ChatGPT defeated by 45-year-old Atari

ChatGPT, one of the most remarkable tools in the field of artificial intelligence, recently made headlines with a highly unusual test. Robert Caruso, an engineer at Citrix, organized a chess match against ChatGPT using the Atari 2600 game console, which left its mark on the late 1970s, and its 1979 game "Video Chess." The winner of this nostalgic yet extraordinary encounter, conducted via an old emulator, was surprisingly the Atari.

During the game, which lasted approximately 90 minutes, the mistakes made by ChatGPT were striking. Basic errors such as confusing the rook with the bishop, ignoring pawn moves, and constantly mixing up the positions of the pieces showed that the artificial intelligence could not establish sufficient control over the game. Initially, ChatGPT attributed these errors to Atari's graphical icon system, but it could not overcome these mistakes even when standard chess notation began to be used.

Robert Caruso stated, "ChatGPT made so many mistakes that it would be the laughing stock of a 3rd-grade elementary school chess club," adding that the artificial intelligence wanted to reset the game many times throughout the match.

This experiment revealed the extent to which ChatGPT is limited in areas that require high computational power and knowledge of game mechanics, such as chess. After all, ChatGPT is not an engine specifically designed for chess like IBM's Deep Blue, which defeated chess masters; it is a language model focused on understanding and generating human language.

According to Caruso, actively playing chess with ChatGPT is not very logical. However, it is quite capable of engaging in theoretical discussions about chess, conducting strategy analyses, or discussing the historical context of moves. The experiment serves as a reminder once again that technological advancements do not bring absolute superiority in every field.