It taught itself to play chess in a few hours hours and then went onto to beat the top computational chess computer. Here is the key sentence.
Indeed, much like humans, AlphaZero searches fewer positions that its predecessors. The paper claims that it looks at "only" 80,000 positions per second, compared to Stockfish's 70 million per second.
We're all doomed, bring on the terminators.
“Results! Why, man, I have gotten a lot of results! I have found several thousand things that won’t work.”
I have to say that it's monstrous, as a chess enthusiast I say that I wouldn't need it, there's no taste. Of course technological progress will certainly not stop there, we should not be surprised.
Oh yeah, I'm finally thinking about creating an extension to play 2D/3D Chess, an old forgotten project.
It taught itself to play chess in a few hours hours and then went onto to beat the top computational chess computer. Here is the key sentence.
Indeed, much like humans, AlphaZero searches fewer positions that its predecessors. The paper claims that it looks at "only" 80,000 positions per second, compared to Stockfish's 70 million per second.
We're all doomed, bring on the terminators.
I hope it doesn't end like in the Terminator movie hehehe.
"The good news is hell is just the product of a morbid human imagination.
The bad news is, whatever humans can imagine, they can usually create." - Harmony Cobel
"The good news is hell is just the product of a morbid human imagination.
The bad news is, whatever humans can imagine, they can usually create." - Harmony Cobel
Most ten year old kids can identify the letter A in thousands of different fonts they have never seen before, that requires intelligence. For computational computer you give it example of every letter A and in can identify the ones you have given it. For AI computer you give a handful of examples and program it it to identify the rest by itself... I don;t know what you want to call that but at the very least it's mimicking human intelligence.
One other important point, the computational computer assuming it has error free data is always right. The AI computer can be wrong. When it's wrong it doesnt file away an example of the letter A it got wrong. It figures out how to arrive at correct answer.
“Results! Why, man, I have gotten a lot of results! I have found several thousand things that won’t work.”
All animals have, for want of a better phrase an EPROM in their heads, the purpose of which is to get us going, growing, breathing, eating, feel, touch, sight, curiosity, walking, speech and movement etc. The rest of our learning is ad hoc, nurture not nature (IMHO) We learn through experience of life growing up and input provided. Each of us human beings has a billion or ten billion choices through life that shape us, some may be become mathematicians, some sportsmen some carpenters, the choice is massively diverse but the point is those choices have taken shape over years of personal experience.
Now, let’s take a normal 21 year old human being. Imagine what has gone on in his head since birth, imagine the input, it’s massive, probably too much to even think about. Now, say we hypothetically build an android, for want of a better word, he’s programmed with identical features, EVERYTHING is the same as the original, mentally and physically. But to get the android to mimic 100% the original human he has to be built, probably the day before, and programmed with everything that’s in the humans head. He’ll remember being 3’6” but never was. At this stage are we sure, are we really sure the android is safe to be let out? After all, supposedly he’ll be a sentient being and capable of changing his mind, he could even take up religion. As far as I see it, if it has to be (pre)programmed, apart from the initial EPROM, it’s not A1 it’s just a super computer.
I can guarantee the girlfriend would tell the difference.
"The good news is hell is just the product of a morbid human imagination.
The bad news is, whatever humans can imagine, they can usually create." - Harmony Cobel
