Google AI creates its own AI Child that is more advanced than Human Designed AI

[reckless toboggan]

I, for one, welcome our AI overlords.

GOOGLE AI CREATES ITS OWN ‘CHILD’ AI THAT’S MORE ADVANCED THAN SYSTEMS BUILT BY HUMANS

The AutoML project is said to make process of designing machine learning models 'much more accessible'

Google has developed an artificial intelligence (AI) system that has created its own “child”.

What’s more, the original AI has trained its creation to such a high level that it outperforms every other human-built AI system like it.

It’s an impressive achievement, but one that could also trigger fears about what else AI could create without human involvement.

Google unveiled its AutoML project in May, with the aim of making it easier to design machine learning models by automating the process.

“In our approach..., a controller neural net can propose a ‘child’ model architecture, which can then be trained and evaluated for quality on a particular task,” the company said at the time.

“That feedback is then used to inform the controller how to improve its proposals for the next round. We repeat this process thousands of times — generating new architectures, testing them, and giving that feedback to the controller to learn from.”

In November, the AutoML plans were used to create NASNet, a “child” AI designed for object detection, which outperformed state-of-the-art machine-learning architectures built for academic competitions by humans.

To test NASNet, Google applied it to the ImageNet image classification and COCO object detection dataset, which it describes as “two of the most respected large scale academic datasets in computer vision”.

On ImageNet, NASNet achieved a prediction accuracy of 82.7 per cent, performing 1.2 per cent better than all previous published results.

On COCO, Google says NASNet achieved “43.1% mAP which is 4% better than the previous, published state-of-the-art [predictive performance on the object detection task]”.

“We hope that the larger machine learning community will be able to build on these models to address multitudes of computer vision problems we have not yet imagined,” said the researchers, who have open-sourced NASNet so it can be used for computer vision applications.

[hatchgreenchile]

What's the child's name? T-800?

[digitaldeath]

you guys know about Elsagate?

its fucked!

[SkiBall]

Somebody get Sarah Connor on the line.

[riser4]

Somebody get Sarah Connor on the line.

I think we are a bit late.

[The Tortoise]

Basically Westworld is now a reality. That ended well for the humans.

[MontuckyFried]

Anybody else here seen the movie, Ex Machina? Hits this nail on the head for sure. Especially with Google-esque overlord sociopath.

[SkiBall]

Ex-Machina was a trip. Good movie, would bang.

[MontuckyFried]

Ex-Machina was a trip. Good movie, would bang.

Indeed. It was my introduction to Alicia Vikander. She really did a great job with the role of Ava. I loved Oscar Isaac's performance in there too. Too bad this one flew under the radar for most. I really liked the movie. Also my first movie with a DTS:X soundtrack, which is also very cool.

[Rasputin]

[mall walker]

there's a guy Eliezer Yudkowsky who is on FB/etc, works for MIRI which is sort of an AI watchdog thinktank organization. he has some pretty fascinating writing on the development of this, from AlphaGo on.

[Mazderati]

Commentary from a poster on another forum:

The program recently defeated stockfish one of the most powerful chess engines in the world and it wasn't even really a close game (link to game in post). Stockfish has an elo (chess strength rating) of 3400. To give a little perspective the top human in the world and the current record holder for the highest elo rating ever achieved by a human is Magnus Carlsen with an elo of 2882. The amazing thing is that Deep mind was not given any access to any chess information such as previous games or recommended strategies, it was only given the rules of the game. It learned everything in four hours simply by playing games against itself. Another interesting part of the game is that Deep mind played an opening called "the kings Indian opening" that is popular among Grandmasters today and has developed over hundreds of years of chess knowledge. The algorithm was able to essentially surpass our knowledge of that opening in just 4 hours! Really crazy stuff shows the power of AI and is maybe a little scary too!


http://www.telegraph.co.uk/science/2...nds-alphazero/

[MagnificentUnicorn]

Commentary from a poster on another forum:

The program recently defeated stockfish one of the most powerful chess engines in the world and it wasn't even really a close game (link to game in post). Stockfish has an elo (chess strength rating) of 3400. To give a little perspective the top human in the world and the current record holder for the highest elo rating ever achieved by a human is Magnus Carlsen with an elo of 2882. The amazing thing is that Deep mind was not given any access to any chess information such as previous games or recommended strategies, it was only given the rules of the game. It learned everything in four hours simply by playing games against itself. Another interesting part of the game is that Deep mind played an opening called "the kings Indian opening" that is popular among Grandmasters today and has developed over hundreds of years of chess knowledge. The algorithm was able to essentially surpass our knowledge of that opening in just 4 hours! Really crazy stuff shows the power of AI and is maybe a little scary too!


http://www.telegraph.co.uk/science/2...nds-alphazero/

A little scary?

[Dantheman]

The singularity is near....

[mall walker]

fwiw the stuff with Go is way scarier than with Chess. Chess has been a "solved" game for a while now, but as recently as 5 years ago people were saying an AI might NEVER be better than the best humans at Go

[Dantheman]

fwiw the stuff with Go is way scarier than with Chess. Chess has been a "solved" game for a while now, but as recently as 5 years ago people were saying an AI might NEVER be better than the best humans at Go

IIRC, AlphaGo was programmed with moves and strategies and reviewed countless games between human players, and then played countless games with human and virtual opponents before it became powerful enough to beat the best human players. Basically a pretty standard machine learning strategy. AlphaZero was given nothing but the rules of chess and allowed to play against itself, and within four hours became so good it could crush the chess versions of AlphaGo. It learned everything on its own from scratch, which is pretty astonishing.

[mall walker]

true that is astonishing - I guess I was mistaken in thinking that AlphaZero had self-trained in Go, not chess

[neckdeep]

[doebedoe]

Commentary from a poster on another forum:

The program recently defeated stockfish one of the most powerful chess engines in the world and it wasn't even really a close game (link to game in post). Stockfish has an elo (chess strength rating) of 3400. To give a little perspective the top human in the world and the current record holder for the highest elo rating ever achieved by a human is Magnus Carlsen with an elo of 2882. The amazing thing is that Deep mind was not given any access to any chess information such as previous games or recommended strategies, it was only given the rules of the game. It learned everything in four hours simply by playing games against itself. Another interesting part of the game is that Deep mind played an opening called "the kings Indian opening" that is popular among Grandmasters today and has developed over hundreds of years of chess knowledge. The algorithm was able to essentially surpass our knowledge of that opening in just 4 hours! Really crazy stuff shows the power of AI and is maybe a little scary too!


http://www.telegraph.co.uk/science/2...nds-alphazero/

I dabble in chess -- looking at some of the reviews of the games they played, one of the crazy differences between Google's AI and Stockfish is the kind of moves they play. Stockfish (and all other engines) are known for "engine-like" moves....things that are so precise and so deep that humans don't really have a shot of finding them without help. This AI's approach is different and appears to be yielding much more intuitive moves that are on par or better than stockfish. It's play is more-human in that sense.

It's also fascinating to see how it developed preferences over time in which openings it like. Stockfish doesn't demonstrate such variety or learning.

As a chess dweeb its pretty exciting. What is surprising to me is that my SO who works w/ ML as part of her comp sci doctorate right now is as equally excited.

[reckless toboggan]

More human than human...

[Dantheman]

How long until "GoZero" comes out and does this to AlphaGo? Over/under at a year?

[Tippster]

The singularity is near....

More human than human...

If we let it. Remember, all it takes is a switch to turn it off.

[old goat]

I, for one, welcome our AI overlords.

But can AI translate the original article into terms a lay person can understand? That would be true AI.

[reckless toboggan]

If we let it. Remember, all it takes is a switch to turn it off.

A singular consciousness that spawned an entire race of machines. We don't know who struck first, us or them. But we know that it was us that scorched the sky. At the time they were dependent on solar power and it was believed that they would be unable to survive without an energy source as abundant as the sun. Throughout human history, we have been dependent on machines to survive. Fate it seems is not without a sense of irony. The human body generates more bio-electricity than a 120-volt battery and over 25,000 BTU's of body heat. Combined with a form of fusion the machines have found all the energy they would ever need. There are fields, endless fields, where human beings are no longer born, we are grown. For the longest time I wouldn't believe it, and then I saw the fields with my own eyes. Watch them liquefy the dead so they could be fed intravenously to the living. And standing there, facing the pure horrifying precision, I came to realize the obviousness of the truth. What is the Matrix? Control. The Matrix is a computer generated dream world built to keep us under control in order to change a human being into this. (Cue the battery)