Over the years, Elon Musk has given plenty of warnings about artificial intelligence and the harmful outcomes it could have.

These days, Musk is harnessing the power of AI to help build out the next phases of growth for his companies Space Exploration Technologies Corp (SPCX) and Tesla Inc. (TSLA).

Musk covered his latest thoughts on AI in a new interview.

Musk on AI

Walter Isaacson profiled Musk for an autobiography that was released in 2023. In a new interview with Isaacson on "The Builders," Musk shares his latest conflicted thoughts on AI.

Musk said there is a chance that AI can be negative and an existential threat to humanity.

"There's nothing I can do to stop it," Musk said.

The billionaire said when it comes to AI he has two choices: "be a participant or be a spectator."

Musk told Isaacson that the keys to a good AI are curiosity and truth-seeking, as forcing AI to believe two things that can't both be true can make AI insane.

"An insane AI model is not a safe one."

On the flip side, Musk said a curious AI "will want to foster humanity."

Musk was asked by Isaacson if we are in a simulation.

The billionaire isn't sure either way and said that he once had dinner with a physicist and computer scientist. When he asked them the question about the odds we are living in a simulation, one answered 100% and one answered zero percent.

Musk said he should answer 50% as a result.

The SpaceX CEO said sometimes he questions this very notion and wonders if the fact that he is working to send people to Mars and build out a multi-planetary society are signs of a simulation.