AlphaGo's next move

> Also, people used to be considered geniuses for knowing a lot of things.

This is still true, and in the eagerness to dismiss "memorization" as a thing of the past you overlook the obvious. For example, anything you care to know about, say, C++ programming or quantum field theory is available to you on the internet. But does that mean you can write a C++ program as if you had already learned it? What if you want to write a C++ program and you have to look up everything? You will do a very poor job if at all, and you will take a lot of time.

So yeah, until looking up stuff in the internet is as quick as effective as looking stuff up in your brain (the quick may happen but the effective I don't think so), then it still is a very worthy skill.

Memory still matters when it matters. For example, when learning a new language. In general, good real-time performance is going to require practice and memorization. It's the trivial stuff you rarely need that's less valued.

There are some other things that I think are less valued these days:

- Informed speculation counts for less when you could do a search instead. Maybe a good thing?

- Cleverness counts less when there are memes everywhere. Jokes are ever more cheap and disposable.

On the other hand, good judgement of what you find still counts.

> Nowadays information is just a Google search away, so knowing a lot doesn't really mean as much as it used to.

You give Google search too much credit. What is a click away is still largely superficial information on any topic and the popular (or specific data set) is often extremely biased or downright incorrect.

Google made fact recollection less important than before. However, Google itself prefers people who do well on whiteboard interview, which could be solved perfectly by some Google searches! Memorization of historical facts is one thing, but training your memory for fast retrieval of facts and associations is still highly valued. Therefore it is still important to train your memory to the extreme. You'll do something better if you do it more than once.

How do you synthesize knowledge that you haven't memorized? How do you know in advance that you can usefully synthesize A and B (and that therefore you should look up B, which you don't know)?

>Nowadays information is just a Google search away

Knowing what is credible and what is not on the internet is a skill in itself. If you don't have that skill you'd likely be telling people all about how Bush did 9/11 or how hillary killed a DNC employee.

AlphaGo is better than humans at seeing patterns and making inferences based on simulations of possibilities, within a limited and perfect information domain. Several professional Go players have described AlphaGo's play as 'creative' and 'beautiful'.

What if technology similar to AlphaGo can be generalized to domains with imperfect information (Libratus from CMU recently beat top Poker players. DeepStack which is NN-based achieved a similar feat.) and to other domains (DeepMind is working on Starcraft.)? What are our remaining competitive advantages against machines?

What are future geniuses supposed to be like or to do (assuming your presupposition)?

I would argue that memory is even more important now since it allows you to think of new things and connect things in your head. Even if we had a neural chip that gave us all of Wikipedia, it would not be the same as intentionally studying something and memorizing it.

While it's a lot easier to get superficial knowledge, the ability to do something deeply is an incredibly valuable skill that is powered by memory. Heck even Euler, one of the greatest mathemeticians of all time had a phenomenal memory and could recite any verse from the Aeneid at will - I don't doubt that memory was a critical component of his success.

>> Basically the "memory" part of human brains have become commoditized without us even realizing.

So, explain this to me like I'm 5. If Google search means you don't need to remember any knowledge any more, why is it that possession of an English-Greek dictionary does not render one capable of speaking both languages fluently?

You can look up all the words you want. Assume you have a grammar of each language at hand, also. Do you think you would be able to speak fulently or understand a fluent speaker of a language you don't know?

There are several activities right now that people keep doing despite machines being better suited: martial arts (and more in point, HEMA[1]), handicrafts, several Olympic sports. Why would anyone attempt to run a marathon under two hours, when any car can do it in 30 minutes?

I think the answer lies less in "I will be the absolute best", and more along the lines of "I will do it better than anyone before me". And sometimes, even "I will do my best" is an excellent reason for doing things.

I don't think Go players were in it due to a need for expertise that machines could not fulfill until now. And if people nowadays keep practicing with swords several centuries after the invention of firearms, Go players will do just fine.

[1] https://en.wikipedia.org/wiki/Historical_European_martial_ar...

The computer supremacy certainly took something out of chess, but it also added a lot : it makes checking tactics trivial even for amateurs, it enormously enriched opening theory, and it made learning much easier for players of all levels (even Grand Masters).

> our world is close to being completely upheaved by intelligent machines, in all areas of intellectual pursuit

I could see myself losing the passion for software engineering and design if an AI can do it better. That would have to be a general AI, and hopefully another couple of decades away.

I wonder if I could enjoy movies or books written by an AI. Scary to think about the psychological manipulation it would be capable of, especially if it lives inside a Google or Facebook datacenter.

Speak for yourself. I paint because I like to paint. It doesn't matter to me if the spectrum of painting technique or theory has been fully explored.

You are expressing a highly misguided viewpoint about what it means to be in touch with beauty.

Nobody has the capacity to build the Pyramids or the Taj Mahal today. So what? Architects haven't shut shop.

I'm a competitive scrabble player, and I suspect it will be less than a decade (and possibly well less) before we have an AI that comfortably outclasses the best human players. (It might never be "unbeatable" due to the luck factor, but it could well, e.g. win 29 games out of every 30). However, it won't affect my enjoyment of the game in any way - it's already clear that board games in general are solvable by an "explore lots of moves with many levels of lookahead" strategy; the thrill lies in the fact that humans playing across a board clearly have to use different techniques to achieve the same result, and you're competing against other humans to see who can do it best.

Indeed, much of the current excitement around AI playing programs lies in the fact that computers are too slow to do the exhaustive brute force tree search either; they need a lot of very clever valuation and pruning techniques to explore more of the tree in less time. It's just a different form of cleverness than what humans do, and there is a lot of feedback between the two communities, with human players helping programmers identify good heuristics, and then computer players uncovering new possibilities for humans to incorporate into their play.

Fish are way better at swimming than humans, it's still great fun. I don't get it.

I played at the local go club this afternoon. Nobody has a problem with AlphaGo and the mystique of the game. Everybody knew go was about to be conquered, even the younger pros. We didn't expect it to happen so quickly but IMHO the sooner the better: we're going to have great tools to analyze games and maybe get better at the game. We're playing against people anyway and it's a great game.

This reminds me of Iain Banks' Culture series, where society is ruled by super-intelligent AIs and drones with intelligence roughly equivalent to people. It isn't really a theme of the stories, more of an undercurrent, but there is an element of humans trying to figure out what their place and purpose is in a society where machines can do literally everything much better than they can.

Bit hyperbolic. Go at it's heart is a pretty basic game, just tricky combinatorics. The fact that you can do unsupervised learning simplifies it even more. I think there are much better true indicators, such as better translation, speech to text, auto driving, etc.

I, too, remember this. I'm very close to the chess engine scene, and while people (such as replies to this thread) have tried to argue that engine development has not taken away from chess, I'm afraid their misguided.

Anybody paying attention to top level chess knows that it has turned into mind bogglingly boring forced draw lines due to engine analysis. I've seen super GMs argue that analysis is so deep now that "e4" openings for white may be unplayable due to how rapidly black can equalize. Romantic play has been all but squeezed out of chess, which is why there has been renewed demand for blitz chess and murmurs that it may one day supplant standard chess as the main World Championship.

The replies miss out that improvements in, say, automobile speed don't impact how marathon runners run their races. But improvements in AI due modify how cognitive (rather than physical) games are played. There is a trade off that is unavoidable. I imagine Go will now become over analyzed just like chess where the top players memorize spreadsheets full of opening moves.

Fortunately, there are variants of chess like Zhouse, which still appear too complex for engine to dominate any position (although they will defeat any human), and for which nearly every move is romantic still.

They don't fully understand the moves but on the other hand, the live commentary on the games suggests it's not completely mysterious. Good moves still tend to look good to them, in retrospect at least.

The games are apparently very interesting to study.

They don't understand the move probably the AI has a memory depth that is way beyond any human, some of those moves are the best possible look aheads for that situation

They did announce that they would release a teaching tool which will show AlphaGo's analysis of Go positions, as well as the paper explaining how to build your own.

Not only do you have the principle and the formula behind it, but also a little physics simulator tool! At this point, it is hard to complain.

"We plan to publish one final academic paper later this year that will detail the extensive set of improvements we made to the algorithms’ efficiency and potential to be generalised across a broader set of problems."

Should be enough, no?

Though if they released the code Tencent would incorporate it in their rival so I can see the argument for delaying a bit.

They're also publishing later in the year with all the details.

Do you know if there's video or audio of Peter's talk online? I just did a bit of a search but couldn't find it.

The microphones used in the post match discussion featured the name “Google” quite prominently. I imagine that was a negotiated detail.

You need to get out of whatever bubble you're living in. Human civilization is doing fine. Machine learning will do some stuff but not major changes at the civilization level in the next 40 years.

Walk into any real world business today. There's a huge amount of need for humans, because fundamentally business is about trust not productivity.

There are ways to greatly improve chances that AI will be beneficial to humanity rather than otherwise.

Check out: UC Berkeley's Center for Human-Compatible AI, led by Prof. Stuart Russell, a co-author of the field's standard textbook. [1] He just gave a TED talk on the issue [2].

Several other noted researchers in AI are working on the issue as well.

For a short primer: https://futureoflife.org/background/aimyths/

[1] http://www.openphilanthropy.org/focus/global-catastrophic-ri...

[2] https://www.ted.com/talks/stuart_russell_how_ai_might_make_u...

> AI systems will surpass me within 2-4 years

Oh lord. 4 years ago was 2013. Was there such a jump from 2013 to today that makes you or anyone claim that within just 24 months machines will actually program better than developers, when there isn't as much as a proof of concept of that yet? Bar a major and unexpected breakthrough you can sleep assuredly that no machine will take your job just yet.

I agree that becoming an AI researcher seems daunting, since the field is moving so fast. However, so far we haven't seen any useful applications of machine learning to ordinary computer programming, so this hopelessness seems rather premature.

(Perhaps applying machine learning to code review might be useful, to spot bugs? The problem would be getting good data to train it.)

Is there really anyone credibly suggesting software developers will be surpassed by AI within a few years? Writing arbitrary software seems dramatically more complex than what "AI" like systems are capable of today.

Even 10 years seems impossibly soon.