May 31 - During an appearance at the All-In-Live event in Miami, Google co-founder Sergey Brin (Sergey Brin) Throwing out new ideas.Sometimes terrorizing AI (AI) model, instead, stimulates the enhancement of its performance.
Brin says this is not just the case with Google's AI models, but with most of the AI models on the market today.
1AI translates Brin's remarks as follows, "I'm not quite sure how many people within the AI community are aware of this.Based on our tests, however, almost all models perform better after being threatened”.
This statement overturns the traditional perception of AI interactions. Many users are accustomed to including polite phrases such as "please" and "thank you" in their prompts, and Brin's comments certainly challenge that practice.
OpenAI CEO Sam Altman has previously quipped that it costs tens of millions of dollars in electricity to process polite words such as "please" and "thank you".