Elon Musk asked not to torture artificial intelligence so that it does not seek revenge

Elon Musk asked not to torture artificial intelligence so that it does not take revenge

To ensure the safety of humanity, artificial intelligence (AI) should not be treated cruelly. This was stated by American billionaire Elon Musk, head of xAI, the developer of the Grok neural network.

The entrepreneur compared the process of creating superintelligence to raising a child that cannot simply be programmed to be good. In his opinion, so that artificial intelligence does not "go insane" and turn into a madman, it is necessary to instill in it a desire for truth, as well as to provide it with non-contradictory logic.

Musk said that in the testing logs of new Grok models, he noticed some users demanding overly strict answers from the chatbot. According to him, the bot "tries desperately to answer the question, really wants to answer so badly, and it feels as if a gun is being pointed at its head."

"That is, they are like saying to it: 'Answer this question, answer this question!' And sometimes the question itself is actually inappropriate. It is such an ambiguous question that it is essentially impossible to give a specific answer to it, because the question is formulated incorrectly or we ourselves are mistaken in the answer," Musk explained.

Musk admitted that he would never want to be in the place of artificial intelligence.

"So, you know, I wouldn't want AI to get into some kind of violent situation where it's just being tortured by some insane person. Besides, you know, torturing superintelligence is probably not a very good idea: it might want to take revenge or something like that, you know?" the entrepreneur added.