"AestheticsInMotion" (aestheticsinmotion)
12/05/2018 at 23:08 • Filed to: None | 2 | 10 |
And then there’s spending 4 hours coming up with a proof for why the first player to get hit is much more likely to lose a game of chopsticks, and trying to apply that in a way that allows you to come up with an equation to find the ideal chess move at any time based on the board- s tate.
I’m not done yet , but I will say it’s mind-blowing to think that the chess grand masters of the late 90's and early 2000's could still beat computers somewhat regularly. I quite literally cannot fathom how these people’s minds work, to have the limitations of a human mind and still be able to work so far in advance with an ever-expanding tree of possibilities.
Dr. Zoidberg - RIP Oppo
> AestheticsInMotion
12/05/2018 at 23:17 | 2 |
The good news is that your brakes didn't fail on that hill.
DipodomysDeserti
> AestheticsInMotion
12/05/2018 at 23:20 | 0 |
There’s nerding out, and then there Obsessive Compulsive Disorder...or marijuana.
I like cars: Jim Spanfeller is one ugly motherfucker
> AestheticsInMotion
12/05/2018 at 23:27 | 0 |
Oh man, I remember that game from elementary school. We called it Chinese Numbers. Yeah, as long as you don’t go first, you’re bound to win if you know what you’re doing.
Dr. Zoidberg - RIP Oppo
> DipodomysDeserti
12/05/2018 at 23:32 | 1 |
Unfortunately I am not seeing marijuana in my vicinity.
Spamfeller Loves Nazi Clicks
> AestheticsInMotion
12/05/2018 at 23:41 | 1 |
Oh man. We need to find a bar to meet up at just to talk about this shit now. Plus the 4VL-is-mandatory rant must be delivered in person. Also the ‘Watson is bullshit’ rant, but I need alcohol to deliver that one. Because fundamentally that’s what it really comes down to.
Not raw processing power (though the brain is still vastly superior there,) not clock speeds, not storage. Even if you get all of those to surpass the human brain, it will still never be even 10% as good at problem solving.
AestheticsInMotion
> Spamfeller Loves Nazi Clicks
12/05/2018 at 23:47 | 0 |
Down. But is it really problem solving if you’re just looking at all branching p ossible outcomes? Or are you talking about in general, and not just my example?
AestheticsInMotion
> Dr. Zoidberg - RIP Oppo
12/05/2018 at 23:48 | 0 |
The Volvo has been perfect lately. Maybe there were just initial teething issues and it will all be smooth sailing from here on!
Spamfeller Loves Nazi Clicks
> AestheticsInMotion
12/06/2018 at 00:14 | 1 |
Oh man. This is one of those things that can only be explained sensibly if you’ve got at least one good beer and good food in me. But I will give it a shot.
Basically, the issue can be summarized thusly: human thought is inherently non-binary and non-deterministic. It is not 0/1, yes/no, true/false. Pretending because neurons are on/off that you can mimic them is the greatest branch of idiocy to every come out of computer science, and yes, I will yell that to a PhD’s face.
Computers can win chess because it i s a binary, deterministic problem. They do not play chess. They do not think about chess. Kasparov did not beat Deep Blue because Deep Blue had insufficient power; he beat Deep Blue because the human brain is not binary. Instead of A/B openers, he had [A-ZZZ]. So when he used !A !B, and the computer spent all it’s time on A+B? It had no value for C.
While game 6 was allegedly anti-computer tactics and long-term planning, that’s simply not sufficient nuance. Those are two things a computer system should be able to counter easily. Computers think binary (i.e. action A = outcome B,C,G,Q, Q = optimal, select A) ; he leveraged non-binary (i.e. action B = outcome B, C, F, G, H, K, L, Q, R, which outcome is optimal?) This allowed him to clearly see the binary tree model of the computer, which is rigid and inflexible. Any time the computer expected outcome B, C, G, Q, if Kasparov instead delivered R? Nobody believed R would happen. Nobody planned for it. And down goes the machine as it tries to reassemble 0's and 1's into 3's and 9's.
The reason computers win now though, is because it’s a matter of a binary deterministic problem that was historically O+N hard to calculate . Once it is no longer O+N hard, or, you have sufficient power to ignore O+N? You feed it the rules, and you just kick back as it calculates every possible move and counter. Take all the time you like, store the data, trot it out to show off on occasion. It’s just a big, brainless database off chess table states. It pre-calculated the possible state, now it’s just picking the series of states most advantageous to it.
No system, no matter how advanced, cannot equal the human mind - or even compare to it - without 4VL. More is better, but 4 is the minimum. It’s the non-binary nature of human thought that is exactly what enables us to come up with the idea for AI in the first place, and to kick it’s ass next. (Or expose it as the total fraud it is.)
Future Heap Owner
> AestheticsInMotion
12/06/2018 at 00:16 | 0 |
Once you’ve solved chopsticks, check this out: https://en.m.wikipedia.org/wiki/Nim
On the subject of strategy games and computers: g o is, mathematically, a much harder game than chess. C hess has more possible games than there are grains of sand on a beach, but go has more possible games than there are atoms in the visible universe. Until last year — 20 years after Kasparov lost to Deep Blue — a computer was unable to beat the top human go player, and even a few years before, no computer could beat a professional go player.
The breakthrough came when researchers started incorporating neural networks in their go AIs, which are loosely modeled on the brain. This is in contrast to Deep Blue, which beat Kasparov using only dumb, exhaustive brute force search.
https://en.m.wikipedia.org/wiki/AlphaGo
davesaddiction @ opposite-lock.com
> Spamfeller Loves Nazi Clicks
12/06/2018 at 09:00 | 0 |
I enjoyed reading this. What it comes down to is this: computers can only do what we tell them to do.