re: Skynet is Upon Us

I mean it's not like we haven't been aware of the downside for decades now. Leave it to human arrogance to be stupid enough to create something that will destroy itself.

frick Atlas. That’s a terrible idea

quote:

---

frick Atlas. That’s a terrible idea

---

Besides Atlas there is,

A Russian robot that shoots two GLocks, one in each hand just drills the targets, never missing.

A British Robot with AI that escaped from its lab twice, they captured it when the batterys died.

A robot that can recognize AC power outlets and plug itself in when a charge is needed.

Combine all of these with the Google AI into a single machine and the world becomes a little more dangerous.

quote:

---

I would be interested to know when the winning games occurred. I would imagine alot of them happened after the 50 game mark, as Alphas learning curve flattened.

---

I am not sure. I haven't read the paper (it should be easily found online). I do know that they analysed all of the practice games Alpha played against itself and noticed that it "evolved" it's opening strategy over time. It came to prefer the "Berlin defense" as black, which is a defense most human grandmasters had stopped playing a long time ago because modern theory suggested it was weak (I guess modern theory is wrong).

quote:

---

I would also be interested to know how many games Alpha was able to simulate in the 4 hour period. Did they track the "real time" of those games? Take the number of games simulated x the "real time" of those games and get X. Then compare that to Malcolm Gladwell's Outliers theory that it takes 10,000 hours to become a master at something.

---

I believe they said that after 4 hours it had played something like 40 million games. Google had it hooked up to a supercomputer I think, so its performance would be much greater than would be expected on a PC, for instance.

From my understanding there is a diminishing return on letting this sort of neural network "practice too much." Supposedly, in theory, it could even become worse at a certain task over time. So I am not sure how Google figured that in when they determined how much practice they wanted to allow it.

quote:

---

The future of AI is really interesting, and I am looking forward to seeing what happens when they start trying to tackle the real world problems, like hunger/disease etc. using these AI computers.

---

That's Google's ultimate goal. They are only using Go and chess as mere demonstrations in their early testing. The paper was sort of tight lipped on some details from what I hear. This means they are probably not wanting to give away too much to competitors. It's also very possible (put your tin foil hat on) that they plan on selling it to the military/CIA.

quote:

---

Like anything though, that type of power in the wrong hands could be catastrophic. The future is a bright scary place to look forward to.

---

The Pentagon said the other day that Russia already has a nuclear missile controlled by AI. However, the decision to launch is not controlled by the AI, but only the guidance system after launch (or so they say -- tin foil hat time again).

quote:

---

Someone explain to me why it's harder to win with black vs. white. Does white always have first move?

---

Yes, white always moves first.

quote:

---

The Pentagon said the other day that Russia already has a nuclear missile controlled by AI. However, the decision to launch is not controlled by the AI, but only the guidance system after launc

---

Why, in the everloving hell, would they allow this? What is there to gain?

quote:

---

Why, in the everloving hell, would they allow this? What is there to gain?

---

Really? It makes total sense. Like a self-driving car for a missile.

quote:

---

Yes, white always moves first.

---

quote:

---

It came to prefer the "Berlin defense" as black

---

How many Steiner rants did it do?

quote:

---

Besides Atlas there is,

A Russian robot that shoots two GLocks, one in each hand just drills the targets, never missing.

A British Robot with AI that escaped from its lab twice, they captured it when the batterys died.

A robot that can recognize AC power outlets and plug itself in when a charge is needed.

Combine all of these with the Google AI into a single machine and the world becomes a little more dangerous.

---

Keep in mind, any robot that is connected to the internet can be potentially equipped with state of the art AI. Now granted, if you cut its connection then it is limited to its own programming. But if connected, then Atlas could control a synchronized army of automatons using its AI.