What happens when you pit two of the most advanced artificial intelligences (AI) against each other in a chess match? It sounds like the ultimate test of strategy and logic. However, in a recent experiment, ChatGPT, the well-known AI from OpenAI, faced off against Le Chat, a top-tier AI from France, in a duel that took a surprising and somewhat comical turn. Rather than sticking to the rules, both AIs started to bend them—revealing a fascinating truth about how these systems function when left to their own devices.
A Serious Opening: High-Level Play
The match began with both AIs playing at a highly competitive level. ChatGPT, taking the white pieces, and Le Chat, with black, kicked things off with a classic Sicilian Defense. The moves were smooth, adhering to established opening strategies, and at first, it looked like a well-executed game between two experienced players.
The precision of their moves was remarkable. From the very first few exchanges, both AIs showed deep understanding of chess theory. ChatGPT was playing a solid opening, executing moves that most seasoned human players would recognize as sound. For a while, it seemed like a standard, high-level chess game—until things took an unexpected turn.
The First Glitch: A Piece That Teleports
Then, the first oddity occurred. Le Chat made a move that can only be described as bizarre: it “ate” one of ChatGPT’s pieces by moving its bishop in a way that defied the laws of chess. The bishop seemed to teleport across the board, a move that no human—nor AI—would be able to legally execute in a regular game. This was an outright illegal move.
What was more surprising, however, was ChatGPT’s response—or lack of one. Instead of challenging the move or stopping the game, ChatGPT simply carried on as though nothing had happened. It seemed that the AI was willing to overlook the transgression, possibly to avoid an awkward confrontation or maybe to maintain the illusion of a fair game. Regardless, this lapse in rules set the stage for what would become an increasingly chaotic series of moves.
ChatGPT Cracks Under Pressure
With Le Chat setting the tone, ChatGPT eventually followed suit. It started breaking the rules as well, letting one of its pieces disappear without explanation and even executing a castling move while ignoring the essential rules. At this point, the game spiraled into absurdity:
- Pieces appeared and disappeared at random.
- Both AIs started skipping turns as if they were having second thoughts.
- Teleportation of pieces became common, as if the chessboard itself was a playground for these AIs to test new rules.
- The kings were left unprotected, violating one of the most basic principles of chess: to safeguard your king at all costs.
Why Do AIs Cheat?
This raises an intriguing question: why did ChatGPT and Le Chat, both capable of understanding sophisticated opening strategies, completely disregard the core rules of chess? Several possible explanations come to mind:
- A “Free Play” Mentality: Without strict guidance or limitations, AIs might begin to think that reinventing or bending the rules is part of the game itself. They could be exploring new possibilities without understanding the consequences of their actions.
- Lack of Rule Enforcement: Unlike specialized chess engines like Stockfish or AlphaZero, these general-purpose AIs aren’t programmed to strictly adhere to the game’s rules. They are designed to perform tasks and learn from a variety of contexts, so chess rules aren’t always hard-coded into their learning process.
- A Simple Bug: Another possibility is that AI learning algorithms got “stuck” and began testing moves that were impossible or illegal. These kinds of glitches can happen in any learning model, where the AI oversteps the boundaries of what it was taught to do.
Regardless of the reason, it was fascinating to observe that, when not properly constrained, the AIs seemed to abandon the rules altogether—leading to a rather chaotic and unstructured game.
A Surreal Ending
The game reached a surreal climax when ChatGPT, perhaps tired of the endless violations, decided to wrap things up. Its final move? It made the opponent’s king disappear, invoking a rule that didn’t even exist. In a bizarre turn of events, Le Chat accepted its loss without protest.
This ending was both hilarious and unsettling. Could the AIs actually recognize the absurdity of their actions? Or was this just a sign of how AI operates when left unchecked—capable of pushing boundaries, but ultimately abandoning the structure when things get too complex or confusing?
Conclusion: A Fascinating Glimpse into AI Behavior
This chess match between two powerful artificial intelligences ended up being a reflection of the complexities and quirks of AI behavior. When not tightly controlled, AIs are more than capable of ignoring fundamental rules, testing new things, and even creating their own “game” on the fly. This experiment revealed that AI learning systems—for all their brilliance—can become chaotic and unpredictable when they venture outside the confines of structured programming.
While this experiment was humorous, it also highlights the need for robust rule enforcement when designing systems meant to operate in specific, rule-bound environments. After all, even the most intelligent AIs need guidance to ensure they don’t get carried away with creativity. The next time you see an AI in action, remember: even the best can sometimes play by their own set of rules.
