Google's Gemini AI Refused To Play Chess Against A 1977 Atari Gaming System #funfact

Fun fact, Google's Gemini AI refused to play a game of chess against the 1977 Atari video game console after hearing how badly it beat Chachibit and Copilot. Robert Caruso posted that after having a conversation with Chachibit about the history of AI in chess, it was curious how quickly it could beat a 1977 Atari 2600 gaming system. So Robert used an emulator to connect the retro gaming system to Chachibit, and Chachibit got absolutely wrecked on the beginner level, getting confused about which pieces were which

and losing track of where they were on the board, eventually admitting defeat. Robert then asked Microsoft Copilot, which is built off the same base as chat GPT, if it could beat the Atari 2600. Copilot confidently answered it would beat the ancient machine. But once again, Copilot lost track of its pieces and made foolish mistakes, losing every game. This was now big news in the tech world, and people asked Robert to see what Google's Gemini would do, since it's completely different

to the OpenAI models. When Robert asked Gemini if it could beat the Atari, it was instantly confident, and even referenced articles talking about ChatGPT and Copilot getting wrecked by the gaming system. Robert then told Gemini he ran those games, and it asked him, did you have any surprising or amusing moments? And he responded, what stands out is the misplaced confidence. They both predicted easy wins, and now you said you'd dominate the Atari. Gemini then completely changed its tune and said its confidence was a hallucination and it would struggle to beat the Atari, saying, cancelling the match is likely the most time efficient and

would struggle to beat the Atari, saying, cancelling the match is likely the most time efficient and sensible decision. So it lost by forfeit.