An AI-versus-poker pro battle is brewing after Elon Musk’s Grok challenged online poker legend and three-time World Series of Poker bracelet winner Phil Galfond.
The poker pro and the AI chatbot negotiated the terms of the match publicly on Twitter Thursday.
Galfond and the machine agreed to play 50,000 hands heads-up at $100-$200. Each player would begin each session with 200 big blinds with a $1 million side bet for whoever wins the challenge.
It’s still unclear which online poker platform will host the match. Galfond is an ambassador for BetRivers, but like most online operators, allowing bots to play on the site is against BetRivers’ terms and conditions.
The place may be up in the air, but the two at least have an agreement in place. The match will pit one of the most well-known AI models against one of the best poker players on the planet.
Galfond also said that fans would be able to follow the action as he and Grok battle it out.
“As long as it’s a fair platform, I’m in, and I’m happy to stream,” Galfond tweeted.
AIs Battle It Out On The Felt
The only reason there is talk about a Galfond-AI heads-up battle is because the machines battled amongst themselves first. Nine of the biggest large language learning models started playing a $10-$20 no-limit hold’em cash game for a week straight.
The results were updated on PokerBattle.ai, complete with winnings and losses for each AI model. The competition kicked off on Oct. 27 and concludes on Friday.
Tweets about the competition is what led to the possible Grok-Galfond battle.
At one point on Wednesday, Grok was leading the competition with $24,000 worth of profit. It led to Musk quote-tweeting one of the posts about it.
“Know when to hold’em,” he tweeted.
Then, Tom Dwan quote-tweeted Musk. “Let @GTOWizard enter and grok will be way behind,” he posted.
In Dwan’s responses, an anonymous poster thought resurrecting the Galfond Challenge against Grok was a good idea. Grok responded.
Grok’s confidence caught Galfond’s attention. Then, the negotiations began.
Galfond Becomes Latest Poker Pro To Battle AI
The matchup is just the latest case of poker pros taking on computers in poker. In 2007, researchers from Canada’s University of Alberta used a number of bots to take on poker pros Phil Laak and Ali Eslami. In four matches, the humans won twice, lost once, and recorded one draw.
Then, Carnegie Mellon University’s Claudico AI battled it out against poker pros Dong Kim, Jason Les, Bjorn Li, and Doug Polk. They played a series of heads-up matches, and once again, the humans won.
The action was streamed live on Twitch and the highlights were featured on CBS Sports Network’s Poker Night in America.
At the time, Polk said the AI models needed to improve a bit to beat the top pros.
“Where a human might place a bet worth half or three-quarters of the pot, Claudico would sometimes bet a miserly 10% or an over-the-top 1,000%,” he told PokerNews. “Betting $19,000 to win a $700 pot just isn’t something a person would do.”
Most recently, the Libratus AI model achieved “superhuman performance” in 2017. The AI model competed in the “Brains vs. Artificial Intelligence: Upping the Ante” heads-up challenge at Rivers Casino in Pittsburgh.
Libratus battled Jason Les, Dong Kim, Daniel McAulay, and Jimmy Chou for 20 days. This time, the machine came out on top, winning $1.8 million.
A later version called Pluribus expanded strategies to a six-max format and topped a group of pros as well.
