Libratus Online Poker Bot Hired by US Army

Libratus Online Poker Bot Hired by US Army

0 comments 08 May 2019, 23:26

In 2017 the world of poker was shocked to witness 4 professional players being defeated by artificial intelligence. The game took place over the course of multiple days, and Bjorn Li, Doug Polk, Dong Kim and Jason Les played against Libratus, a poker bot.

Even though they’ve played the most complex version of all (the no-limit Texas Holde’em), the bot became better and better with each day, complex algorithms running on supercomputers and coming up with improved strategies. Libratus eventually won $677,000 against the 4 pro players. This complex bot was made by two researchers from the Carnegie Mellon University, based on the technology called computational game theory.

Poker is more difficult to be learned by AI compared to other games like chess. This is because each player only sees his own cards, but not the ones of the opponents. So the information available is not complete. The bot came with various betting strategies and even learned how to bluff.

Libratus did not have a pre-programmed strategy, but took decisions on the go. Its algorithm relied on “action abstractions” and “card abstractions”, which allowed him to take the possible decisions from 10161 down to 1012.

One of the two developers created a company called Strategy Robot, where he adapts the technology of the poker bot to the needs of the government. And they soon managed to get the first contract with the US army, which hired this complex robot for no less than 10 million dollars!

Libratus will be working for the next 2 years for the Defense Innovation Unit, an agency of the Pentagon. The United States is not the first country to show interest in AI, both China and Russia exploring the potential of AI in military applications. Even president Putin said that the country that will be the leader in AI will rule the world.

Most of those who are developing AI are against its use for military purposes. For example, some of the researchers for AI at Google joined thousands of other employees in a protest against the involvement of the company in the Project Maven. This program wants to implement commercially available AI to US military missions.

Tuomas Sandholm, one of the developers of the Libratus poker bot says people should not be worried and that AI can make the world a much safer place. What do you think about this? Let us know in the comments area below!

No Comments

No Comments Yet!

You can be first one to write a comment

Leave a comment