Any poker player with a winning bankroll will tell you that to become a true great at the table, you uncover what opponents are trying to hide — noticing a slight physical tic, or the way one shuffles their chips, or even how they breathe. Those “tells” can reveal as much to an elite poker pro as if their opponent turned over their hand and showed them their cards. The best of the best understand and play the game on a level few will ever truly achieve.

And a bunch of them just got their asses kicked by an advanced artificial intelligence program.

As reported by New Scientist, the AI program Libratus from Carnegie Melon University has now bested four professional players after 120,000 hands in a 20-day, no-limit Texas Hold’em heads-up tournament (the AI played directly against one player at a time, with each rotating through). At the end of the nearly three-week-long tournament, Libratus was ahead by more than 1.7 million “dollars.” (The players didn’t lose any real money, but did get to split a 200,000 dollar purse for playing.)

https://twitter.com/nicknystrom/status/826455558341402625

Libratus’ performance is more impressive than you think. Libratus’ algorithms were not specific to playing Poker — the A.I. simply solved problems. All it “knew” were the rules of the game. It learned as it went along and analyzed previous hands well enough to dominate the human players in what is considered a “sizable victory” that is “statistically significant and not simply a matter of luck.”

By the end, Libratus not only learned concepts like bluffing and concealing the value of its own hand, but also the strengths and weaknesses of its opponents. It’s an impressive feat considering the machine never saw the human players’ cards afterwards.

Image: Miramax

Like any real poker player will tell you, however, often one of the biggest issues to overcome at the table has nothing to do with other good players, it’s “plugging your own holes,” self-inflicted mistakes, often minute, that can lead to big losses over time.

In a article from Carnegie Mellon about the “Brains vs. Artificial Intelligence: Upping the Ante” event, one of the two computer scientists behind Libratus, Tumoas Sandholm, explained how it learned about itself from examining how the real pros took advantage of its mistakes.

After play ended each day, a meta-algorithm analyzed what holes the pros had identified and exploited in Libratus’ strategy. It then prioritized the holes and algorithmically patched the top three using the supercomputer each night. This is very different than how learning has been used in the past in poker. Typically, researchers develop algorithms that try to exploit the opponent’s weaknesses. In contrast, here the daily improvement is about algorithmically fixing holes in our own strategy.

Self-reflection as a means of improvement — it sounds so easy, but without any preconceived notions about “the right way” of playing, and without any emotional baggage that can make us stubborn or blind, Libratus became a better poker player, all while processing huge amounts of information in a short time.

Image: Universal Pictures

Some might argue fairly that a computer has unfair advantages over a real player. It never gets tired or hungry, it never needs to use the bathroom, and it can’t be distracted by outside forces (sometimes casinos get loud, and the guy next to you is sitting too close). An AI also doesn’t have to feel the very real pain of knowing its own money is on the line.

None of that is an indictment of real poker players though. The best players shut out the distractions and pressures of playing, all while continuously evaluating opponents and analyzing their own play. Rather, the victory speaks to the power of advanced computer algorithms, who are free of things like recency or confirmation bias, and all of the other failings of human judgement that can hinder our decision making.

What do you make of this news? Is it a great advancement for the future of mankind, or the beginning of the end for our dominance (or being able to win any money in online poker)? We’re not bluffing when we say we want you to lay your cards out on the table in our comments below.

Featured Image: Fox