Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

AlphaGo's next move

deepmind.com

Post reply on HN
Print view View original

131–140 of 179 posts

Re: AlphaGo's next move

#131

Post by nickpsecurity »

computerex wrote: post #111
nickpsecurity wrote: post #78

Earlier quoted context omitted.

The author isn't hyperbolic. The problems that require what we call common sense continue to defeat all AI's. These board games don't require that. It comes down to pattern recognition, heuristic search, and good hardware. The examples of language and driving are great as the AI's only seem to do well on really narrow situations that require no greater understanding of context. And far as common sense, there's only a…

Modern AI, things like AlphaGo are examples of applied AI. "Common sense" falls within the realm of artificial general intelligence, which is a line of research that's largely abandoned now in favor of applied AI. Modern AI solutions are engineered to solve very very specific problems. You are never going to see attempts to teach "common sense". https://arxiv.org/abs/1705.08807 With that said, the above is what the w…

"The goal is to do things like self driving cars. "

Replacing a human driver takes an AGI at least for the exceptional or new situations. It's why we're including it as a counter instead of supporting point.

Re: AlphaGo's next move

#132

Post by andrepd »

baq wrote: post #125
andrepd wrote: post #102

Earlier quoted context omitted.

Only because you lack information.

Actually at the lowest level there's this whole quantum business that says exactly otherwise.

That is unlikely to affect the microscopic world. The physics of everyday life are deterministic, no need to invoke quantum mechanics.

Re: AlphaGo's next move

#133

Post by andrepd »

jayd16 wrote: post #99
andrepd wrote: post #95

Earlier quoted context omitted.

> 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…

But you've just proven cocktailpeanuts point that its about the ability to use the knowledge, not simply recall it.

You can't use the knowledge well if you have to look everything up every time, that's my point.

Re: AlphaGo's next move

#134

Post by Cyphase »

binarymax wrote: post #23
matt4077 wrote: post #19

Earlier quoted context omitted.

It does feel like there's a difference–maybe Go and Chess are just so much closer to what defines humanity than any physical sports?

Yes and physical prowess was never dominated by humans anyway. We are surrounded by creatures which can fly and swim and sprint far better than any human ever could. But our minds is what separates us from the beasts. We've never been second place in the thought competition.

http://www.slate.com/articles/sports/sports_nut/2012/06/long...

"There's no denying it—our kind started substituting brains for brawn long ago, and it shows: We can't begin to compete with animals when it comes to the raw ingredients of athletic prowess. Yet being the absurdly self-enthralled species we are, we crowd into arenas and stadiums to marvel at our pathetic physical abilities as if they were something special. But there is one exception to our general paltriness: We're the right honorable kings and queens of the planet when it comes to long-distance running."

Re: AlphaGo's next move

#135

Post by cfmcdonald »

cocktailpeanuts wrote: post #88

There was a time not too far back when people used to be considered a "genius" for their ability to memorize things well. Nowadays nobody thinks of them as geniuses. Also, people used to be considered geniuses for knowing a lot of things. Nowadays information is just a Google search away, so knowing a lot doesn't really mean as much as it used to. What matters more nowadays is your ability to learn synthesize the thi…

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)?

Re: AlphaGo's next move

#136

Post by mrob »

zem wrote: post #128
binarymax wrote: post #3

I remember vividly in 1997 when Deep Blue defeated Kasparov, and I was a competitive chess player. The mystique of the game was immediately lost for me, and I never found the passion for the game that I once had. My heart goes out to the sea of Go players now searching for meaning in the game. At the very least we can take this signal as a true indicator that our world is close to being completely upheaved by intelli…

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 "ex…

How good does it have to be to count as "comfortably outclasses"? Wikipedia claims Maven (Scrabble AI) was beating top humans in 2006:

https://en.wikipedia.org/wiki/Maven_%28Scrabble%29

Re: AlphaGo's next move

#137

Post by orthoganol »

probably_wrong wrote: post #12
binarymax wrote: post #3

I remember vividly in 1997 when Deep Blue defeated Kasparov, and I was a competitive chess player. The mystique of the game was immediately lost for me, and I never found the passion for the game that I once had. My heart goes out to the sea of Go players now searching for meaning in the game. At the very least we can take this signal as a true indicator that our world is close to being completely upheaved by intelli…

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 so…

At the same time martial arts in the age of firearms has become relegated to the fringe, and running in the age of automobiles (probably even starting with the age of horses) is also kind of silly to take too seriously, and few people care outside of a couple days every four years, if even.

It's kind of like what happened to painting when the era of photography began.

Machines taking over human disciplines still kills the disciplines, some may just not be aware how this type of death works.

That said, I disagree with most here in that I don't think machines are anywhere close to taking over creative human disciplines that don't follow incredibly specific rules, like boardgames.

Re: AlphaGo's next move

#138

Post by lloydde »

cocktailpeanuts wrote: post #88

There was a time not too far back when people used to be considered a "genius" for their ability to memorize things well. Nowadays nobody thinks of them as geniuses. Also, people used to be considered geniuses for knowing a lot of things. Nowadays information is just a Google search away, so knowing a lot doesn't really mean as much as it used to. What matters more nowadays is your ability to learn synthesize the thi…

> 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.

Re: AlphaGo's next move

#139

Post by zem »

mrob wrote: post #136
zem wrote: post #128

Earlier quoted context omitted.

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 "ex…

How good does it have to be to count as "comfortably outclasses"? Wikipedia claims Maven (Scrabble AI) was beating top humans in 2006: https://en.wikipedia.org/wiki/Maven_%28Scrabble%29

top humans have been improving too; i'd estimate that quackle is stronger than most human players today, but the very best (especially nigel richards! https://fivethirtyeight.com/features/what-makes-nigel-richar...) can hold their own against it. i'd like to see the top-level-human win rate drop below 25% before i'd call us decisively outmatched, and below 10% before it hits the "don't even expect to win a game unless you're an expert" stage.

also, there's an (afaik) unexplored area of scrabble AI where you take into account the fact that some plays are harder to spot than others, and some words less likely to be in your scrabble vocabulary, so if you're playing a weaker opponent you can play risky moves that might cost you against an expert but will win big against a weaker player. unlike in go, this really matters because tournament scrabble uses spread as a tiebreaker; learning how to not just win but win big against weaker players is an important skill in some tournaments.

Re: AlphaGo's next move

#140

Post by lambdadmitry »

resf wrote: post #106
ac29 wrote: post #97

Earlier quoted context omitted.

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.

People said the same thing about Go. That it was far too difficult and that a pro-level AI was 10+ years away. What happened was new mathematical tools and new hardware were developed, and suddenly it was all too possible. It's clear that with our current tools, general AI is out of reach, and new tools must first be developed. But because nobody has any idea what those new tools are , it could happen overnight or ov…

It didn't come from nowhere, the current streak of ML achievements rides on the back of deep learning, which is an elaborate pattern matching at its core. What makes Go "harder" than chess is that it's difficult to estimate how good or bad a particular position is, so we employed a "magical box" of deep learning and learned to estimate how good a particular move is. That's pretty good, but let's not forget

- it took a ton of very hard work

- it's not transferrable to other domains per se ("elaborate pattern matching" can be, but it's not even an AI)

- this has nothing to do with qualia, consciousness or the theory of mind.

Programming is not about elaborate search or pattern matching at the end. It's about formalizing a domain, stripping it from some subset of real-world complexity, and inventing a solution to a problem in that domain. A rift between beating someone in Go and deducing a fact that doubles wouldn't do well in financial calculations is immense.

This sort of over-extrapolation of current trends is surprisingly prevalent in the tech crowd to be honest. It's like folks in mid-20th century who saw both airline and car industry exploding and made a "logical" guess about flying cars being the obvious next step. Guess what, physics doesn't work that way and flying cars are a dumb idea. The current AI craze seems very, very similar to me.

Post reply on HN