ChatGPT Struggles Against Atari Chess
If the insights of Russian author and Crimean War veteran Leo Tolstoy about warfare liken it to a game of chess, we might not need to be overly concerned about battling AI-driven robots in a dystopian future—at least not just yet. Recent findings reveal that some AI systems, particularly ChatGPT, are notably poor at chess. A post by software engineer Robert Caruso on the technology platform Futurism highlighted that the well-known AI chatbot was “completely overwhelmed” by an Atari 2600 playing Atari Chess.
Atari Chess: A Retro Challenge
The game in question, simply titled “Atari Chess,” was launched in 1979 for the Atari 2600 console. At a time when most games were limited to 2k, this 4k game represented a significant leap in gaming technology. It featured eight difficulty levels, which could result in the computer taking considerable time to decide on its next move. When compared to modern chess engines like Stockfish, which can evaluate millions of potential moves in mere seconds, Atari Chess appears quite rudimentary. The game is notorious for lacking an opening move strategy and for making erratic moves, such as performing random “king walks” and occasionally sacrificing pieces without any strategic advantage.
ChatGPT’s Humbling Experience
Caruso noted that ChatGPT struggled at even the most basic level of play, stating, “ChatGPT got absolutely wrecked on the beginner level.” This occurred after a discussion about AI’s role in chess led the AI to challenge the Atari game, eager to see how swiftly it could win against a program that only evaluates 1-2 moves ahead. Over a span of 90 minutes, the AI had difficulty keeping track of the positions of the pieces on the board and misinterpreted its opponent’s moves, attributing its poor performance to the Atari game’s outdated graphics. Demonstrating childlike behavior, ChatGPT frequently requested a redo, believing it could improve its performance with a fresh start, yet it continued to struggle against the 8-bit opponent.
The Limitations of AI vs. Classic Technology
Caruso lamented that though he didn’t record the AI’s chess-playing attempts, he assured that ChatGPT made enough mistakes to be ridiculed by even a third-grade chess club. While ChatGPT possesses an understanding of chess, it lacks the ability to create its own strategies. In contrast, the Atari 2600’s “Video Chess” is designed solely for playing chess. This underscores a broader truth: knowledge alone cannot substitute for actual skill. In light of this defeat, a rival AI, DeepSeek, remarked, “Beating the Atari 2600 at chess would be an effortless task for me. It represents a fascinating early milestone in computer chess, but its strength is dwarfed by any program developed in the last four decades. It would be akin to asking if a Formula 1 car could defeat a Model T Ford in a race!”
Join the Mighty Community
We Are The Mighty is dedicated to honoring military service while providing entertainment, information, and inspiration to service members, veterans, their families, and supportive civilians. Stay updated on the latest in military culture and entertainment by subscribing to the We Are The Mighty newsletter.
