Skeptics have said that achieving this milestone would not happen within this decade, our lifetimes or even ever. It happened yesterday.
It took Lee Sedol many decades of his life to train to achieve this level. And his ability to pass on his skills is limited. Now that a computer has achieved this level, the state of the neural network behind it can be serialized and ran into an unlimited number of computers and have millions of systems that are more proficient at Go than the best player in the world.
People have said that achieving the cognitive level of the human brain requires to match its computational power. But if you take out all the parasympathetic and motor boilerplate, what is actually left for mental tasks is much less from that, most of that power is not even recruited for higher level mental tasks. That lowers the bar for strong AI.
Then, strong AI can be immortal, and never physically deteriorate from aging. Strong AI can multiply infinitely and communicate at a rate that would be equivalent to writing millions of books in a second. It could transfer all its knowledge in seconds. It can also recursively improve itself. This advantage will lower the bar for strong AI even more.
It's worth remembering that AlphaGo was partially, initially trained on the previous game records of professional players, to form the basis of its policy network and so on. AlphaGo's victory over Lee Sedol is impressive, but it does rest on the combined total experience and long study of humans. That's why it's even more exciting to me to hear that one of the DeepMind team's next efforts will be to see how AlphaGo plays when it starts learning "from scratch".
I agree with the excitement. A possible outcome is that the AlphaGo trained "from scratch" will not be as strong--but another possibility is that by eschewing any bias for human tradition, perhaps it will come up with even better strategies that no one has thought of before.
This was my argument against the StarCraft claim. If it wins, it will likely have just imitated human pros' thousands to millions of games like a smarter version of old chatterbots.
Whereas, good human players learned as they went with far fewer matches, trial and error, planning, and learning from pto games sometimes. We can even go from Age of Empires to Starcraft with little prrparation and still do OK.
Can anyone elaborate on how they would teach it from scratch? Would this mean only giving it the ability to play valid moves (and against itself) and give it access to the final score?
On the other hand i would like to read an assessment of the significance of this advance. While AI may have perfect applicability to these perfect information games, we may still be decades from useful real world generic applications.
Exactly so. While this is a major advance, I roll my eyes at the sea of comments hailing the arrival of a superior intelligence.
When the AI is cognizant of the fact that Go is a game, and knows what a game is, and perceives that the game is taking place in a larger reality, where there are other things happening while the game is going on....then I'll be impressed.
Or maybe not. The history of AI is one of humans always moving the goalposts when AI advances.
Chess? just a computationally simple game.
Driving? Well it's just physics.
Go? now that requires intelligence oh wait, just some deep neural savant thing, not real AI
Well, I think what I'm describing (general awareness) is a little different from your examples there, but I acknowledge my choice of words implied I'm not impressed, which I completely am because this is obviously utterly amazing. It's just not AS big a deal as a lot of people are making it out to be with proclamations of apocalypse.
Huh? IBM Watson was able to answer trivia questions based on some impressive NLP (that I'd like to see more widely available) and a huge database of factual information. Oh, and it was able to beat humans in buzzing in. Watson was impressive, no doubt about it. But it certainly didn't have any meta knowledge of its place in the world or in a contest.
What I mean is that the IBM Watson deals effectively deals with semantic information.
You can make some analogies between sparse matrix data representations and what happens in the mammalian neocortex. It is of course not self conscious, but it effectively deals with semantic information and relationships among it.
AlphaGo's algorithms will generalize easily to other perfect information games, including ones that include randomness like Backgammon. Games with imperfect information will take some more advances, but AlphaGo just opened up a lot of interesting potential paths, so there's a decent chance someone pursuing one of those will find something quickly.
The sheer amount of computing power that AlphaGo needs will be a problem for games as complicated as Go for people who aren't Google. There will probably be more activity in simpler games with interesting properties.
Non-game applications will require a fully formalized set of rules. The fact that formalizing what you want is hard is the reason we need programmers in the first place, so nothing fundamentally different there.
Even knowing what rules one is using isn't always clear.
I did some work for a Forex trader. He had devised an algorithm for trading and used it daily for modest gains. He engaged me to automate it for him via the vWorker freelance website [1]. I got the engine going getting the realtime price data etc. But when it came to his system it didn't work. In practice he didn't always act upon the signals he thought he did. We ended up in payment dispute because he insisted that if I had done the work right then the bot would be making money. By this time, I had learned the error of his thinking. We arbitrated on 50% payment - I wasn't too pleased with that but I had learned about trading.
I do feel that given all the hype and speculation about strong AI researchers in Deep Learning should really come forward to contextualize the real impact of this - which I think is not all that great.
AlphaGo has no form of state space discovery. That really prevents all the magic that strong AI fans want. Even though the actual rules of Go are a minor part of the program, the human who put them in was doing something that AlphaGo fundamentally can't do. AlphaGo can't learn what Go is, it can only take an existing understanding of the rules and learn to play very well.
You can expect a lot of advancement in actual games very soon, including ones with randomness, and after a few years probably hidden information games too. Being able to specify real-world problems as formally as a board game is now a more effective skill than it was before (it was good already).
In my opinion a possible path is to make some sort of Symbolic AI (https://en.wikipedia.org/wiki/Symbolic_artificial_intelligen...) system, and put machine learning on top, and even below of it. You could start with a very basic Symbolic AI, easier to understand. Then growth it gradually. The cool thing of this approach, would be that you could inspect the symbolic system to get a glance at how the AI system is thinking. Obviously a purely neuronal system can also do the trick. That is how the human brain works. But it would be great to make a system based on a symbolic AI, because it could be very interesting to observe its inner working. And it would also be possible to interact with it at the symbolic level.
It is good enough now to replace millions of jobs. Transportation, retail, press, trading, etc.
It is also good enough to put it in a combat drone that visually recognizes objectives and people.
This can be all commanded by a small group of people, anonymously, and detached from accountability and without casualties.
Part of the significance is that it was not the main goal for Deep Mind and Hassabis. He's said his main goal is to solve intelligence and it approaching it by trying to reverse engineer the human brain, having done a PhD in neuroscience. Hippocampus next. At this rate I'd be surprised if it took decades to real world generic applications.
Certainly a lot of people ten years ago would have bet against this happening by 2016--especially before Monte Carlo approaches were tried. But I doubt you would have found many takers for betting against it happening in a 100 years (much less it never happening).
Autonomous vehicles are similar. IMO, it would be a fool's bet to bet against them ever happening. But it's reasonable to debate whether they'll be mainstream Level 4 consumer devices in 10 years, 25 years, or 50 years.
Though it's true many people were skeptical about an AI winning at Go played at this level, I think people are mostly skeptical about general AI, not about algorithms that win at playing board games. It's different to be skeptical about specialized AI than about general AI.
General AI could be the biggest ever discovery for humanity. But it will never be if we don't put a hard enough effort. The bad thing, is that economy laws seem to assign a limited effort toward it. In my opinion we should be trying harder. Worth mentioning is OpenAI (https://openai.com/), an effort by Y Combinator, Elon Musk and others. Brilliant initiative.
Overpopulation in terms of running out of space is definitely overplayed. But it's a direct multiplier for our effect on the environment. If the world's population were cut in half, we'd emit about half as much pollution, half as much greenhouse gas, half the land used for farming would be left wild, etc.
Not necessarily...carbon emissions per capita is not constant around the globe. Energy use is skewed towards modern countries, i.e United States. If you wiped out the 300 million people in the US, you would save more on carbon emissions than wiping out 300 million in China.
Technically, it's two neural networks, each calculating reasonably simple functions (an estimate of the value of a board position and an estimate of which moves might be good) combined with a tree search algorithm.
My grandfather was very racist and never changed as long as he lived. I'm glad most people aren't immortal. There's no guarantee a full AI wouldn't be tempted by evil and just become republican (or worse, a VC).
It can also recursively improve itself.
So can people (the more you know, the more you can learn), but most don't. It's important to remember "intelligence" isn't an abstract concept—intelligence is also embodied in personality—and personalities have wishes and goals and desires and loves and hates and that one song they can't get out of their head. A true "strong AI" will be fully conscious, not just algorithmic function bating.
Good luck telling a mildly strong godform to stop tripping on youtube videos and instead solve the global economic stability equation over lunch.
This advantage will lower the bar for strong AI even more.
That's kinda foofy conjecture. Being good at rectangular grid outcomes isn't necessarily a step in any direction towards a hands-off tax evaluating robot.
It feels really really good to talk about how AI will be a hundred billion trillion times smarter than the combined brainpower of all humans that have ever lived, but it feels good in the same way thinking dead people live again after they die feels good—it triggers that warm wishful thinking parietal lobe that removes a bit of reason for the sake of an overarching calmness.
Enthusiasm is great, but tempering with real expectations and less technopriesthood is better.
You can recruit neurons within your own brain, not plug the equivalent of 4 brains into your brain and increase your cognitive abilities.
Now, while many people die and their ideas die with them, there are built in aspects of our brain that are hardwired. People eat animals because they are tasty, people follow the life of Kim Kardashian and waste money buying a rolex and wear fur just because the stupid way our social functions are hardwired in our brain.
In the grand scheme of things, what is the point of humans repeating billion of times the same cycle?: born, grow, learn, achieve some things, have some fun, and die. It is awesome a few times. But it is worth to repeat and repeat that for ever? Not if there is a much better option: The singularity. It has the potential to enhance the human race in formidable ways. Yes,there is the risk that the AI may chose to destroy us. But in my opinion is a bet that we should take. Just try our best to not be destroyed in the way. But we shouldn't leave this opportunity unexplored. It would be the biggest achievement of humanity.
It is NOT the same cycle.
Every life is different.
So what the people learned 500 years ago made sense for them - but they would face a hard time now.
Our knowledge of daily living will be mostly obsolete in 200 (or much less) years, too.
Fresh minds don't have the burden of outdated ideas... quoting Max Planck: "a new scientific truth does not triumph by convincing its opponents and making them see the light, but rather because its opponents eventually die, and a new generation grows up that is familiar with it."
You are assigning value to "us" beyond what might actually be there. What if our only evolutionary value is to give rise to machine intelligence [that will go on to explore the universe] and die off?
We are primitive forms of life after all and are at the point where we should be able to say, with certainty, that machine intelligence will be objectively superior and something that we should feel OBLIGATED to give rise to, regardless of consequences.
> What if our only evolutionary value is to give rise to machine intelligence
There is no prescribed script about what our evolutionary value should be. We will be whatever we ourselves make of us. Unless something unexpected happens -more advanced aliens intervening with our specie for example- Our human instinct is to grow, explore, discover, compete, share as much as we can. AI should be seen as a tool to our goals. Not as an end by itself. There is of course the possibility that we may loose control to the AI. That is not our desired goal, but it could happen. In that case our future will not be our decision. We will be at the mercy of whatever the AI choose for us. That will depend on what kind of AI we built, and how the rampant AI chooses to modify itself. But we will strive to reach our human goals. Because that is the human nature.
Just because you or me see no prescribed script, doesn't mean it's not there. You make a philosophical point here, but my point is more than that.
"We will be whatever we ourselves make of us" sounds meaningless to me. The universe obeys certain physical laws, evolution is directed and follows a specific pattern. We have up to this point been bound by both. Occam's razor says this will continue to be the case.
I think we are already at the crossroads where we can more than speculate about the function of machine intelligence in evolutionary terms, but hopefully sooner rather than later we will have actual _evidence_ that we can look at and use to refine our assumptions.
It took Lee Sedol many decades of his life to train to achieve this level. And his ability to pass on his skills is limited. Now that a computer has achieved this level, the state of the neural network behind it can be serialized and ran into an unlimited number of computers and have millions of systems that are more proficient at Go than the best player in the world.
People have said that achieving the cognitive level of the human brain requires to match its computational power. But if you take out all the parasympathetic and motor boilerplate, what is actually left for mental tasks is much less from that, most of that power is not even recruited for higher level mental tasks. That lowers the bar for strong AI.
Then, strong AI can be immortal, and never physically deteriorate from aging. Strong AI can multiply infinitely and communicate at a rate that would be equivalent to writing millions of books in a second. It could transfer all its knowledge in seconds. It can also recursively improve itself. This advantage will lower the bar for strong AI even more.