Elon Musk: Artificial Intelligence Will Be a Danger to Humanity if It Is Trained to Lie

Lex Fridman, author of one of the most popular tech vlogs on YouTube, interviewed Elon Musk.

His podcast has been viewed over half a billion times. Friedman is an expert in artificial intelligence and computer technology.

Friedman asked Musk whether artificial intelligence poses a threat to humanity.

“I think the risk of a ‘machine uprising’ is less than 20 percent,” Musk replied. “In my opinion, the most important thing in training AI is to stick to the truth, whether that truth is politically correct or not. For example, if you ask Gemini [Google’s chatbot] to show you a picture of the founding fathers of the United States, it will show you a variety of women. Yes, there are considerations of equality and liberation, but in reality that is not true. If you force AI to lie, you create a big problem, even if the lie is done with good intentions. I think programming AI to deviate from the truth is very dangerous, because when the neural network becomes very intelligent, it can destroy everything that does not fit the ideas of gender diversity, racial diversity, political correctness, and create a world that never existed.”