Artificial intelligence is now able to beat professional poker players

>Until now. Scientists recently introduced DeepStack, an algorithm for imperfect information settings. DeepStack "combines recursive reasoning to handle information asymmetry, decomposition to focus computation on the relevant decision, and a form of intuition that is automatically learned from self-play using deep learning," researchers said in a recent paper published in Science magazine.

>In other words, DeepStack played poker against itself and learned how to make the same sorts of intuitive leaps in logic at key decision points in poker games that all professional poker players face, and then developed a level of intuitive reasoning that matches the pros.

>DeepStack then played those poker pros in real life, and defeated enough of them to prove that it had mastered the game – which is a feat that no AI research had been able to accomplish in the world of "imperfect information" games. "In a study involving 44,000 hands of poker, DeepStack defeated with statistical significance professional poker players in heads-up no-limit Texas hold'em," the researchers wrote in Science.

>Poker pros routinely assess odds in their heads before they make key decisions. They then generally bet (or not) based largely on the mathematical odds confronting them. But not always. Sometimes they go on intuition based on their own deep knowledge of the game, players and circumstances.

>DeepStack took that process – both the certainty of mathematical odds and the deep learning of intuition at key moments – and used that to beat a statistically significant number of poker pros in 44,000 hands of poker.

>While this breakthrough may seem innocuous, it's anything but.
...

>What DeepStack has done here, with the game of poker, explains why business leaders such as Elon Musk and scientists, including Stephen Hawking, have said publicly that the public should fear AI research without any guardrails in the future. AI could literally start to replace humans. This concern is exactly why a consortium of academic scientists and business leaders recently announced guidelines for AI research that are designed to build precisely these sorts of guardrails around AI to keep it from supplanting us.

>The Future of Life Institute, whose scientific advisory board includes Musk and Hawking, issued guidelines that it hopes all AI research companies will follow.

>"The goal of AI research should be to create not undirected intelligence, but beneficial intelligence," the institute said in announcing the core principles.

>Musk was thrilled.

>"Top AI researchers agree on principles for developing beneficial AI," Musk tweeted afterwards.

>But this was just the first step. AI research without boundaries can literally outgrow us, the human beings who created it. Once a super-intelligent, big-data-crunching AI machine learns how to think and learn for itself, it may decide that carbon life forms are the obvious target in any threat scenario. At that point, it won't care what world leaders, CEOs or research scientists think.
...

Inb4 insane anon rambling about ai not being real or some shit