Artificial Intelligence
- Thread starter Pogue Mahone
- Start date
How far away are intelligent sex robots?
I'd settle for a stupid one tbh.
I'm a third through the read... whew. Not sure if I have the time or attention span to finish this.
Stick with it. The juiciest stuff is near the end.
Your man badly needs a good editor. Often takes 500 words to say what could be said in <100.
Another skeptical link about predicting the development of AI.
https://intelligence.org/files/PredictingAI.pdf
https://intelligence.org/files/PredictingAI.pdf
Aye, ditto.Mentioned before that researchers have been known to give super-optimistic predictions. On the fifties, they thought that they can solve everything with perceptrons (a very primitive form of neural networks) and then Marven Minsky showed (and gave proofs) that they can't solve even trivial problems like XOR. He also showed that even other forms of ANN which are based on that idea won't work. Then researchers left ANN, until on the eighties Rumelhart, Hinton and co. discovered back propagation which is a network that propagates the error backward and then using the derivatives tries to fix it. Then they became awesome again.
Anyway, now I think that the prediction holds more weight. Back then it was just enthusiasm, now we are seeing real results. And interesting algorithms are solving difficult problems (for example an ANN made by Ng and co which gets 10 billion parameters was able to detect cats even without showing the program what is a cat in the first place). I guess that this time they may be right.
On the other side, we have those who say that it will never happen because we have some survival instict, are created by God or other nonsense reasons. Likely people were saying the same that no robot will be able to defeat a chess world champion a few decades ago.
Exactly. But then again, it is very difficult to find solutions to the problem we currently don't have. What I think though, is that when we'll have an AI smarter than us (which will also be able to either change it's source code or create other AI which are even smarter) then all bets will be off. Even in relatively primitive forms of AI, we see many times results which shouldn't happen or generalization which don't make sense.
That isn't AI though. But then again, what is to stop an AI using nanobots to do exactly that. And even without AI, humans have a history of either making errors of judgement or just being careless (like for example: http://sunnyday.mit.edu/accidents/Ariane5accidentreport.html - which is such a banal mistake).
I haven't read the piece (Pogue's warning about its length frightened me off - I'll get around to it eventually), but I don't believe we're any closer to AI now than we were in the early 70s, when everyone was predicting that machines like HAL 9000 - the computer in 2001: A Space Odyssey - would be a reality before the date in the film's title. We know how that worked out.
Can something be intelligent without being conscious? And how is consciousness programmed? Computers can be programmed to perform certain intellectual tasks superlatively well - for instance play chess - but they're still just glorified adding machines, whose every output is entirely predictable.
There's a qualitative difference between what a brain does and the processes of an 'intelligent' machine, and ever increasing computing power isn't going to turn apples into oranges.
Can something be intelligent without being conscious? And how is consciousness programmed? Computers can be programmed to perform certain intellectual tasks superlatively well - for instance play chess - but they're still just glorified adding machines, whose every output is entirely predictable.
There's a qualitative difference between what a brain does and the processes of an 'intelligent' machine, and ever increasing computing power isn't going to turn apples into oranges.
Yeah, I heard someone else being very cynical about the 20 year timeframe predicted by the experts. As he put it, predict something will happen in 50 years and you won't get a grant to research it, predict it will happen in 5 years and you'll actually have to deliver so a 20 year timeframe is perfect! He also confirmed what you've said in that quote. Historical surveys have repeatedly generated the same timeframe. Back in the 60s everyone was predicting ASI in 20 years too.
God I hated that movie.
From what I can gather that's kind of the point isn't it? They'll make decisions differently to us, in a calculated way, making it potentially much more severe and powerful. If they are self learning like has been said they'll become better and better on their own and be able to fend for themselves entirely, doing as they wish with humans.
Intelligent machines have to be about the scariest thing you can have coming after you, right?
Interesting! Any good? Like stuff like that but don't get much chance to read any more.
I haven't seen it for more than 3 decades, which, I suppose, says something about its entertainment value, but it's certainly highly regarded by the critics.
I will only accept Artificial Intelligence if it watches that movie and tells me that the movie's crap.
Sure is. Banks is by far the best scifi writer imo.
AI should be something that makes all human life to become more at ease. With capitalism though that is something which I fear, the manual and even such tasks that can be computed with out such labour should ease those in the lowest equity. It won't though, the class at the bottom will be further punished by those who will lose such income. With the advancement of AI and computing on a whole through such services that use such means should lead to a more socialist society, I very much doubt that will happen. At the very least those they do continue to be employed should have the same wage with less working hours and share their job with someone else to conform to the hours/labour required. It will certainly be interesting times ahead with the current philosophy, growing populous and lack of human demand in all of the job sectors around.
Creating machines that intellectually supersede us, render us obsolete and then decide to destroy us for wasting resources would surely win the human stupidity competition.
That's not necessarily true. I can't recall the persons name but the most advanced computer code in terms of binary matches the binary code of humanity. I shall try to recall the persons name who come up with such theory (in the 50's) and that who is currently investigating such in this period of time. They/we have identical binary codes.
Yeah I assume we'll always have the 'pull the plug' option.
An article with some quotes from Andrew Ng (one of the leading AI researchers and a great guy):
http://fusion.net/story/54583/the-case-against-killer-robots-from-a-guy-actually-building-ai/
http://fusion.net/story/54583/the-case-against-killer-robots-from-a-guy-actually-building-ai/
Long reading but interesting:
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
Reading it now, pretty good. Stuff like this pisses me off though:
"As of now, the human brain is the most complex object in the known universe."
Incorrect - cetacean brains are more complex. No matter which way you measure the complexity of a brain - volume, crenellations, number of lobes, number of neurons, structural complexity - cetacean brains are more complex. People are depressingly fixated on the idea that technology = intelligence. Are the people around today more intelligent than Albert Einstein because they have an iPhone and he didn't? Physiological intelligence is not directly related to cultural technology level. Hell, we don't even have a clear definition of what "intelligence" really means, that we strut around congratulating ourselves on being the best at it is ridiculous!
That was a very good read. No idea how accurate it is, but very interesting
From what I've read it's pretty accurate. I have a book (which is now a few years old) about the technological singularity, and have read articles by the people behind the ideas (Vernor Vinge for example) and it's all very much legit. Amazing really.
It is very speculative to be fair, but what is pretty sure is that when AI reaches human intelligence (and assuming that it will have enough energy) then for a very short time it will reach superhuman intelligence and it would basically be a God. The time when this will happen, cannot be estimated but according to most experts, it will happen soon enough.
The good news are that there aren't many (if any) researchs in strong AI (though there might be projects from states which are kept at secret), while there are plenty of researchs (we're talking for billions of dollars) on weak AI. And weak AI is awesome and helps on improving the quality of life.
But on the other side, I am a believer that technological singularity will happen within 50 years or at worst, within this century. And then all bets are off and 'will it be a nice God' might be a legitimate question.
I've been pondering all this and the Space/Astronomy thread over the last day or so and was thinking about how a galaxy could be colonised by machines in 3.75m years and combined with Super Intelligent AI it makes for some interesting possibilities.
What if Organic life was always the precursor to something much bigger.
So let's say we are the peak of existence as far as organic life goes and the great filter is Super Intelligent AI. So we create this AI which then thanks to its makeup of metal, plastics, silicon and other hard wearing materials and ability to repair and replicate in a way biotic organisms never could, immortality is realistic for them and time becomes somewhat irrelevant.
So this AI sets out to colonise the Universe, harness all its materials and energys and turn it into something entirely different. At the moment it's a big empty space full of nothing but raw materials and possibly life on planets so far and wide that the majority of them are unaware of each other and the time it takes to communicate let alone travel between even the smallest of distances makes it completely impossible.
What AI could actually do is set out to change the universe in such a way that it is no longer such an empty, ufathomable place. If machines could colonise the Milky Way in 3.75m years by spreading out exponentially then what progress could they make in 1 billion years? How could AI shape the Universe in that time?
The big question for me would be what would be its purpose in doing this? After all we would be its creator and as such it's God, what purpose could we come up with for machines to spread out amongst the Universe and harness its energy and materials?
It could take human DNA and forge life on exoplanets that have potential but no current life yet. In fact it could take the DNA of every animal from Earth and drop it all over the universe in habitable places. Maybe the technology could be in place to preserve our brains, clone ourselves many times and have the bodies farmed so they are ready for our brain to be implanted and achieve immortality ourselves.
aargh too many possibilities.
What if Organic life was always the precursor to something much bigger.
So let's say we are the peak of existence as far as organic life goes and the great filter is Super Intelligent AI. So we create this AI which then thanks to its makeup of metal, plastics, silicon and other hard wearing materials and ability to repair and replicate in a way biotic organisms never could, immortality is realistic for them and time becomes somewhat irrelevant.
So this AI sets out to colonise the Universe, harness all its materials and energys and turn it into something entirely different. At the moment it's a big empty space full of nothing but raw materials and possibly life on planets so far and wide that the majority of them are unaware of each other and the time it takes to communicate let alone travel between even the smallest of distances makes it completely impossible.
What AI could actually do is set out to change the universe in such a way that it is no longer such an empty, ufathomable place. If machines could colonise the Milky Way in 3.75m years by spreading out exponentially then what progress could they make in 1 billion years? How could AI shape the Universe in that time?
The big question for me would be what would be its purpose in doing this? After all we would be its creator and as such it's God, what purpose could we come up with for machines to spread out amongst the Universe and harness its energy and materials?
It could take human DNA and forge life on exoplanets that have potential but no current life yet. In fact it could take the DNA of every animal from Earth and drop it all over the universe in habitable places. Maybe the technology could be in place to preserve our brains, clone ourselves many times and have the bodies farmed so they are ready for our brain to be implanted and achieve immortality ourselves.
aargh too many possibilities.
That's the article Pogue linked to in the OP (albeit he linked straight to part two)
Still, it's a marvelous article, and blog in general. Everyone should read it. Elon Musk gave it his blessing by calling it a good primer on exponential advancement of technology.
Interesting that you mention this kind of thing. I don't think anything like that could ever work.
Take humans, for example. What percentage of our behaviour is hardwired into our genes, and what percentage is aquired through experiencing life? A lot of people like to think that humans act completely outside the influence of their genes (often referred to as derogatory "base instincts"), but that's definitely not the case. Regardless, a huge percentage is aquired during your life, and the source of that behaviour is interaction with others in (and outside) our culture. Take a perfect human clone and place it on a world without that culture, and it would not grow up to be like us, because it does not have any of the guiding references or interpersonal interactions that we learn from.
The only way colonisation via cloning would be possible would be to replicate human society on the other planet - either by transporting a small number of humans to that place (which defeats the point of cloning), or by replicating that small number of humans using humanoid robots (which begs the question: if you can create robots that are just like humans, why would you subject some poor souls to our imperfections when you could just populate the universe with an artificially improved version?).
It's usually pretty easy for people to understand that argument when it comes to humans. What they find hard for some reason is to accept that for all other sentient species the situation is the same. And sentience is present in at least all other mammals, all birds, and many others including invertebrates such as octopuses. Recreating the Earth's habitat on another world is a hell of a lot more difficult than just growing plants from stored DNA.
This actually seems pretty likely in a sense - scanning a brain and emulating it using software is a very clear possibility. Nondestructive scanning would be more difficult though, but would have the slight advantage that it wouldn't kill you
in the process
Last edited:
The biggie would be to transfer consciousness to a safe replicable host ie immortality. Who needs god or to be good?
What if the God is an ulta-advanced AI of a far more advanced species than us? Then the AI took over, exterminated them and started playing games on galaxy (universe) by planting simpletons like humans.
Well we have achieved far more than dolphins: the wheel, New York, wars and so on.Reading it now, pretty good. Stuff like this pisses me off though:
"As of now, the human brain is the most complex object in the known universe."
Incorrect - cetacean brains are more complex. No matter which way you measure the complexity of a brain - volume, crenellations, number of lobes, number of neurons, structural complexity - cetacean brains are more complex. People are depressingly fixated on the idea that technology = intelligence. Are the people around today more intelligent than Albert Einstein because they have an iPhone and he didn't? Physiological intelligence is not directly related to cultural technology level. Hell, we don't even have a clear definition of what "intelligence" really means, that we strut around congratulating ourselves on being the best at it is ridiculous!
We are more intelligient than them despite that they have more neurons and more synapses. I think that the reason might be that they don't have hands. I really think that if we didn't manage to free our hands, then we wouldn't ever become an 'intelligient' specie.
What use would they have for any of these things? They are already at the top of their food chains. We had to use our brains and thumbs to get to the top of ours.
An idea expanded upon in this science fiction book. The farmed bodies are called "sleeves" and can have all sorts of funky implants and upgrades put in them. It's a fun read.
It was a joke borrowed from 'The Hitchhikers Guide to the Galaxy'.
I think that humans managed to became top of Earth species thanks (to a large degree) of the ability to use our hands (most of the animals cannot).
Lol wars
Yeah I realised it was at least tongue in cheek. I was just making the point that we evolved as we did because we had to. I think that's also something that is missed when discussing extra terrestrial life. The chances are that the majority of planets harbouring life probably have a very simple food chain that doesn't see the need for one species to achieve human like intelligence to survive, just being a leathal hunter is enough.
When you add in the sheer number of planets that can potentially harbour life that we're discovering though (extrapolated, there are billions in this galaxy), even a tiny percentage of them creating a human-like species is a very large number. But after that's when you get into the hazier of the estimates in the Drake equation.
Good rationalization, and something that I didn't think of before. It may be that if the planet has more resources than Earth, then species won't ever become that much intelligent, simply because they don't need to be intelligent in order to survive.
Obviously, it might be the other extreme but anyway, all bets are off.