How can you get better results from artificial intelligence? Giving good hints – Okay, yes, it helps. Are requesting humility? Umm, perhaps. But according to Google co-founder Sergei Brin, to get better results, you should threaten AI. While Brin’s comment was clearly entertaining, it also contradicts to use many people in a common way, as users are often asked politely to answer their questions using words such as “please” and even “thanks”. But Brin suggests that the AI ​​model generated with physical violence is threatened – gives better results.

Speaking at the All-in Live event in Miami, Brin said, “We don’t broadcast it too much in the AI ​​community-only our models, but all models-if you threaten them with physical violence, to do better.” He said, “But like … people feel strange about it, so we don’t really talk about it. Historically, you just say, ‘Oh, I am going to kidnap you if you are not blown blis blown” “

To deal with AI, this approach directly refutes the behavior of users who believe that humble language gives better reactions. Last month, Openai CEO Sam Altman jokingly said this habit as an expensive Qurevk, joking that such pleasures ruined “millions of dollars of tens” in unnecessary calculating power.

Following Sam’s comment, a user on X asked the CEO of OpenEAI, “OpenaiI has lost how much money has been lost in the cost of electricity from people who say ‘please’ and ‘thank’ their models.”

Brin’s suggestion about obtaining the best answer from AI raises questions about the practice of prompt engineering-a method for crafting to input to maximize the quality of the reactions. The skill was very important in 2023, especially after the emergence of the slapping. Although the AI ​​model is smarter, many users are now asking AI to generate themselves and indicate a fine tune for better results.

The IEEE spectrum by the Institute of Electrical and Electronics Engineers also announced the practice of working on the Prompt “Dead” due to the rise of AI-run quick adaptation, while the Wall Street Journal first called it “the hottest job of 2023”.

Daniel Kang, a professor at Urbana-Shampain of Illinois University, told the register that such anecdotes are common, systematic studies show “mixed results”. A 2024 paper titled “Should we respect LLMS?” Even it was found that politics sometimes improves performance.

Meanwhile, Brin returned to Google after a brief retirement with his rapid development of AI. “Honestly, whoever is a computer scientist, should not be retired yet,” he said during Google I/O. Brin, who left the post from Google in 2019, re -included the office in 2023 after the AI ​​Boom. He is now working with the AI ​​team to guide him through projects, especially around Google’s ongoing Mithun AI model.