World's top Go player loses second match of five against Google AI

[kurros]

Meanwhile, in other news;

A Ferrari beats Usain Bolt in the 100 metre sprint.

If you don't think this is a significant milestone in AI then you should go read about it more, because you are wrong.

I don't like being told I'm wrong. *sniff*


In my opinion a Ferrari actually CAN go faster than Usain Bolt. But you seem to disagree.

I'd be impressed if the Go software was written by a non-human.
Self-sufficient, autonomous AI writing its own AI software script. THAT would send shivers up my spine.

But for the time being, all we have is the same old boring "if/then" method merely sped up faster and faster so it does more 'stuff' in the same time - thereby achieving exactly what we would expect.

Fast doesn't equal 'intelligent'.

That's not how this AI works. You should read up on neural networks and learning algorithms. They are quite fascinating. It beat Se-dol in a qualitatively different way than DeepBlue beat Garry Kasparov at chess 20 years ago.

[Nelvan]

Kurros,

AI Watson has the definition of "colors" memorized in the same way I could memorize, "Un dia voy hacer un hombre muy rico" without knowing what it means even though it is meaningful to others.

[kurros]

Kurros,

AI Watson has the definition of "colors" memorized in the same way I could memorize, "Un dia voy hacer un hombre muy rico" without knowing what it means even though it is meaningful to others.

Yet you wouldn't know if that phrase was the answer to a Jeopardy! question, while Watson would. So it understands color better than that too. What else is "knowing" other than understanding how various data fit together cohesively?

[Lion IRC]

You would 'know' it was the answer to a Jeopardy question if someone told you - programmed you.

Bigger programs (with more Jeopardy answers) running of faster machines - yep, of course they are going to beat the human game show contestant.

[Lion IRC]

I would argue than every one of the human-vs-AI contests are comparable to the Ferrari/Usain Bolt analogy.

[Nelvan]

A computer may know every medicine in existence and know exactly what the medicine cures.  That would be soft AI.  But in order for it to be strong AI, the computer would need to know what "medicine" is, and what "sickness" or "illness" is.  I'm afraid AI currently does not know this.  AI is very good at organizing, calculating, categorizing, but not good at knowing what abstract concepts "mean". 

[kurros]

I would argue than every one of the human-vs-AI contests are comparable to the Ferrari/Usain Bolt analogy.

Sure... in that it is man vs machine. But I would argue that there is a pretty important difference between man-legs vs machine-legs, and man-brain vs machine-brain.

[kurros]

A computer may know every medicine in existence and know exactly what the medicine cures.  That would be soft AI.  But in order for it to be strong AI, the computer would need to know what "medicine" is, and what "sickness" or "illness" is.  I'm afraid AI currently does not know this.  AI is very good at organizing, calculating, categorizing, but not good at knowing what abstract concepts "mean".

That's not what people usually mean by strong AI. They mean that it is functionally equivalent to a human in general reasoning skills. Note "functionally". What you are talking about is some ephemeral unmeasureble thing beyond the realms of science. AI researchers do not care about such things.

[Nelvan]

Kurros,

I'm sure some scientists would love for AI to understand abstract concepts like "medicine" and "sickness".  And if they could program such an AI, they would.  It would be the continuing undermining of humanity, which is what some materialists are after.  "Humans are not so special after all, see?  Told you so."

[kurros]

Kurros,

I'm sure some scientists would love for AI to understand abstract concepts like "medicine" and "sickness".  And if they could program such an AI, they would.  It would be the continuing undermining of humanity, which is what some materialists are after.  "Humans are not so special after all, see?  Told you so."

Not sure what you are talking about. People are exactly developing AI like that, right now. They are called expert systems. Though actually that might be an old word for specific kinds of decision making AI. Modern neural net AI has much greater potential than traditional approaches.

[Lion IRC]

kurros,

Don't you agree that an invincible software game program which always beats its human opponent is actually behaving MORE artificially than the human?

The software can't resign from the game - unless it's 'told to'

The software can't strategically 'offer' a draw - why would it? It's programmed to win.

[Lion IRC]

Note also that Deep Blue vs Kasparov was a time competiton as well as the moves.
Obviously the software has more 'time' up its sleeve.

[kurros]

kurros,

Don't you agree that an invincible software game program which always beats its human opponent is actually behaving MORE artificially than the human?

The software can't resign from the game - unless it's 'told to'

The software can't strategically 'offer' a draw - why would it? It's programmed to win.

alphaGo will resign from the game if it thinks it is defeated. That is how almost all games of Go end. There is no equivalent of "checkmate". This problem of accurately understanding the state of the game was one of the big hurdles for previous Go-bots.

[Lion IRC]

I would argue than every one of the human-vs-AI contests are comparable to the Ferrari/Usain Bolt analogy.

Sure... in that it is man vs machine. But I would argue that there is a pretty important difference between man-legs vs machine-legs, and man-brain vs machine-brain.

See what you just conceded there?

The human brain isn't as fast as the computer for the EXACT same reason as the humans' legs aren't as fast as a machine. But so what? We already knew that.

[kurros]

I would argue than every one of the human-vs-AI contests are comparable to the Ferrari/Usain Bolt analogy.

Sure... in that it is man vs machine. But I would argue that there is a pretty important difference between man-legs vs machine-legs, and man-brain vs machine-brain.

See what you just conceded there?

The human brain isn't as fast as the computer for the EXACT same reason as the humans' legs aren't as fast as a machine. But so what? We already knew that.

No we didn't, and we don't. We know that machine brains are faster than humans at certain tasks, but until now Go has not been one of those tasks. And still isn't until you throw a couple thousand processors at the problem.