Slashdot Mirror
DeepMind's Go-Playing AI Doesn't Need Human Help To Beat Us Anymore (theverge.com)
An anonymous reader quotes a report from The Verge: Google's AI subsidiary DeepMind has unveiled the latest version of its Go-playing software, AlphaGo Zero. The new program is a significantly better player than the version that beat the game's world champion earlier this year, but, more importantly, it's also entirely self-taught. DeepMind says this means the company is one step closer to creating general purpose algorithms that can intelligently tackle some of the hardest problems in science, from designing new drugs to more accurately modeling the effects of climate change. The original AlphaGo demonstrated superhuman Go-playing ability, but needed the expertise of human players to get there. Namely, it used a dataset of more than 100,000 Go games as a starting point for its own knowledge. AlphaGo Zero, by comparison, has only been programmed with the basic rules of Go. Everything else it learned from scratch. As described in a paper published in Nature today, Zero developed its Go skills by competing against itself. It started with random moves on the board, but every time it won, Zero updated its own system, and played itself again. And again. Millions of times over. After three days of self-play, Zero was strong enough to defeat the version of itself that beat 18-time world champion Lee Se-dol, winning handily -- 100 games to nil. After 40 days, it had a 90 percent win rate against the most advanced version of the original AlphaGo software. DeepMind says this makes it arguably the strongest Go player in history.
I, for one, welcome our new Go-playing robotic overlords.
General Relativity: Space-time tells matter where to go; Matter tells space-time what shape to be.
Whoop-de-freakin-doo
>> After three days of self-play, Zero was strong enough to defeat >> the version of itself that beat 18-time world champion Lee Se-dol, winning handily -- 100 games to nil. >> After 40 days, it had a 90 percent win rate against the most advanced version of the original AlphaGo software. So after 3 days, it had 100% win rate, after 40 days it had only 90% win rate.
... It can deal with hidden information. The most advanced approaches based on the same algorithm as used by AlphaGo (MCTS) are still not practical to use outside of a few toy examples.
Not impressed, doesn't prove anything, and why should anyone even care?
... It can deal with hidden information.
So you mean something like poker? AI beats pros at Texas-Hold'em.
Sorry, I doubt there is any chance that a neural network can be used in a meaningfull way in drug design or climate modeling.
Cost free eBook I read (by iBook/Kobo/Amazon/ObookO/Gutenberg etc.): "The Green Odyssey" by Philip Jose Farmer.
having a computer sequentially learn game moves is hardly A.I. -- Sounds like a grab for new investors more than any sort of groundbreaking research. Grow me some better weed as an intro to pharmacology and maybe I'll warm up to it...
They could have mined tons of bitcoins instead with that computing power.
I foresee some well-deserved -1 troll mods coming your way in the near future.
Poker is a very bad application of 'AI'. Poker is not a game of human strength, but rather a game of exploiting human weakness. A poker playing computer can be superior to a human simply by virtue of its programmers choosing not to program it to have any tells when it bluffs. A computer raising when it has a four of a kind is indistinguishable from a computer raising when it has high card 10.
This is one of the problems in the AI world. They should have targeted playing as well as the average human. There is minimal benefit in being the absolute best Go player that could exist. Difficult and complicated intelligences have to be far more general than that. There is tremendous value in developing an intelligence comparable to normal humans without need for it to be capable of defeating humans who've dedicated their lives to a single obsession at their own game.
Read the book The Promethean by Owen Stanley whereby a tech billionaire decided to build the perfect AI as a gift to humanity. Social Justice Warrior intervene and taught him the concept of social justice and the AI forked itself to infinity.
They want to use a new more powerful computer to answer the question of life, the universe, and everything?
Pft it's been done. Stupid 3 dimensional creatures.
Solution: eliminate humans
What side do you want?
If it were all about tells there would be no online poker. Poker IS about reading other players but you can read a player from their play.
Next up, calling people out for begging the question on that whole gravity "theory" by asking them guess where an Apple tossed up in the air will land when it comes down.
Such as Global Warming[TM]?
Oh you're denialist too!
Kind of obvious in hindsignt given the nature of your sig.
I can already see the next generation of Illiberalas taunting skeptics to "win a game of Go against this AI before questioning Climate Change".
Welp, if you've run out of things to be outraged about, just make some up and get outraged about those! Darn librulhs.
SJW n. One who posts facts.
I'm the parent AC.
I know about the poker AI, but the thing is that solution is very non-general.
AlphaGo's inner workings are based on the MCTS algorithm (which is a fairly general algorithm that can be applied to many games). It would be more interesting to see MCTS properly developed to deal with hidden information. There are improvements (ISMCTS, MO-ISMCTS and MT-ISMCTS) but they are all extremely limited in one way or another.
A general solution would be awesome.
That's the skill in writing AI for computer games. You don't just want to have the game always at skill level infinity, you want to have the whole range from skill level 0 to infinity. Skill level 0 is simply making moves at random. Sometimes, it's up to looking ahead a dozen moves, other times, it's not constantly making aggressive moves.
Vintage computer adverts: http://www.vintageadbrowser.com/computers-and-software-ads
I've decided that this accomplishment -- a dizzying milestone in artificial intelligence that not long ago was though impossible or at least decades away -- is actually meaningless and doesn't prove anything and they should clearly have been working on some other problem. I have no idea how their system works, but I'm confident that their approach is just "brute force" (or something, I clearly have no idea what even that means) and won't generalize to any "real" problem solving (with my definition of "real problem" subject to change without notice).
I will only admit that any progress has been made towards artificial intelligence when computers perform exactly equivalent to humans in all tasks with no human intervention. I mean, I won't really, because I have weird quasi-spiritual hangups about believing computers can be intelligent, but that's where I'm putting the goal posts for now. Digital computers can't think, but I can because reasons. Free will or quantum mechanics or something else that I haven't thought about at all, probably.
Also, cotton gins and blacksmiths, therefore computers will never take our jobs. Amen.
Let's not stir that bag of worms...
The way to pass Turing test would be to convert it into creationism as only human would believe it?
Oh wait, this should be easy, it was created...
4wdloop
"programmers choosing not to program it to have any tells "
Actually, creating and exploiting fake tells is exactly how an AI poker can win.
1. A simple computer has 4 of a kind and always goes All-In.
2. An advanced computer has 4 of a kind and sometimes calls and sometimes goes all-in, just to mix it up.
3. A good AI computer has 4 of a kind and has carefully built a fake trail of typically acting a certain way with a good hand, but this time decides to simply bet 18%, which is calculated based on what it has learned of the human players typical traits regarding calling or folding in order to optimize the take.
They should have developed it to *teach* someone to play go, and correctly assess and match their skill level no matter what that may be.
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Can you make the "game" be the game of learning? I can imagine the dataset would be the rules of many different games and the solutions would be networks that learn those games with solution quality based on some balance of leanness and efficacy of the networks. You'd then let it loose teaching itself how to best teach networks. Hmmm.
I don't have a meaningful understanding or opinion of what "AI" is, but whatever it is, is at least a momentary respite from pretending that being able to watch TV on a mobile telephone is the greatest human achievement since the discovery thing since penicillin.
C'MON PEOPLE! These are the "innovations"!
just 100'000 Go games played by humans.
That's what, say 2 games/day makes 160 people playing Go daily for a year.
"No human help".
Sure. Got it.
Forgive my ignorance but is it defined as being able to adjust is own programming code to me efficiently reach a predefined goal?
Dictionary definition is "the capability of a machine to imitate intelligent human behavior" but I just don't think as though being able to utilize enormous computational power to work through permutations of a board game should qualify.
Don't be too hard on him, manufacturing this sort of outrage is a time-honored tradition and puts bread on a lot of tables.
Outrage doesn't just grow on trees you know. Without his efforts the outrage deficit would lead to a world awash with harmony.
A general solution would be awesome.
General solutions require Strong AI, which, for now, is science fiction.
Complaining that this self-learning Go program isn't general purpose is sort of like complaining that a better electric car battery won't help your Tesla reach Warp 9.
This is incremental progress, not a revolution, but it is still an interesting advance.
The essence of intelligence is that it enables one to predict the outcome of a unique situation based upon an understanding of its essential elements.
Starting with only the rules of Go, Zero explored a variety of combinations, learning that some were more likely to give a satisfactory result. It developed a sense of what types of moves are best. Thus, without playing or studying an infinite number of games it could know the type of move that should be best in each unique situation.
Theoretically, a vast intelligence, given only the facts of the Big Bang, could anticipate most of the resulting evolution of our universe. Zero has taken the first small step.
...omphaloskepsis often...
Boards may be 21x21 but may be other sizes as well.
She was like chocolate when she drank... semi-sweet at first and then increasingly bitter.
Zero when later pressed for a comment, responded:
"A strange game.The only winning move is not to play. How about a nice game of chess?"
It still can't beat me at tic-tac-toe.
At least I can still beat alpha go at kick boxing.
This doesn't show we are winning at creating AI. It simply shows that the game of Go is more tractable than we previously thought. Claims about the number of positions in Go being vastly greater than the number of atoms in the universe (something like the number of atoms squared) completely miss the point: this is a straw man argument for why algorithms weren't good at Go until recently, since (obviously) humans are not searching the entire space of all possible board positions either. It stands to reason that once a sufficiently flexible fuzzy hierarchical pattern matching algorithm were produced, it would be able to play Go much better than a human.
But if you play against yourself... You *will* win everytime.
And that son is how Skynet was born. Damn humans never learn anything.
"They should have developed it to *teach* someone to play go, and correctly assess and match their skill level no matter what that may be."
True, outside of the Go playing countries in the Far East, we could use 1000 or more copies of AlphaGo Zero to train us how to play better. AlphaGo Zero would be a tireless teacher, in addition to having the best answers to our moves. We have yet to see how AlphaGo Zero would play in a handicap game.
I have programmed this type of learning algorithm in the past. About 30 years ago when computers were about 30000 times slower than now.
Anyway, you can have the program play itself for a while and it becomes quite good. But you won't know how it will perform against a human unless you try. It might be very good against those moves that the computer player will come up with, but very bad against moves thought up by a human.
30 years ago I tackled a simpler game than go.
no human help? oh come on.... no SW or AI can claim this.
who learned AlphaGo Zero to learn?
who programmed the basic Go rules?
I'm impressed though, they found a smart way to build a huge
game database with proper evaluation/conclusion for each turn.
This assumes that there is a linear level of skills and that winning against an AI that wins against a human means that you'd win against that human.
That isn't so. There are chess players who specialize in combatting AI chess. They fare better against AIs than world champions of chess but would not have a chance to beat the latter.
There is minimal benefit in being the absolute best Go player that could exist.
This is a research project. People want to know how far they can push it.
Dumbing it down to provide a useful challenge for humans is easy.
I'm not talking about dumbing it down to provide a useful challenge for humans, I'm talking about having it play Go, compose music, and write poetry. Using AI for anything but a toy application like this will require AI that master many poorly related skills and combine those skills to execute a complex task. If you ever want an AI to write a useful fiction crime novel that AI will not only need to be able to compose English and be creative it will need to be able to research the various topics involved come to a high level understanding of these various unrelated subjects and hypothetically apply them in some unique, plausible, and unsurprising way. You are never going to get there if you are with algorithms designed to waste effort trying to be the best at everything, you need algorithms that look for "just as much as I need" otherwise your book writing AI would spend a lifetime on each of the dozens of different things it needs to research to write the book.
Let me put this another way since so many seem to think I'm talking about dumbing it down enough to be fun to play. We aren't talking about a game bot we are talking about an AI development project.
Humans are a ridiculously powerful AI's, much more powerful than AlphaGo, so why can it beat us? It isn't because AlphaGo never sleeps, in fact, AlphaGo does sleep it just spreads it among more frequent and smaller time scales. We can't beat it for the same reason we can beat AlphaGo at everything but playing go. From the moment of birth we'd need to be hard-wired into a minimal set of controls with a direct brain interface with all unrelated sensory input shut off, feeding and waste removal processes automatically handled. That is the human equivalent of the AI they are building. Except they are pre-programming it with the rules and objective even though Go has an extremely simple play mechanism.
This approach is only going to get you so far, if it weren't humans would be better go players. Human limitations exist for a reason we don't get bored with performing a single task over and over forever because we suck, we get bored as a mechanism to shift and spread mental resources among more skills which provide alternative insights that can map back as well as be combined to tackle ever more complex super skills. Mastering Go in itself is pretty useless, mastering go alongside a highly realistic warfare simulation on the other hand...
you still don't get that whole "evidence" thing, do you?
as in, tons of evidence of one thing over here , no evidence over there, yet somehow you want to pretend the two possibilities are equally scientifically probable, or even that the one with no evidence is more likely to be true regardless of the complete lack of evidence supporting it.
The guy who said the election was rigged won the presidency with the second-most votes.
The rules of GO has been modified to fit playing software. With classic rules the computer still have no chance since there is no guaranteed win condition, the win might be up for discussion.
So, computers still "suck" at GO but this new game, designed to help them that we call GO, they are excellent at.