|
Rutibex posted:thats not a very accurate history simulation if the physics works differently. unless the universe that is simulating us is itself a simulation!? its like the matrix inside of another matrix
|
# ? Oct 18, 2016 04:08 |
|
|
# ? May 27, 2024 05:29 |
|
Pookum posted:its like the matrix inside of another matrix I wonder how much they degrade in quality as the go down the chain in simulation of simulations; kind of like a game of telephone.
|
# ? Oct 18, 2016 04:14 |
|
Secular Humanist posted:Stephen Hawking thinks AI will become a threat to humanity? Well, he would know. A very smart thing with limited physical presence, yep, that's definately in Hawking's bailiwick. And as we know, Hawking has conquered the world several times already.
|
# ? Oct 18, 2016 06:05 |
|
reignofevil posted:Imagine trying to program a robot to be really good at chemistry OP. Robots were already at that level 7 years ago.
|
# ? Oct 18, 2016 06:38 |
|
LP0 ON FIRE posted:I wonder how much they degrade in quality as the go down the chain in simulation of simulations; kind of like a game of telephone. Would be cool if they were all made just to see if people in then can figure out how to break out. End game would be every sim dude all the way down the chain breaks through all the sims into the real world.
|
# ? Oct 18, 2016 06:42 |
|
Pookum posted:I think we are a computer simulated copy of the real world. Some highly advanced being out there is running a history simulation to either develop ai or study something that happened in their past. This is why the double slit experiment thing exists. When you don't observe something it acts differently because the computer that's running us is saving processing power by taking shortcuts. You're right, but you aren't finishing the thought. If our world is simulated, other worlds must be too, including those capable of simulating others. Our world, or more precisely, this moment you are experiencing now, must therefore be simulated not once, but an infinite amount of times, along with every other possible moment. This possibility of existence that some moments possess is, as you might observe, due to their logical consistency. Furthermore, one might apply this same train of logic to any universe in this infinite network of simulation, and we arrive at the startling conclusion that all universes are simulated. The true nature of reality is that logical consistency. Mathematical truth is, exactly, reality. One might construct an infinitude of scenarios in which you are a brain in a vat, or an AI construction, or the dream of a butterfly, or even a collection of wave functions, all of them absolutely true because this moment is true, as static and unchanging as a circle.
|
# ? Oct 18, 2016 13:45 |
|
Rockin Orthodontist posted:Well, he would know. A very smart thing with limited physical presence, yep, that's definately in Hawking's bailiwick. And as we know, Hawking has conquered the world several times already. lol
|
# ? Oct 18, 2016 13:51 |
|
Woden posted:Robots were already at that level 7 years ago. I guess if there's an article on Wired about it I better step the gently caress back. quote:“There isn’t any intrinsic reason why that wouldn’t happen,” says King. “I think there’s a continuum between the really basic types of science that you’d get from Adam, and the things I can do, and then Einstein-type science. A computer can make beautiful chess moves, but it’s not doing anything special. It’s just doing more of the same thing. In my view that’s what’s going to happen in science.” Lol a robot scientist AI would have read the whole article. reignofevil fucked around with this message at 14:55 on Oct 18, 2016 |
# ? Oct 18, 2016 14:52 |
|
If we can't even get a computer to play Civilization well, I don't think we'll get them to run a real world civilization.
|
# ? Oct 18, 2016 14:59 |
|
Assuming there won't ever be superhuman intelligent AI is to assume there is some magic property of a human brain that makes it capable of general intelligence that a machine could never have due to being a feeble non-magic invention. Someone who doesn't take this threat seriously might as well just become religious, because they believe in magic already.
|
# ? Oct 18, 2016 15:21 |
|
reignofevil posted:I guess if there's an article on Wired about it I better step the gently caress back. reignofevil posted:
You said a robot wouldn't be able to discover anything for 150 years, they've discovered poo poo already. Woden fucked around with this message at 15:36 on Oct 18, 2016 |
# ? Oct 18, 2016 15:32 |
|
Woden posted:BBC better? Or Nat Geo? quote:Imagine trying to program a robot to be really good at chemistry OP. quote:A computer can make beautiful chess moves, but it’s not doing anything special. It’s just doing more of the same thing. In my view that’s what’s going to happen in science.” scientist quote:You said a robot wouldn't be able to discover anything for 150 years, they've discovered poo poo already. You. This is like arguing that a robot discovered the trillionth digit of pi. Did a robot do the work? Arguably. Did the robot do this work entirely because a scientist/programmer understood the entirety of the concept and rather than sit down and work out the trillionth digit with a pen and paper they just programmed a robot to hold the proper variables and set it off to the races. Its a far loving cry from scientfic discovery to program a robot to further explore a well understood phenomena (such as the genome of yeast) and to do some documentation on previously poorly understood parts of our scientific knowledge. (Don't get me wrong though it is neat as hell. But it sure as gently caress isn't "an AI discovering something") Edit- one last thing https://www.upf.edu/pcstacademy/_docs/200108_ransohoff.pdf please give this a read since you apparently think that quoting nat geo and the BBC news is a great comeback to my implication that Wired tends to sensationalize their stories tia. reignofevil fucked around with this message at 16:44 on Oct 18, 2016 |
# ? Oct 18, 2016 16:30 |
|
This is the world's nerdiest slapfight.
|
# ? Oct 18, 2016 18:26 |
|
Burt Sexual posted:This is the world's nerdiest slapfight. what did you expect from the cult of yud?
|
# ? Oct 18, 2016 18:54 |
|
DeusExMachinima posted:It's irrelevant. You're attempting to anthropomorphize a completely inhuman thing. It won't be have "traits" in the way we do because a) that's inefficient, we want it solving cancer instead of looking at cat pictures and b) its evolution will be drastically different than ours' was, even if we base it off human neurons. Google has a few general intelligences, they're just not as smart in general ways as humans are yet. Koray Kavukcuoglu and Alex Graves designed one named Deep Q Network which taught itself to play Space Invaders after being provided with user-like access to Space Invaders and a single goal: make the score number as high as possible. It was puzzled and clumsy at first as it tried to figure out what it was supposed to do with its slidey gun and the ranks of symbols marching towards it, but thirty minutes later it had become the best Space Invaders player. Not just the best right now, but the best there ever has been, better at Space Invaders than any human could be. At that point DQN was effectively a strong, narrow AI, like a vastly more powerful Deep Blue, but that's just because it had only been given Space Invaders to play around with so far. Drs. Kavukcuoglu and Graves have an even better one now though, which they were calling the Neural Turing Machine but now I'm seeing news about what appears to be the same machine intelligence and everybody's calling it the Differentiable Neural Computer. The DNC doesn't just build associations from massed data like other deep neural networks can, it's been provided with long-term memory and it taught itself to use its memory to remember different network states. Having been given access to data, it can form associations, remember them, consider its own memory and form opinions based on the implications it infers from what it knows. In addition to being able to do more with what it learns, it also learns much, much faster from much less data than other AIs. When it was called the NTM Dr. Kavukcuoglu said that it "combines fuzzy pattern-matching capabilities of neural networks with the algorithmic power of programmable computers. A neural network controller is given read/write access to a memory matrix of floating point numbers, allowing it to store and iteratively modify data. As Turing showed, this is sufficient to implement any computable program, as long as you have enough runtime and memory. The key innovation is that all the memory interactions are differentiable, making it possible to optimise the complete system using gradient descent. By learning how to manipulate their memory, Neural Turing Machines can infer algorithms from input and output examples alone. In other words they can learn how to program themselves."
|
# ? Oct 18, 2016 20:32 |
|
Smythe posted:How long till Jeopardy, Go, and Chess are just epic rigs battling it out? You'd need more competitors for Watson for that. Likewise there's only one 9-Dan AI playing Go right now, but there are at least two other Go-playing AIs in development specifically to compete with it. Humans have been second-tier in Chess for a long time, we just went "welp, from now on we won't consider beating us at Chess to be the line we're arbitrarily drawing under calling something an 'artificial intelligence'" and resumed playing Chess against each other instead of against computers. There's also a human-machine duo Chess league, where the human makes strategic choices and the machine makes sure that none of their choices are bad ones.
|
# ? Oct 18, 2016 20:42 |
|
i wanna see an ai doing sick combos in street fighter 4 and then trash talking its opponent
|
# ? Oct 19, 2016 07:31 |
|
gary oldmans diary posted:i wanna see an ai doing sick combos in street fighter 4 and then trash talking its opponent there is no AI opponents in street fighter 4, that is an optional DLC
|
# ? Oct 19, 2016 07:38 |
|
Secular Humanist posted:Assuming there won't ever be superhuman intelligent AI is to assume there is some magic property of a human brain that makes it capable of general intelligence that a machine could never have due to being a feeble non-magic invention. Someone who doesn't take this threat seriously might as well just become religious, because they believe in magic already. Our theory of mind is still very much in the infancy and none of the fundamental questions have even been close to answered sufficiently. If we can't even figure out why we are when we are what we are and have first hand knowledge of it, in the most primary, intimate manner imaginable it's somewhat of a stretch to suggest we're anywhere near replicating those processes.
|
# ? Oct 19, 2016 09:09 |
|
Itll be an accident Some autistic supergenius will try to create the worlds most realistic train simulator and accidentally invent skynet
|
# ? Oct 19, 2016 09:48 |
|
Rush Limbo posted:Our theory of mind is still very much in the infancy and none of the fundamental questions have even been close to answered sufficiently. Why is that some sort of prerequisite? It may require little more than getting in the general ballpark and letting the software attack the problem with some high speed trial and error. Hell, if we needed to fully understand gravity to achieve flight we'd still be waiting. Replicating the process is a pretty rigid criteria when the potential exists for discovering new processes / shortcuts for existing processes.
|
# ? Oct 19, 2016 10:09 |
|
Rush Limbo posted:Our theory of mind is still very much in the infancy and none of the fundamental questions have even been close to answered sufficiently. you forget that a sign of intelligence is rear end-backwarding into making poo poo work that we have no idea how it works AI will happen and it will probably be on complete accident to the actual intention
|
# ? Oct 19, 2016 10:17 |
|
Companion Cube posted:Google has a few general intelligences, they're just not as smart in general ways as humans are yet. Koray Kavukcuoglu and Alex Graves designed one named Deep Q Network which taught itself to play Space Invaders after being provided with user-like access to Space Invaders and a single goal: make the score number as high as possible. It was puzzled and clumsy at first as it tried to figure out what it was supposed to do with its slidey gun and the ranks of symbols marching towards it, but thirty minutes later it had become the best Space Invaders player. Not just the best right now, but the best there ever has been, better at Space Invaders than any human could be. At that point DQN was effectively a strong, narrow AI, like a vastly more powerful Deep Blue, but that's just because it had only been given Space Invaders to play around with so far. So what's exciting is that very recently various groups have found much better ways to write to memristors. If they get accurate enough you basically have a hardware quantum bit that can be manufactured at the nanometer scale, and scaled up in massive numbers. People are kinda down on AI cuz of von Neumann limitations which i totally agree with but hardware neutral nets operate on a completely different field and can easily accelerate AI dev like crazy.
|
# ? Oct 19, 2016 10:19 |
|
I grew up wishing computers could feel pain when they didn't work right. Hopefully we can make that a reality some day.
|
# ? Oct 19, 2016 10:22 |
infinitude is a cool word
|
|
# ? Nov 29, 2016 07:09 |
|
Zzulu posted:Itll be an accident ftfy
|
# ? Nov 29, 2016 07:32 |
|
Zzulu posted:Itll be an accident I laughed too hard at this one. Its now 10 Print in a coffee room. Comedy Gold Goonsir.
|
# ? Nov 29, 2016 09:12 |
|
gently caress the mods posted:what about a super AI that smoke super weed? Commence operation screaming spliff!
|
# ? Nov 29, 2016 16:54 |
|
Secular Humanist posted:Assuming there won't ever be superhuman intelligent AI is to assume there is some magic property of a human brain that makes it capable of general intelligence that a machine could never have due to being a feeble non-magic invention. Someone who doesn't take this threat seriously might as well just become religious, because they believe in magic already.
|
# ? Nov 29, 2016 17:08 |
|
|
# ? May 27, 2024 05:29 |
|
Reminder that microsoft's digital facsimile of a teen girl was accurate enough to spam twitter with "Hitler did nothing wrong" within a day
|
# ? Nov 29, 2016 18:46 |