Dec 29, 2021 07:34 AM
(This post was last modified: Dec 29, 2021 08:38 AM by Yazata.)
This is a video where Lex Fridman, a very smart AI guy from MIT, talks with Elon (a certifiable genius) for 2 1/2 hours about any topic you can imagine. Perhaps the best Elon interview that anyone has ever done.
They start with SpaceX, the DM-2 mission and then about Starship, and Elon drops some information nuggets. (He expects to reach Mars in 5 years best case, to 10 years worst case.) Elon sounds more excited about Moon bases than he's seemed in the past. (Maybe the HLS influence.)
Then they switch to Tesla and get deep into the technical details of the AI of self-driving cars. (That's Fridman's thing.) Lots of stuff about machine perception and data representation in layers upon layers of neural networks. Elon talks about Tesla's AI team (some of whom Fridman knows and have lectured at MIT) and how they implement their theoretical ideas in computers and software. Fascinating stuff about how the human brain handles these tasks and about how neural networks are taught. (They are taught rather than programmed.) Some of it was above my head. Elon reveals that he expects Tesla self-driving AI to surpass human drivers in accidents per mile next year. (Though it will have to be several times as good, before government regulators make self-driving cars legal.) Then they moved from automobile AI to the recently announced Teslabots. These humanoid robots are a natural evolution of the automobile AI, will use similar machine perception and neural networks, and should step out in about two years. It became very clear that Tesla isn't just an automobile (and battery) company, they aspire to be the world's leader in AI and robotics, and already have some of the world's most advanced AI technology. Eventually Elon expects there to be more Teslabots than Tesla cars. Then they discussed the impact of robots on society more generally. Robots replacing human workers (Elon foresees work becoming optional and endorses universal basic income). They had a fascinating discussion of robots developing enough unique personality to be dog-like friends and companions to lonely humans as they learn and adapt to their owners' idiosyncracies.
So while we are all watching Elon make science fiction real with his Boca Chica rockets (FAA willing), Tesla is doing their own less-visible sci-fi things.
Elon discussed the nature and function of money when Fridman asked him about cryptocurrencies. And he explained why he likes Dogecoin, even though it started as a joke. (It does important things right, entirely by accident.) Politics came up several times, though always in the abstract with no discussion of divisive American partisan politics. (Elon makes clear that he wants to be a uniter by inspiring people about future possibilities, as opposed to those who use fear to divide people.) He emphasized that reality isn't a zero-sum game where one person wins by making others lose. The goal should be to grow the pie, so to speak. Then they discussed history (and the folly of dismissively applying our own moral standards today to people of the past, who might have been very good by the standards of their day). He endorsed nuclear power (and argued that fear of radiation can often be irrational).
They started to wrap it up by discussing humor and the possibility that Elon might try doing standup and ended with Elon on love and the meaning of life, and with his advice to young people.
I was very pleased to find that I agree with Elon on most of what he said. (Irrefutable proof that Elon is a genius!)
A few things Elon said that caught my interest:
Physics is the Law. Everything else is just recommendations.
Mars is the least inhospitable planet in the solar system besides Earth, but it's still a fixer-upper.
Money is a database for resource allocation.
Friendly competition is great. Governments are slow. The only thing slower is a collection of governments.
Rules and regulations never die, they just add up as time goes on until freedom disappears and society is slowed to a halt. The silver lining of war is that oftentimes it results in a reset.
Everytime he accepts cookies on his computer, he worries about the slight chance that he's opened a portal to hell.
https://www.youtube-nocookie.com/embed/DxREm3s1scA
They start with SpaceX, the DM-2 mission and then about Starship, and Elon drops some information nuggets. (He expects to reach Mars in 5 years best case, to 10 years worst case.) Elon sounds more excited about Moon bases than he's seemed in the past. (Maybe the HLS influence.)
Then they switch to Tesla and get deep into the technical details of the AI of self-driving cars. (That's Fridman's thing.) Lots of stuff about machine perception and data representation in layers upon layers of neural networks. Elon talks about Tesla's AI team (some of whom Fridman knows and have lectured at MIT) and how they implement their theoretical ideas in computers and software. Fascinating stuff about how the human brain handles these tasks and about how neural networks are taught. (They are taught rather than programmed.) Some of it was above my head. Elon reveals that he expects Tesla self-driving AI to surpass human drivers in accidents per mile next year. (Though it will have to be several times as good, before government regulators make self-driving cars legal.) Then they moved from automobile AI to the recently announced Teslabots. These humanoid robots are a natural evolution of the automobile AI, will use similar machine perception and neural networks, and should step out in about two years. It became very clear that Tesla isn't just an automobile (and battery) company, they aspire to be the world's leader in AI and robotics, and already have some of the world's most advanced AI technology. Eventually Elon expects there to be more Teslabots than Tesla cars. Then they discussed the impact of robots on society more generally. Robots replacing human workers (Elon foresees work becoming optional and endorses universal basic income). They had a fascinating discussion of robots developing enough unique personality to be dog-like friends and companions to lonely humans as they learn and adapt to their owners' idiosyncracies.
So while we are all watching Elon make science fiction real with his Boca Chica rockets (FAA willing), Tesla is doing their own less-visible sci-fi things.
Elon discussed the nature and function of money when Fridman asked him about cryptocurrencies. And he explained why he likes Dogecoin, even though it started as a joke. (It does important things right, entirely by accident.) Politics came up several times, though always in the abstract with no discussion of divisive American partisan politics. (Elon makes clear that he wants to be a uniter by inspiring people about future possibilities, as opposed to those who use fear to divide people.) He emphasized that reality isn't a zero-sum game where one person wins by making others lose. The goal should be to grow the pie, so to speak. Then they discussed history (and the folly of dismissively applying our own moral standards today to people of the past, who might have been very good by the standards of their day). He endorsed nuclear power (and argued that fear of radiation can often be irrational).
They started to wrap it up by discussing humor and the possibility that Elon might try doing standup and ended with Elon on love and the meaning of life, and with his advice to young people.
I was very pleased to find that I agree with Elon on most of what he said. (Irrefutable proof that Elon is a genius!)
A few things Elon said that caught my interest:
Physics is the Law. Everything else is just recommendations.
Mars is the least inhospitable planet in the solar system besides Earth, but it's still a fixer-upper.
Money is a database for resource allocation.
Friendly competition is great. Governments are slow. The only thing slower is a collection of governments.
Rules and regulations never die, they just add up as time goes on until freedom disappears and society is slowed to a halt. The silver lining of war is that oftentimes it results in a reset.
Everytime he accepts cookies on his computer, he worries about the slight chance that he's opened a portal to hell.