chat GPT was defeated in a game of chess by an emulation of an Atari 2600 console, which is nearly half a century old, when it accepted the challenge.
According to Citrix engineer Robert Jr. Caruso, the face-off initially started as a casual test. He mentioned that the advanced language model on its own decided to engage in a game of Video Chess, aiming to determine how swiftly it could outperform the game.
In 1977, the Atari 2600 made its debut. Two years after that, the game called Video Chess became available on this console. With it, players could engage in games against a computer-managed adversary whose skill level could be adjusted.
Caruso remarked, “ChatGPT was thoroughly outplayed at the beginner level.” He further explained that this occurred following our discussion about the history of AI in chess, which prompted it to offer to play Atari Chess. It was curious to see how swiftly it could conquer a game that can only strategize one or two moves ahead on a 1.19 MHz CPU.
In more depth, Caruso explained that ChatGPT struggled to consistently grasp the purpose and boundaries of each chess piece, often forgetting about earlier moves played on the board.
For approximately 90 minutes, I found myself intervening frequently to prevent it from executing poor decisions and enhancing its understanding of the game’s layout on numerous occasions during each turn, Caruso explained.
A fair game?
As a devoted admirer, I’d like to share that while ChatGPT excels at text processing and conversational responses, it isn’t specifically designed to be a chess grandmaster. However, should you ask it if it can play chess, it will happily oblige! In fact, it offers the unique opportunity for users to engage in a game of chess directly within the app itself.
For skill level, the AI lists the following limitations.
- Not infallible: May blunder occasionally, especially in complex positions.
- No deep calculation: Doesn’t evaluate millions of positions like an engine does.
- No rating: Hard to give an exact Elo, but likely plays at a beginner to intermediate club level, depending on the position and version you’re using.
Assessing the ability of Video Chess’ AI can be quite challenging due to its intricacy. Given the specifications of the Atari 2600, it’s clear that it doesn’t possess the capacity for sophisticated simulations.
Essentially, Caruso’s study is quite similar to trying to compare apples with oranges, but it serves as a valuable lesson: ChatGPT may not be omniscient and should primarily be utilized according to its designated function.
Read More
2025-06-11 17:19