ESPN's broadcast of the 2026 World Series of Poker Main Event introduced a new AI-driven 'tells detection' tool, but the system's small training dataset is raising doubts among professionals. According to Wired, the tool was designed by Luke Geel, an AI engineer for the US Air Force, and appeared periodically during the tournament's early July live broadcast. A text overlay displayed live metrics on players' movements and a 'hand strength model' chart estimating the likelihood of hand types.
How the AI Tells Detector Works
Wired reported that the system watches every hand captured on camera in the 2026 WSOP Main Event to build a tells database on various players. The inputs include eye movements, blink rate, posture, chip handling movements, and 'hand fidget' metrics. It then analyzes that data and the outcomes of each hand to predict the likelihood of a player holding a strong made hand, a drawing hand, or a bluff.
The Small-Data Problem
Despite the tournament drawing over 9,000 entries, Wired noted that the vast majority of players never appeared on the three tables that were recorded by cameras. Those same camera feeds, used for the broadcast, were also used to train the AI tool. Because even featured players spent limited time at those tables, the dataset was too small to cover the range of poker situations. The poker experts Wired spoke with expressed skepticism about the tool's effectiveness.
| Metric | Value |
|---|---|
| Tournament entries | Over 9,000 |
| Tables recorded by cameras | 3 |
| Top prize for the finalist | $10 million |
| Gagliano's poker experience | 17 years |
| Gagliano's final-table start | 8th chip position |
A Professional's Perspective
Michael Gagliano, a 17-year poker pro who made the Main Event final table and is playing for the $10 million top prize, told Wired that the varied streams mean "you don't get the same players too frequently." Gagliano, who started the final in eighth chip position, spent the two-and-a-half-week break after the final table was reached in mid-July reviewing every second of ESPN's live streams. He combed for tells or information on his remaining opponents, but the limited screen time constrained his analysis.
"I don't know how much actual information I'm going to be able to act on from what I saw." — Michael Gagliano, as quoted by Wired
Implications for AI Reliability
Wired also noted that most nonplayers' exposure to poker tells comes from the 1998 film Rounders, in which Matt Damon's character, Mike McDermott, folds a monster hand after spotting a tell in John Malkovich's Teddy KGB. The real-world debate, however, centers on whether a camera-based AI can outperform a trained human. For technology decision-makers evaluating AI tools, the poker controversy illustrates a core principle: a model trained on a small dataset may not generalize to the full range of real-world scenarios. The source reported that the AI tool was trained on only three tables' worth of footage, and even a 17-year professional found it difficult to extract usable information from those feeds — a cautionary data point for any organization deploying AI in complex, high-variability environments.
Sources:
