ESPN’s World Series of Poker streams are adding an AI tool that studies players’ body language to predict whether they are bluffing or holding the nuts, according to Sportico. The feature is being built into coverage of the Main Event, the game’s marquee tournament.
On March 26, ESPN and the WSOP announced a multi-year agreement to bring the Main Event back to ESPN platforms beginning this summer. The package includes more than 100 hours of multiplatform coverage, with the Main Event starting July 2 on the ESPN App, edited episodes on linear networks, and a three-night live finale from Aug. 3-5, from 9 p.m. ET to midnight ET.
The Main Event is the $10,000 No-Limit Hold’em World Championship. ESPN’s press release said it first broadcast the event in 1987 and has been a broadcast partner in every decade since.
The AI system was developed by independent engineer Luke Geel. Sportico reported that he spent roughly six months building it and said the process was harder than he had expected.
Geel said he could not simply upload a YouTube URL and ask the system to find players’ tells. Instead, the model analyzes past results and visible activity, including posture and blink rate.
Peyton Manning’s Omaha Productions, newly in charge of the event’s TV coverage, enlisted the model. Sportico said Omaha plans to be cautious about how it uses the feature, focusing only on players after they have been eliminated.
After he posted about the work online, Geel said multiple players reached out about using it. He also said some people asked whether he could run the tool on Daniel Negreanu, but he said there would be no tells.
Geel said he sees no reason computer-vision prediction models cannot find other uses beyond poker if the technology improves further. Pokerfuse separately reported that similar technology could eventually be used in other sports broadcasts.
