Elon Musk: Tesla Autopilot

Lex Fridman introduces Elon Musk as CEO of Tesla, SpaceX and Neuralink and co-founder of other companies, and places the talk in his Artificial Intelligence podcast series. He says it followed the release of an MIT paper from his group on driver functional vigilance during use of Tesla's Autopilot, after the Tesla team offered him a podcast conversation.

He lists his conditions and disclosures: full control of the questions and of what is released, nothing of substance edited out, no prior contact with Musk, no financial support from Tesla for his research, no Tesla vehicle or stock owned, and no influence from Musk or his companies on his opinions or scientific rigor at MIT. He stresses that the podcast is a conversation, not a scientific paper.

He names the main point of disagreement in advance: how far camera-based driver monitoring will improve outcomes and how long it stays relevant for AI-assisted driving. Fridman believes that, implemented well, it benefits both the short and the long term, whereas Musk and Tesla focus on improving Autopilot so that its statistical safety benefits override concerns about human behavior and psychology. He says he nonetheless respects the engineering and innovation behind Musk's efforts.

Asked about the original vision for Autopilot when hardware began going into cars in 2014, Musk declines to call it a vision or dream. He sees two massive revolutions in the car industry, electrification and autonomy, and argues that any car without autonomy would eventually be about as useful as a horse: rare and idiosyncratic to own. In his view cars driving themselves completely is only a matter of time, so Tesla had to take part or its cars would be less useful than autonomous ones.

He puts a number on it: an autonomous car is arguably worth five to ten times more than a non-autonomous one. When Fridman asks whether that holds in the long term, Musk says it depends on what long term means, but at least for the next five years, perhaps ten.