ESPN introduced an AI-based 'tells detection' tool during the 2026 World Series of Poker Main Event, aiming to analyze player behavior through visual and audio cues. The tool displayed live metrics on player movements and hand strength predictions, sparking debate among poker professionals about its effectiveness and relevance to the game's human elements. The system was trained on footage from the tournament, which drew over 9,000 entries, but the data collected was limited due to the short time many players spent on camera. This raised concerns about the tool's ability to accurately interpret complex human behavior in a high-stakes environment. Source: wired
The AI tool, developed by Luke Geel, an AI engineer for the US Air Force, analyzed data such as eye movements, blinking rates, posture, and chip handling to predict the likelihood of different hand types. However, poker experts like Michael Gagliano, a 17-year professional, expressed skepticism about the tool’s effectiveness, noting that the small dataset made it difficult to build a robust model for the diverse situations in poker. Gagliano, who made the Main Event final table, said he reviewed all available footage but found limited actionable information due to the lack of consistent screen time for any single player. Source: wired
Poker professionals argue that the nuances of tells extend beyond what a camera can capture, including verbal and physiological cues that are hard to quantify. Shaun Deeb, a two-time WSOP Player of the Year, emphasized that physical tells are far more complex than the AI tool could handle. He noted that factors like breathing patterns and pulse could reveal more about a player’s strategy than visual data alone. Despite these limitations, the tool’s creators admit it is not intended to be a perfect oracle, and they are exploring ways to improve its performance with larger datasets. Source: wired
Source: wired
