Poker is one of the most vexing games for professionals and amateurs, alike. There are many professionals that can make millions at poker, only to lose to an amateur who got lucky with a good hand or just didn’t play the game the way professionals would think.
Suffice it to say, poker is tough for humans to really master. And that’s what makes it so compelling. It’s also what makes a collaboration between CMU and Facebook published recently in Science that examined whether a robot could beat five professional poker players even more interesting.
That robot, which is really an artificial intelligence platform, is called Pluribus. It took on five professional poker players at a game of No Limit Texas Hold ‘Em, one of the more popular poker games in the world, which requires players to deliver the best hand with a combination of the two cards they have and the five community cards everyone at the table can use.
Pluribus was outfitted with a slew of tweaks and algorithms to help it decide what to do on a hand, including fold, raise, call, or bluff. It also needed to analyze the other five players’ moves and make on-the-spot decisions on whether those players had the better hand.
Ultimately, Pluribus played 10,000 hands against those professionals. And although it didn’t win every hand, the researchers said that Pluribus ultimately won and most importantly, didn’t expose any weaknesses or odd playing behaviors that the professionals could have countered.
“The bot doesn’t take any of these short cuts and has an immensely complicated/balanced game tree for every decision,” professional Jimmy Chou said in the Science report.
Indeed, the researchers called Pluribus’ play as “ruthless consistency.” The bot was able to analyze every hand and every move and even after the hand was over, the machine would analyze how it played that and consider other possibilities, like raising or calling in a different situation, to see how it might have turned out. It then used that information to inform future decisions.
Of course, that’s something that humans can’t easily do. And that’s a reality that professional Sean Ruane bristled at in the Science piece. He noted that poker requires you to consistently exhibit “mental discipline, focus, and consistency.” But humans are humans. And bots are bots.
“Competing for hours on end against an AI bot that obviously doesn’t have to worry about these shortcomings is a grueling task,” Ruane said.
But Pluribus beating poker players isn’t the most important feat.
Poker is a very complex game that requires sophisticated thinking to be successful. And Pluribus has proven that it can get the job done. It also means that the technology could be used in the future to power other sophisticated tasks and eliminate the need for humans to perform them. It also suggests that as artificial intelligence gets better, companies will benefit in a myriad of ways.
Indeed, as a business owner, eyeing artificial intelligence and machine learning and identifying how that could change your own business is critical. And as Pluribus clearly shows, machines might do a better job than you might think at handling even complicated tasks in your business.
And perhaps that’s the most imporant component in evaluating the future of machine learning. There has been some thought that artificial intelligence can replace simpler tasks or job functions. But as technology has improved and machine learning has grown in its sophistication, it’s now clear that it can be used for a variety of tasks. And best of all, it can reduce costs.
But as those who worry about artificial intelligence, like Elon Musk, note, where’s the line in determining what’s best for companies and what goes too far? And is there a possibility of artificial intelligence taking on a too-important role in the enterprise?
The fact is, no one knows. But what is clear is that machine learning and artificial intelligence can be used in increasingly important ways. And it’s incumbent upon all business owners to know when to use it — and how.
All Rights Reserved for Don Reisinger