The new forums will be named Coin Return (based on the most recent vote)! You can check on the status and timeline of the transition to the new forums here.
The Guiding Principles and New Rules document is now in effect.
Whatever happened to artificial intelligence?
Thirith
Registered User regular
Registered User regular
I remember being pretty interested in artificial intelligence as a teenager, having grown up on movies like 2001 and Wargames. When games like Creatures announced that they would feature *real* AI - in other words, self-learning algorithms - I was excited; I was doubly excited when I read about Creatures being used (in the loosest sense) to create AI for jet planes (they'd give the AIs a number of 'genes' based on behaviour and reactions to the world, and they'd keep 'breeding' those AIs who survived the longest without crashing horribly, or at least that's how I remember).
I also remember programmes based on 20 Questions that would learn new criteria and questions and therefore become better at what they're doing.
However, these days it seems that artificial intelligence isn't really a big thing, or at least it's not spoken about outside the inner circle of AI afficionados.
Whatever happened to AI? Do any of you know what's going on these days? Are we any closer to machines or programmes that could pass the Turing Test without cheating (or without cheating too much)? Or is HAL 9000, like, so '60s?
I also remember programmes based on 20 Questions that would learn new criteria and questions and therefore become better at what they're doing.
However, these days it seems that artificial intelligence isn't really a big thing, or at least it's not spoken about outside the inner circle of AI afficionados.
Whatever happened to AI? Do any of you know what's going on these days? Are we any closer to machines or programmes that could pass the Turing Test without cheating (or without cheating too much)? Or is HAL 9000, like, so '60s?
"Nothing is gonna save us forever but a lot of things can save us today." - Night in the Woods
Thirith on
0
Posts
I too am hoping some company will feel a bit more ambitious one day though, and provide us with something new.
"Nothing is gonna save us forever but a lot of things can save us today." - Night in the Woods
We do have things like facial recognition and other limited knowledge/logic/perception programs which is a form of AI though.
I don't think we have anything remotely close to "real" AI yet however. I also don't think we've made much headway into the subject for the past 50 years. Hopefully something will break through soon though.
It would be pretty odd if we managed to truly create artificial intelligence when we don't even fully understand our own intelligence yet.
Everyone is banking on motion control.
3DS FC: 5343-7720-0490
http://www.singinst.org/
They're pretty much looking at what happens in Terminator and saying to themselves "This is the next step in human evolution! The human race will become obsolete, possibly extinct or enslaved, but that's evolution!". It's kinda scary that these scientists hope to bring about the doom of the entire human race and nobody actually has the foresight to stop them. I guess the foreshadowing of science fiction isn't a compelling enough argument against these people
Luckily, true AI is likely a pipedream. From all the advances I've seen during my whole life, nothing points towards computers achieving true sentience. I'm not sure if it's actually possible, I sure as hell hope it isn't. Humans really would have no place on this planet if AI sentience is achieved.
the whole "SKYNET WILL BECOME REAL AND THEN IT WILL ASSUME ULTIMATE POWER SOMEHOW AND KILL US FOR SOME REASON" is the doomiest and gloomiest of predictions possible
They're still trying to get robots to walk properly, nevermind think and speak for themselves.
I want to know more PA people on Twitter.
I guess
"Nothing is gonna save us forever but a lot of things can save us today." - Night in the Woods
http://en.wikipedia.org/wiki/Expert_system for the interested.
No, that seems pretty apt.
They are different but naturally, they're related.
It was just an example really, and I mentioned it because I was watching some robotics shows on Discovery awhile ago and they usually discuss A.I to an extent when they discuss robotics.
Basically where they are at now is creating programming for the A.I to recognize faces, objects and words, then commenting/answering regarding any questions they are asked regarding that thing or person.
It seems to be coming along well, but the reaction times and all that are really slow, and aren't 100% accurate.
I want to know more PA people on Twitter.
If you really want to know where game AI is today, go read some articles on http://aigamedev.com/
AI in games though, is not AI at all. It's just "Event A will react to Event B and then X happens". It's just a set of pre-determined actions.
Wat?
That really makes very little sense. The elements of a human brain are all mechanistic elements that behave according to strict laws of cause and effect, yet the net effect is intelligence. Why would that not be true if we change the parts to silicon rather than carbon?
Steam ID: slashx000______Twitter: @bill_at_zeboyd______ Facebook: Zeboyd Games
AI are made up of sets of rules that react to other sets of rules. If A happens then do B. If B happens then do C. Thats how they work. The rules are set within them and they follow what theyre programmed to do. Thus, they follow instructions. We, as humans, can react differently, while the AI reacts how they were programmed.
As far as I know, AI research is still going strong. You don't hear much about it because it's not very sexy, and most people have a pretty big misconception on what AI really encompasses (hint: it's nothing like what you see in the movies). When I was in grad school, one of my profs was really into natural language processing, automated text summarization, etc.
Wikipedia has a nice overview of the major topics in AI. Artifical Intelligence: A Modern Approach is a good introductory textbook, if you're into that sort of thing.
Could you make this more clear?
You still haven't explained the difference. Yes, there is structure in most AI, as there is structure in the human mind - consider the almost unvarying nature of the processing performed by the visual cortex - so what's the big deal?
AI does not consist of rules of behaviour in a strict sense; they consist of a computational system that result in behaviour. If you were defining how an AI would respond to every conceivable input, then yes they would be limited - but you aren't.
Think of it this way, if you go down a hallway in HL3 with a pistol and a grenade and come up to a barricade with 3 combine standing behind it, what good will the AI do?
I mean, it's easier just to say "Hay fuckers, when someone approaches, seek cover and try to kill them!" than program some elaborate AI that is going to reach that conclusion anyway.
Do androids dream of electric sheep? (hint: No.)
"Wat?" really you want to be that guy?
There is no intelligence there, as someone else pointed out they are just following instructions based on previous inputs. I'm not downplaying computers or the amazing things we accomplish with them but they are just machines and so far no amount of clever software or circuitry can get past that.
A computer doesn't understand, it just does what it's told albeit sometimes in a more roundabout way that gives the impression of intelligence. No matter what, it is still just reacting to some past instruction.
I want to know more PA people on Twitter.
The part I'm disagreeing with is the no matter what part.
If you break down any intelligent system the parts at the bottom are unintelligent. The problem is that peoplein this thread seem to assume that because modern AI is pretty bad, that it is in principle impossible to have an intelligent machine, which is quite literally a bizarre claim.
To take a silly but logically powerful example; I take a computer and simulate the input/output behaviour of a single neuron. I link this computer together with 100 billion other computers and simulate an entire brain. Unless you believe the emergent properties would be automatically different, you have simulated intelligence using computers. Hurrah.
"Nothing is gonna save us forever but a lot of things can save us today." - Night in the Woods
I've seen bits of research here and there over the years that suggest the human brain is a lot more program like than we'd like to admit. Like it or not.. when we make a decision, we're just reacting to some past instruction as well.
Clearly there's a difference between computers and humans, but I think it mostly revolves around humans having superior capacity to accept and process the input our senses provide us.
But that is the difference. Humans can react in a multitude of ways that no one could possibly expect. While AI will react in only the ways we have programmed into them. They cant make up a whole new way to react like a human could. A happens, then they search what can happen if A happens and choose the best scenario and bam, thats what you get.
Humans can react to any possible thing to any possible way. AI react with whatever theyre programmed to react with. Its that simple. There is no sentient AI all knowing god that can react how it wants like a human mind. They have rules they follow based on lines of code input into them.
For instance. Someone points a gun at you. You could do any number of things to respond to this. Maybe talk him down, back down, knock it away, those are probably the most likely scenarios. Now if an AI had a gun pointed at it and these are the rules set within it, and say maybe there was this perfect path to run away and he couldve made it. But theres only those 3 rules set within him. Hell only think of those 3 and react with the most likely scenario to keep him alive. While a human could make a break for it because he isnt restricted to only what he has programmed.
Well sure, and thats obvious. But what were trying to get at here is that AI is created by us, the humans. and will be given rules that we think of as what they are allowed to do. The AI cant find a way out of those rules and will only follow the set rules within itself. While a human can do whatever he wants. It doesnt matter how stupid or useless. They can do it.
What you guys arnt yet understanding is that AI is made up of rules and reactions. I know. I know. Youll say humans are too. But again, they will only react with whatever we as humans have programmed into them. Not anything else. Thats what Im trying to get at. They will not, by any means, do something we havent programmed. They stay within the boundaries of what we've created and follow that.
Ahh, I see the problem. You have not encountered the concept of emergence.
http://en.wikipedia.org/wiki/Emergence
Have a look. And I don't mean this as a brush-off, I quite genuinely think that reading that will satisfy why the AI can behave in the fashion you believe it can't.
I want to know more PA people on Twitter.
What is the distinction between flawless mimicry of behaviour and intelligence?
If your zombie is like a person in every respect, in what sense is it a zombie?
on top of that, we don't really have a good operating definition of intelligence, and we certainly don't really know how intelligence or consciousness exists in detail, so we can't yet reproduce it. i suspect we will do so by accident far before we actually understand how our own brains work in a mechanistic sense.