Managing Editor
ESPN’s coverage of this year’s World Series of Poker features a new gimmick no one asked for — AI-based analysis of players’ expressions and body language, which purports to determine whether or not they’re bluffing.
For game integrity reasons, the new technology will only be applied to players who have already been eliminated from the tournament. Even if it were applied to players during their live streams, there’s no clear need for it, since the “hole cam” (or its RFID successor) has been in effect since 2002.
Fans were quick to point this out on social media when the feature was announced.
Can’t we tell that they are bluffing by the fact that we know what cards they have
— Baseball Aslan (@BaseballAslan) July 8, 2026
The tool was developed by independent AI engineer Luke Geel, who describes himself as an avid poker player. He’s also adept at self-marketing, having written about his tell detector for Poker.org and landed a contract with Omaha Productions, the new producer for the televised WSOP this year.
If you ask me, however, this project highlights the biggest problem with the AI industry in general. It’s not that it hallucinates, or that it’s bad for the environment, or that it’s going to eliminate everyone’s jobs, or any of the other usual criticisms. I find most of those a little overblown.
It’s that no one actually seems to know what it’s for. AI has found a home here and there — for instance, in producing images to accompany op-eds like this one. However, in many cases, it is still a solution in search of a problem.
Supply-Side Economics as a Business Plan
This issue isn’t unique to the AI industry. It’s common across the tech sector. For instance, the entire wearables business seems predicated on the idea that because we can pack high-end electronics into a ring or a pair of sunglasses, there must be some reason someone would want that.
There’s been a popular meme among engineers since the late 1990s that follows the format:
- Step 1 and 2: [whatever it is your startup is doing]
- Step 3: ???
- Step 4: Profit!
It’s probably no coincidence that the technology industry first began to take on its current form during and immediately after the Reagan years. Famously or infamously, depending on your perspective, that era of politics marked the rise in popularity of supply-side economics.
Reagan and his economic advisors popularized the idea that industrial output was the primary driver of the economy, and that tax cuts for the wealthy and for large businesses were the means to stimulate that.
It was an inversion of the long-held economic belief that consumer demand and consequent spending are what make for a healthy economy. Whereas once, economic policy focused on making sure the populace had money to spend on goods, the emphasis became on ensuring that output continued to grow under the assumption that someone, somewhere, would generate the requisite demand to match the supply.
That logic, which I happen to disagree with, seems to have infested the entrepreneurial world, especially on its high-tech fringes. Once upon a time, the business world would have started with a real problem — housing shortages, the plague of ticks descending on the East Coast, or the dire state of bacon packaging — and gone about finding a way to address it profitably.
Now, instead of that, we get smart rings and poker tell detection AI, and businesses that spend a lot of their time and energy on Step 3 in that meme. That is, trying to figure out how to get people to pay for the things that they built just because they could.
Poker Tells Are Overrated
Like those video-enabled smart glasses, it’s easier to imagine undesirable uses of AI tell detection than desirable ones. In fact, those two things together would be worse than the sum of their parts. And, as some have pointed out, it’s the sort of thing that government agencies might get excited about if it works too well, and we all know where that leads.
That said, the charitable view of the AI tell detection is that its novelty value might be worthwhile for one season, at least. Whether it succeeds at its task or fails, viewers might enjoy getting an answer to the question of whether this sort of thing is possible.
Seems more like a silly fun thing that has zero meaning but rec players will think it’s cool. We already know if players are bluffing we can see their cards on stream
— sammie g (@thefakesammieg) July 8, 2026
Beyond that, it’s less clear what such technology would add to a broadcast, even if hole cards weren’t available.
Any poker player knows that the importance of tells to the game is vastly overstated in pop culture. A player adept at interpreting behavior will use tells as a “tie-breaker” of sorts when the situation is otherwise a close one. However, it always plays second fiddle to strategic fundamentals. Zachary Elwood, author of Reading Poker Tells and Verbal Poker Tells, repeatedly emphasizes in his books that a tell should never be your primary reason for making a decision.
In terms of applying AI to determine bluff frequency, the results for any competent player would be more accurate if using board texture and bet size as the inputs, rather than video data.
But that would put utility ahead of technological “wow factor.” And that’s not how the AI industry rolls.
