebody figures out how to make an artificial intelligence. Why? What advantages would an AI bring to computers?

[someone_else]

Uhm, let's see, creating an uncontrollable thing able to fuck you over and over isn't a goal for anyone.

Sillyness aside, imho the only reason to make an AI (with the popular meaning of "making a human mind with code in a computer") is to study the brain itself.
You know, to learn about how something works you must try to fuck some specific part up and see what happens, but doing it on actual human brains is uhhhhh... immoral.
Waiting for bad things to casually happen on people and studying them is very inefficient, and chimps are just related to us, but not the same.

Making a batch of say 100 to 1000 half-decent AIs and then fucking their minds up with sadism that would make a SS go pale would be a good thing for neurology.

[FireNexus]

What advantages would something that could become orders of magnitude smarter than any human who has ever lived bring? It could design, plan and act faster and more effectively than anyone has ever been able to.

If you get it right, it could make everything better. If you get it wrong, it might kill everyone.

[Rabid]

Could you precise what you mean by an "Artificial Intelligence" ?

Do you mean an AI that is "like" a human mind, but far more powerful ;
Are you talking about something totally alien to our comprehension (well, apart from the the fact that WE designed it, there's just the fact that we can't relate to the things way of thinking), and possibly with its own goal/purpose ;
Or are you just talking about a [self-adapting, evolving] or [static, fixed] system that we can reliably use to micro/macro-manage our lives without real worry about it turning rogue or saying "Fuck you, I'm outta there".


The first one, well, you could imagine people wanting to have them as some sort of companions, or even to upload their [memory][mind-patterns] into it, for whatever reasons (immortality, or a hunger for MOAR POWER !!).
The third one, well... Deep down, admit it, we all want our very own, personal, ethic-dilemma-free slave, yeah ?
As for the second one, well, apart from some singularity-wankers trying to play God... I don't see any sane reason to do it. But I may be biased on the matter...

Making a batch of say 100 to 1000 half-decent AIs and then fucking their minds up with sadism that would make a SS go pale would be a good thing for neurology.

My Moral Sense is tingling... Help me there :

If it resemble a human mind,
If it work like a human mind,
...
... Doesn't it mean that we are basically screwing with what I'll basically (for the understanding of all...) call a HUMAN SOUL [a person's 'spirit', 'mental essence', 'personnality', whatever term suits you best, really...] ? FOR SCIENCE ?!

*wipe a little tear and smile proudly*

Josef Mengele would be so proud...

What advantages would something that could become orders of magnitude smarter than any human who has ever lived bring? It could design, plan and act faster and more effectively than anyone has ever been able to.

If you get it right, it could make everything better. If you get it wrong, it might kill everyone.

Well I think you answered your question, here. The only real question is :

How do we design it so that it doesn't go Skynet on us ?

Some thinks they have the answer. I don't know. Only time will tell.

[Formless]

Sillyness aside, imho the only reason to make an AI (with the popular meaning of "making a human mind with code in a computer") is to study the brain itself.
You know, to learn about how something works you must try to fuck some specific part up and see what happens, but doing it on actual human brains is uhhhhh... immoral.
Waiting for bad things to casually happen on people and studying them is very inefficient, and chimps are just related to us, but not the same.

What. The. Hell are you smoking? If you want to make an AI similar to the human mind/brain, you have your research agenda ass backwards. FIRST you figure out the brain, THEN you make an AI. Not that most AI researchers want to emulate the human mind/brain in the first place. Also, there are far better research methods than "break it and see what happens" like brain imaging technology, and even if you do want to go that rout there are methods for temporarily disrupting brain activity in select parts that have been used to study moral reasoning.

[Chirios]

Could you precise what you mean by an "Artificial Intelligence" ?

A "thinking" machine, a machine with problem solving ability that shows complete human-like sapience, but with a computer's processing power.

[Rabid]

So, basically, a system that emulate the functioning of a human brain, with superior "computing" ability, and that isn't based on some sort of Wet-ware (flesh is weak, metal endure).

Am I understanding it right ?

[Broomstick]

Assuming you're talking about synthetics that are not squishy organics in any way, one reason to build an AI is space exploration. Communication lags are enormous in space due to the speed of light limitations and a probe, mining, or other device out past a few light minute's distance that could identify and solve problems on its own would be much better than telepresence.

[Lord of the Abyss]

Assuming you're talking about synthetics that are not squishy organics in any way, one reason to build an AI is space exploration. Communication lags are enormous in space due to the speed of light limitations and a probe, mining, or other device out past a few light minute's distance that could identify and solve problems on its own would be much better than telepresence.

Nor would it need a massive life support system, and even if it is capable of boredom unlike a human it can just shut itself off while it takes months or years to get to the destination.

[Singular Intellect]

The benefits of AI are blindingly obvious if one merely imagines how they'd personally be affected by having the computational power, memory stability and capacity of current machines interfaced directly with their own intelligence.

Sure, we get hand waving dismissals and appeals to human shortcomings as the 'important' aspects of being human. The multi millions of users of compact information devices (like smart phones), on the other hand, shows how weak the argument is about people not being interested in enhancing their information processing capabilities as much as they possibly can.

Advancing AI is just more steps down the road people like to pretend we're not walking down, because they are either fearful, ignorant or in denial of where we're heading.

[Broomstick]

The benefits of AI are blindingly obvious if one merely imagines how they'd personally be affected by having the computational power, memory stability and capacity of current machines interfaced directly with their own intelligence.

"Interfaced with their own intelligence" is different than having their intelligence displaced or replaced.

Sure, we get hand waving dismissals and appeals to human shortcomings as the 'important' aspects of being human.

Actually, it's the human capacity to recognize deviation from routine and improvise a new solution that is our real asset. An AI might have a similar capability, but I don't see it as inevitable.

The multi millions of users of compact information devices (like smart phones), on the other hand, shows how weak the argument is about people not being interested in enhancing their information processing capabilities as much as they possibly can.

Meanwhile, we'll just ignore all the traffic accidents that result from people "enhancing their information processing capabilities" instead of paying attention to their driving. It circumstances like people texting while driving that leads me to think not all "enhancements" are actually a good thing. There is a right way and a wrong way to do such things.

Advancing AI is just more steps down the road people like to pretend we're not walking down, because they are either fearful, ignorant or in denial of where we're heading.

As long as I can turn the shit off when I want some peace and quiet I'll consider the upgrade.

[Korvan]

The benefits in education alone would be staggering. Imagine every student with their own personal teacher who is top in whatever field is relevant, is able to tailor the course to fit the student's exact learning needs and abilities and is able to devout all their (run)time to that student.

[Singular Intellect]

"Interfaced with their own intelligence" is different than having their intelligence displaced or replaced.

I'm assumng the OP means AI in the sense of having human capabilities for prediction and pattern recognition. By default, such intelligence will be integrated with computer systems with their superior traits.

Actually, it's the human capacity to recognize deviation from routine and improvise a new solution that is our real asset. An AI might have a similar capability, but I don't see it as inevitable.

Again, I'm assuming the OP meant AI reflecting human intelligence traits.

Meanwhile, we'll just ignore all the traffic accidents that result from people "enhancing their information processing capabilities" instead of paying attention to their driving. It circumstances like people texting while driving that leads me to think not all "enhancements" are actually a good thing. There is a right way and a wrong way to do such things.

Of course there are bumps, mishaps and problems to be worked out. AI driven cars and restricted use of wireless devices enforced by a operated vehicle are just a couple of solutions in the field.

As long as I can turn the shit off when I want some peace and quiet I'll consider the upgrade.

Even if that peace and quiet is your personally designed virtual world relaxation experience?

[Broomstick]

There's a lot to be said for experiencing reality. For one thing, I don't have to go through all the exhausting work of making up every little detail.

[Singular Intellect]

There's a lot to be said for experiencing reality. For one thing, I don't have to go through all the exhausting work of making up every little detail.

That's like saying an artist needs to worry about every molecule of paint.

Furthermore, you of all people know there's plenty of shit in reality many of us don't want or need to experience.

[Broomstick]

Well, yes, of course there's shit in reality - that doesn't mean it's ALL shit!

There's crap on the internet, too. I'm not interested in just any virtual reality, I want a VR that's at least as good as the good stuff in reality. I don't adopt stuff just because it's new or shiny or high tech, I only want to bother with stuff that will actually make my life better in some way. I insist on quality.

[Singular Intellect]

Well, yes, of course there's shit in reality - that doesn't mean it's ALL shit!

Obviously not, but experiencing a vacation whereas you can selectively cut out the bad parts is hardly something you could complain about, is it?

There's crap on the internet, too. I'm not interested in just any virtual reality, I want a VR that's at least as good as the good stuff in reality. I don't adopt stuff just because it's new or shiny or high tech, I only want to bother with stuff that will actually make my life better in some way. I insist on quality.

I'm pretty sure most people would insist on high quality simulations. I suspect it will actually be easier to fool the human brain than people tend think. Hell, the damn thing fools itself on a regular basis.

Anyhow, I don't intent to hijack the thread for discussion of virtual reality.

[Skgoa]

OK, there is a lot of bullshit in this thread, so lets look at the question in a more methodic way:

A system is called intelligent when it can react to stimuli/achieve goals in ways that were not directly programed into it. I.e. by intelligently choosing a course of action. Systems that do that exist. (Most likely your spam filter is one.) Actualy, they have been around for a while. They are a great way to make life easier for people, since they:
- filter spam from your inbox
- provide opponents in computer games
- can remember assloads of information and thus can provide doctors with likely causes for a given set of symptoms
- can shoot down missiles before they hit their target
- make it unneccessary to sent humans into space only to resupply the ISS
... and many more, that list is only what I came up with while writing it.

There is no doubt that this limited or "narrow" AI has huge benefits and is economically valuable. And thats why we make them: they provide a significant benefit that offsets the cost of developing them.


But the OP has clarified that he is talking about human-like sentient AI, i.e. the so called General AI. GAI has not been achieved. And its not the focus of AI research. Yes, there are a bunch of AI wankers who dream about it, but no one has come even close. The benefits are... difficult to imagine. No task that an AI can help us with needs sentience. As the Chinese Room thought experiment shows: the system's internal state doesn't really matter, its all about it's interactions.
Actually, the huge increase in complexity (i.e. cost) means that gAI is an inferior solution to any problem we might want to throw it at.
Note: I am talking in a utilitarian sense. Obviously its an interesting field FOR SCIENCE!

[Sarevok]

I think the biggest appeal of "AI" that can handle typical human tasks is the prospect of never working again for a living. Most people would not really care about the semantics or the notion of a machine being able to think once this becomes obvious. If for instance a group of networked machines could build houses, dams and factories all by themselves like RTS videogames; then suddenly the question of AI becomes different. It is no longer about deep philosophical pondering but rather where do my own AIs do the things I could never do myself alone.

[Sarevok]

Let's assume for the sake of argument, that in the future somebody figures out how to make an artificial intelligence. Why? What advantages would an AI bring to computers?

My current job is a shitty one. I have to write programs by hand and I hate it. If there was an AI that could take specifications from me and turn it into a working software it would be a dream come true.

That is an example of an advantage AI would bring to computers in real world use. Right now computers relieve people from manual labor. But a working AI would also get rid of much intellectual labor.

I am not saying a world without much thinking required would necessarily be a net positive outcome. But not having to do the typical repetitive grunt work with computers everyone has to contend with anymore would be a welcome change of pace.

[Skgoa]

Well, we already have "executable pseudo-code" (e.g. Python, markup languages), interpreted higher level languages and automatic code creation from ontologies. Thus the need for low level coding is getting less, I am with you on hoping that its going to be totally obsolete rather sooner than later. Cheap labor in China and India might make it more cost effective to simply sent the design documents over there, though.

[adam_grif]

Let's assume for the sake of argument, that in the future somebody figures out how to make an artificial intelligence. Why? What advantages would an AI bring to computers?

The future is now! I present to you a working artificial intelligence:

http://www.google.com

Assuming you meant "Strong A.I.", then the answer is "lots of reasons". Some people want to do it purely as an achievement milestone (like landing on the moon for neuroscience and computing), but mostly because having human level intelligence combined with everything computers are good at (heavy mathematics, blindingly fast calculations, interfacing with the internet etc) nets you an extremely useful entity. When you move beyond human intelligence with them, then it's like having a brilliant human mind that never gets tired, exhausted etc, always has perfect concentration, can multitask with no issues, is completely loyal and thinks faster than any normal human.

[Lagmonster]

I'm no specialist in the field, but it seems to me that a lot of what slows down a human mind are biological input - hunger, fatigue, pain - and limitations of the way our brains work, namely self-delusion, boredom, and focus/attention span issues. Assuming I'm not wrong, I'd figure that removing all of that from the equation could only be a good idea - providing a mind with the ability to work without many of the limitations we endure.

And I never got the whole 'skynet evil AI' thing. I'd be like Hitler's head in a jar - yeah, it's thinking evil things, but it can't do anything other than express itself. Now, if some dumbass gives it control over something important, that might be dumb up until the point where you realize how to use an off switch or a hammer.

[Lord of the Abyss]

And I never got the whole 'skynet evil AI' thing. I'd be like Hitler's head in a jar - yeah, it's thinking evil things, but it can't do anything other than express itself. Now, if some dumbass gives it control over something important, that might be dumb up until the point where you realize how to use an off switch or a hammer.

Well, for one thing if you keep it isolated you not only prevent it from doing anything dangerous; you also prevent it from doing all sorts of useful things too. For another, what if some other nation/group puts less restrictions on their AIs and as a result begins to economically or militarily overtake you? I recall a line from a Norman Spinrad story about that scenario; "The side that gives the machines the most freedom always wins". If the things are genuinely superhuman in their capabilities, then they'd be too big an edge not to use, and the guys keeping their AIs in an isolated box won't get nearly as much out of them as the guys who don't. For that matter, it seems logical that a superhuman general AI could do an awful lot just by talking to people; subtly manipulating them.

As for "an off switch or a hammer", if the thing is a genuine superior intelligence we could easily be blindsided by the machine before we realized there was anything to worry about. Especially if its goals are more subtle than "kill all humans".

[someone_else]

Doesn't it mean that we are basically screwing with what I'll basically (for the understanding of all...) call a HUMAN SOUL [a person's 'spirit', 'mental essence', 'personnality', whatever term suits you best, really...] ? FOR SCIENCE ?!

I really hope a human soul isn't a hardcoded programming in a biological brain. But if that is true, then I'd have no complaints in emulating it for research purposes, to improve the life of copuntless people afterwards.
As an aside, not being made of meat will make a pain in the ass to legislate this kind of AI out of existence without fucking the rest of non-AI computers.

Frankly, I don't approve a lot even experimenting stuff with animals. But research has always had a "fuck you all non-human living beings" attitude.

FIRST you figure out the brain, THEN you make an AI.

Clearly speaking out of your ass. Never studied scientific method at school?
FIRST you obseve the phenomenon, SECOND you make theories, THIRD you test theories to see if those are true or not.
And in the THIRD point, making a machine that follows your theories becomes useful. You can place the machine in situations that you cannot recreate on humans due to either moral implications or time constraints (hell, if you have to wait 20 years for the result of a single experiment you're going quite slow). If the machines react in the same way as the human would (more or less) the theory is valid and can be used as a basis for new observations.
If the machines do not react the way human would, then you trash or correct the theory and try again.
Rinse and repeat.
Of course the first machines won't be particularly remarkable, but in time your computer models will become much more similar to human minds, maybe reaching what most people think an AI is.

Also, there are far better research methods than "break it and see what happens" like brain imaging technology

That is true if you're talking of human beings (that shouldn't be harmed by your research). If you have a batch of AIs you can do it much faster by "breaking up" things since it is just lines of code.
Also, having a blank memory slate allows you more control on what causes what in its behaviour.

My current job is a shitty one. I have to write programs by hand and I hate it. If there was an AI that could take specifications from me and turn it into a working software it would be a dream come true.

You would also lose your job.

Meanwhile, we'll just ignore all the traffic accidents that result from people "enhancing their information processing capabilities" instead of paying attention to their driving.

This is a very good point. Human minds aren't easy to control, but you can exploit their weaknesses (and most politicians do so very heavily these days). If you make a thing that can think faster and better than you without the obvious weaknesses of human minds.... you can only prey your favourite gods that its agenda matches your goals. But you have no real control whatsoever over it.
So you cannot place it in places where it can use its power.
That means someone will ask you what the hell are you going to use it for. Mopping floors?
Marvin is a good example.