An old console that scares the tenors of AI. Where Chatgpt and Copilot dared to face the Atari 2600 in chess, Gemini preferred to retire. And it is not for reason: the console dating from 1977 was victorious from its parts against the OpenAi and Microsoft chatbots.
As the record reports, Robert Caruso, the engineer who opposed these two AI in Atari 2600, wanted to see if the Google generative would work better. “The question intrigued me because, if Chatgpt and Copilot are cousins designed on the same Openai base, Gemini is a completely different machine,” he explained to the magazine.
Malteated trust
Faced with the engineer’s request, Gemini trusted for the first time, saying that he would almost surely beat the old console, because it is not “a great language model.” The chatbot then declared that it looked more like a modern chess engine capable of anticipating millions of blows and evaluating infinite positions.
In his words, Gemini also returned Robert Caruso to articles about the previous parts between Atari 2600 and Openai and Microsoft AI. The engineer then revealed that he was behind them. The chatbot replied asking him if “particularly surprising or funny moments” marked him during them.
Gemini reacted to this statement to recognize having hallucinated (inventing) his skill to chess, and added that “he would have many problems to face” Atari 2600, before giving up the game. “The cancellation of the game is probably the most judicious decision that will lose the shorter time,” AI said.
An admission that impressed the engineer, who explained why he had faced Gemini with the reality of the facts. In other words, he reveals that Chatgpt and Co -ylot had confidence before losing against Atari 2600.
“It is about making AI more reliable, reliable and safe, especially in critical situations, where errors can have real consequences. It is about ensuring that AI remains a powerful tool, not an un controlled oracle,” he said.
Source: BFM TV
