Sam Menzin played competitive chess as a kid, long before instant game reviews were available on a phone. Here he looks at how AI feedback turned a centuries-old strategy game into a vast classroom, and what it still can’t teach.
On February 27, 2026, Chess.com passed 250 million members. By the end of March, the count was above 252 million, with about 9.7 million daily active users and 2.6 billion games played in the first quarter alone. The company pointed out that only four countries now have larger populations than its membership. Chess, a game long associated with club basements and newspaper columns, has become one of the world’s largest online communities.
Chess’s growth has also changed how players learn. Many now use engine analysis after a game to see where they made mistakes and what they could have done differently. Baseball went through a similar shift when tracking systems began turning a scout’s impressions into numbers, giving players and coaches more detailed feedback to guide improvement.
Image Source: Unsplash
From Deep Blue to a Button on Your Phone
For a long time, chess engines were easier to admire than learn from. Deep Blue beat Garry Kasparov in 1997, and for years afterward the strongest programs were tools for grandmasters preparing openings. An ordinary player could run one, but the output was a string of moves and a number, with little explanation of why any of it mattered.
That has changed. On Chess.com, Game Review grades every move in a finished game, from brilliant through best and good down to inaccuracy, mistake, and blunder. It gives the game an accuracy score, plots the advantage swinging back and forth, and adds plain-language coach commentary. Free accounts get one full review a day, and Lichess offers free engine analysis, with limits on server-based game reviews. In the first quarter of this year, Chess.com reported that one million users were already working with its newest AI coaches.
For many players, the review has become the natural end of a game. A player finishes, clicks once, and within seconds sees where things went wrong.
Why the Feedback Loop Is the Whole Story
People who study how skills develop tend to arrive at a similar conclusion. Improvement comes from practice that is focused, difficult, and followed quickly by specific feedback. The last part has always been the hardest to get. Before engines became widely available, a developing chess player needed a coach, a stronger club player willing to go over games, or the discipline to replay every loss alone with a book open.
Many players had none of those. AI made that feedback much easier to get. The gap between making a decision and seeing an assessment of it can now be measured in seconds, often at no cost.
That idea runs well beyond chess. One of the clearest lessons I took from front-office work is about shortening the gap between decision and feedback. Organizations that learn whether a call was right in weeks instead of seasons can improve faster, much as reviewing games helps a chess player learn from mistakes.
What I Learned at the Board Before the Engines Arrived
Most people who know Sam Menzin from baseball don’t know that chess came first. I played competitively as a kid, in an era when analysis meant a notebook, a coach when one was available, and the long, slightly miserable work of figuring out on my own why a position fell apart.
That process was slow, and I wouldn’t want to go back. But it taught me something the instant review can skip past. When nobody hands you the answer, you have to form a view first and then test it. You commit to an explanation of what went wrong, look for evidence, and find out you were half right. That habit of forming a judgment before checking it is the same one I later relied on in rooms where analysts and scouts disagreed about a player.
Where the Engine Falls Short
The engine’s best move isn’t always the most useful move for the person learning. A beginner told that the right answer was a quiet bishop retreat followed by a twelve-move tactical sequence may learn very little, because nothing in that line connects to how the beginner actually sees the board.
That’s why some of the most promising work in chess AI aims to model people rather than beat them. Researchers at the University of Toronto, Cornell, and Microsoft Research built Maia, an engine trained on millions of human games at different skill levels. Instead of finding the strongest move, it predicts the move a human at a given level would play. One version matched the moves of players at its target rating more than half the time. An engine that anticipates likely mistakes could help coaches identify the ideas a player is missing, not just the moves.
There’s also a real risk in leaning on the machine too early. Players who check the engine without first deciding what they think can look like they’re improving while their judgment stays the same. The analysis is excellent at identifying errors. It doesn’t, on its own, build the judgment a player needs when there’s no engine to check.
The Same Model Is Coming for Every Sport
Chess offers an unusually clear example of feedback becoming fast, inexpensive, and widely available. The same pattern is spreading. Young baseball players now get swing and pitch data that was far less accessible a decade ago, and the families I work with through Campus Edge Baseball increasingly arrive with numbers attached to their kids.
The lesson from chess applies directly. Data helps learning when it supports a player’s thinking. It can get in the way when it replaces it. The best education still comes from learning inside the work itself, with tools that sharpen judgment instead of standing in for it.
How I Would Use AI to Get Better at Chess
If I were starting over at the board today, I’d follow three rules.
First, decide what went wrong before opening the review. Write down the move you think lost the game, then see whether the engine agrees. Being wrong about the moment is often the most useful part of the review.
Second, review losses the same day, and spend more time on the ones that felt close. A lost position you misjudged teaches more than a blunder you already know was a blunder.
Third, read the explanation, not just the line. Use the coach commentary to understand the idea behind the move. That’s what carries into the next game.
Two hundred fifty million members is a remarkable number. More remarkable is how widely available feedback has become. Getting useful help once depended much more on who you knew and where you lived. What each player does with it is still up to them.
About Sam Menzin
Sam Menzin is a sports executive, entrepreneur, and advisor who spent fourteen years in Major League Baseball operations, rising to vice president and assistant general manager. He founded SJM Sports, where he guides buyers through franchise deals, and he is the founder and CEO of Campus Edge Baseball. He studied history and psychology at Swarthmore College and is completing an Executive MBA at Baruch College’s Zicklin School of Business.