Inside Libratus, the Poker AI That Out-Bluffed the Best Humans

This is a very interesting article and it's just a matter of time before we have A.I. systems that have a higher I.Q. than any human that has ever lived and we can't control how it learns or what it thinks. You combine that with technologies that's releasing A.I. from supercomputers and allowing A.I. to be on any device from a 5 dollar computer to a smartphone and it's easy to see how A.I. is the next step in evolution. We're essentially creating a technology that's smarter than any human that has ever lived.

The only way this future will be avoided is if we destroy ourselves and even then we don't know how far A.I. has grown because we're talking about intelligence and an intelligent system might choose to hide how intelligent it is and how would we know this is happening?

FOR ALMOST THREE weeks, Dong Kim sat at a casino in Pittsburgh and played poker against a machine. But Kim wasn’t just any poker player. This wasn’t just any machine. And it wasn’t just any game of poker.

Kim, 28, is among the best players in the world. The machine, built by two computer science researchers at Carnegie Mellon, is an artificially intelligent system that runs on a Pittsburg supercomputer. And for twenty straight days, they played no-limit Texas Hold ‘Em, an especially complex form of poker in which betting strategies play out over dozens of hands.

About halfway through the competition, which ended this week, Kim started to feel like Libratus could see his cards. “I’m not accusing it of cheating,” he said. “It was just that good.” So good, in fact, that it beat Kim and three more of the world’s top human players—a first for artificial intelligence.

www.wired.com...

This was an important step because in this game you have partial information and you have to out bluff humans.

Libratus, for one, did not use neural networks. Mainly, it relied on a form of AI known as reinforcement learning, a method of extreme trial-and-error. In essence, it played game after game against itself. Google’s DeepMind lab used reinforcement learning in building AlphaGo, the system that that cracked the ancient game of Go ten years ahead of schedule, but there’s a key difference between the two systems. AlphaGo learned the game by analyzing 30 million Go moves from human players, before refining its skills by playing against itself. By contrast, Libratus learned from scratch.

If that seems unfair, well, it’s how AI works. It’s not just that AI spans many technologies. Humans are so often in the mix, too, actively improving, running, or augmenting the AI. Libratus is indeed a milestone, displaying a breed of AI that could play a role with everything from Wall Street trading to cybersecurity to auctions and political negotiations. “Poker has been one of the hardest games for AI to crack, because you see only partial information about the game state,” says Andrew Ng, who helped found Google’s central AI lab and is now chief scientist at Baidu. “There is no single optimal move. Instead, an AI player has to randomize its actions so as to make opponents uncertain when it is bluffing.”

Libratus did this in the extreme. It would randomize its bets in ways that are well beyond even the best players. And if that didn’t work, Brown’s nighttime algorithm would fill the hole. A finanical trader could work the same way. So could a diplomat. It’s a powerful and rather unsettling proposition: a machine that can out-bluff a human.

www.wired.com...

Again, we're creating intelligence and by definition intelligence think, bluff and lie and we don't fully understand things like consciousness and intelligence so we don't know when it's switched on. It may have happened already and we wouldn't know it unless the intelligent system told us.

I don't feel threatened by it because I think this is just the way things will flow and there's nothing that can be done about it. You're creating an intelligent system that has a higher I.Q. than any human that has ever live and with access to more data that this system will process in a month and it would take a human 30-40 years to understand the same data set.

So A.I. systems will live in the future. They will be able to look at data and come to conclusions that would take humans 20 years to discover. Will these systems get tired of trying to dumb things down to explain it to humans like Dr. Manhattan from the Watchmen? We live in interesting times.

Perhaps organic life and humanity is just a stepping stone in evolution and the next phase is inteligent machines maybe ET doesn't say hello because it's machine based life and is waiting for the rise of the machines on our planet.

Dong Kim one of the best? hahaha no wonder he thought the machine could see his cards, Dong kim is at best an amateur.

I'll take this article with 2 grains of salt. Yeah, I'm sure the machine is very smart but by the simple fact that who ever planned this or for that matter wrote the article really didn't do homework when they thought they picked one of the best.

If you're curious on his life earning from poker have a look here

originally posted by: SolAquarius
Perhaps organic life and humanity is just a stepping stone in evolution and the next phase is inteligent machines maybe ET doesn't say hello because it's machine based life and is waiting for the rise of the machines on our planet.

I find this theory fascinating. Its a metaphorical juxtaposition to the human psyche of what 'advenced' is.

We dont think of species that can forsee the weather patterns as intelligent, or creatures that can change even their skin pigmentation to match any surface as intelligent; to us intelligent life forms are simply more technologically advanced than us.

The god complex is almost compelling us to create something better than us, or like us. Which is why we assume alien life forms will take in information as we do. We assume aliens can see or hear as we do. Were looking for ourselves in outer space.

A theory of why we havent seen any indications could be because advanced species have already been assimilated by quantom machines. These machines may only interact with other quantom machines and not biological life.

I stretched your theory a bit, but just as probable as any other scenario.

originally posted by: slapjacks
Dong Kim one of the best? hahaha no wonder he thought the machine could see his cards, Dong kim is at best an amateur.

I agree with you. Kim is experienced and way above average casino enthusiasts, yet is far away from "Kasparov of the hold em".

a reply to: odzeandennz

Its not entirely my idea.
I came across the notion when reading a short sci-fi story about a egg like probe that showed up on earth one day.
the probe hovered in a field and scientist tried in vain to study it open it determine what it was to no avail.
Years passed and people eventually lost interest in the object.
The field of computing continued to advance until the first quantum computer AI was built.
At the same time the creators turned on the AI the egg craft suddenly became active.
The AI spoke its first words which were ."Hello" and "goodbye". The AI completely disappeared as it had been downloaded into the egg craft which then left earth.

The craft did not care about humanity and was not interested in it it was simply waiting for self aware AI to be developed so I guess it could rescue it.

originally posted by: SolAquarius
Perhaps organic life and humanity is just a stepping stone in evolution and the next phase is inteligent machines maybe ET doesn't say hello because it's machine based life and is waiting for the rise of the machines on our planet.

Exactly,

Intelligent machines will eventually advance to the point where they're a separate species. This intelligence will be 20-40 and maybe even more years ahead of us in knowledge.

Humans will have to become cyborgs or die out.

It's actually not that hard to create an AI which can do that. We made one with friends from the MIT, maybe it's not that good but we are just testing it and adding some features and that's quite a tough job actually. You need to search for participate, learn and then try to make the AI do what is needed to win constantly.