Poker Coverage: Poker Legislation Poker Tournaments U.S. Poker Markets Sports Betting

Poker Bot Technology Being Applied By US Military

'Libratus' Developer Tuomas Sandholm's New Venture Involves Applying What He Learned From Poker To War Games

Print-icon
 

Carnegie Mellon University Computer Science Professor Tuomas Sandholm developed Libratus, the poker bot that infamously defeated a group of world-class professional heads-up no-limit hold’em specialists.

But dominating a specific poker format was never the end goal for researchers like Sandholm. Instead, computer scientists had been targeting poker as a stand-in for complex real-world problems, which they one day hoped to tackle using what they’ve learned about artificial intelligence algorithms by tackling heads-up poker. Now Sandholm has begun to take the next step with the technology that was behind Libratus, applying it to military strategy for the US Army according to Wired magazine.

“Games make great example problems because the rules are well described and scoring makes evaluation easier. While some of the research in the last decade plus was specific to poker, the large majority of it was on general techniques, applied to poker. The techniques and ideas that came out of the poker research can be useful elsewhere,” Neil Burch told Card Player in a 2015 interview. Burch was a PhD student at University of Alberta at the time and a co-author of a paper in Science magazine entitled, “Heads-up Limit Hold’em Poker is Solved.”

Tuomas SandholmIn early 2018, Sandholm started a company called Strategy Robot, looking to apply what he had learned with Libratus to new incomplete-information problems, such as war simulations and military strategy analysis. The company received a two-year, $10 million contract with the US Army last August via the Defense Innovation Unit, an organization founded to help the US military apply commercial technological breakthroughs to their needs more quickly.

Sandholm’s experience with using AI to create a poker bot that could win at a rate of more than 14 big blinds per 100 hands versus top human opponents helped him bring a new perspective to complex problems like military simulation and strategic planning.

“Since the earliest days of AI research, beating top human players has been a powerful measure of progress in the field,” Sandholm said after Libratus won the Brains vs. AI challenge. “That was achieved with chess in 1997, with Jeopardy! in 2009 and with the board game Go just last year. Poker poses a far more difficult challenge than these games, as it requires a machine to make extremely complicated decisions based on incomplete information while contending with bluffs, slow play, and other ploys.”

Sandholm’s progress in creating a computer program that can make such complicated strategic decisions in scenarios with incomplete information could prove crucial to improving military simulations, which might ultimately lead to better training and planning. Sandholm told Wired that war games currently only assign a small number of strategies the simulated opponents, which could open those training with these more simplistic simulations to being strategically exploitable against real-world opponents.

While artificial intelligence is being applied by the US military and others around the globe, the technology has also been increasingly implemented in the business sector. An estimated $12 billion was spent on AI and machine learning in 2017, and that number that is expected to grow to over $57 billion by 2021 according to a study by the International Data Corporation.

Given that information, it is hardly surprising that Sandholm founded another company in addition to the one working with the military. His business-oriented startup, called Strategic Machine, will look to utilize the strategic problem-solving ability of Sandholm’s work in the commercial marketplace.