Person
20 episodes · 41 quoted claims
Lex Fridman says Elon Musk's xAI built the Colossus supercomputer in Memphis in about four months, and that it had reached 200,000 GPUs.
“Colossus supercomputer, probably in record time, in just 4 months. It's now at 200,000 GPUs and growing very quickly.”
Lex Fridman cites Nvidia's Vera Rubin pod as containing 7 chip types, 5 rack types, 40 racks, 1.2 quadrillion transistors and nearly 20,000 Nvidia dies.
“we're talking about 7 chips, 7 chip types, 5 purpose-built rack types, 40 racks, 1.2 quadrillion transistors, nearly 20,000 Nvidia dies”
Lex Fridman describes Nvidia's Nemotron 3 Super as a 120-billion-parameter open-weight mixture-of-experts model.
“Nemotron 3 Super, which you can also use inside Perplexity to look stuff up now, which is, uh, 120 billion parameter open weight, uh, MoE model.”
Lex Fridman asked about a story that Morris Chang offered Jensen Huang the chief executive role at TSMC in 2013; Huang confirmed the story and said he declined.
“in 2013, the founders of TSMC Morris Chang offered you the chance to become TSMC's chief executive”
Weatherford says Genghis Khan's father Yesuke named his son Temujin after a Tatar he had just killed, whose name meant 'person of iron'.
“He killed two Tatars. One of them was named Temujin Uge, which is sort of person of iron is what it means from the Turkic.”
Lex Fridman cites a 2003 paper, 'The Genetic Legacy of the Mongols', estimating that 0.5% of the world's men descend from Genghis Khan.
“So, that paper has a finding that estimates that 0.5% of the world's male population is descendant direct descendants of Genghis Khan.”
Lex Fridman says he has held a paid research position at MIT since 2015.
“I have been at MIT for over ten years in a paid research position from 2015 to today.”
The Clay Mathematics Institute lists seven Millennium Prize problems, and solving any one carries a million-dollar prize.
“The Clay Foundation has these seven Millennium Prize problems, has a million dollar prize for solving one of these problems.”
Of the seven Millennium Prize problems, only the Poincaré conjecture has been solved, by Grigori Perelman.
“Of these seven, only one of them has been solved. Uh, the Poincare conjecture, by Perelman.”
The Kakeya needle problem originated as a puzzle posed by the Japanese mathematician Soichi Kakeya around 1918.
“it came from a little puzzle by the Japanese mathematician Soichi Kakeya, uh, in like 1918 or so.”
Terence Tao published a paper on finite-time blowup for an averaged three-dimensional Navier-Stokes equation in 2016.
“In 2016, you published a paper finite time blow up for an averaged three dimensional Navier Stokes equation.”
Grigori Perelman proved the Poincaré conjecture after working largely alone for about seven years.
“who proved the Poincare Conjecture and did so by working alone for seven years, with basically little contact with the outside world.”
Grigori Perelman famously declined both the Fields Medal and the Millennium Prize, which came with one million dollars.
“Famously declining the Fields Medal and the Millennium Prize, which came with 1 million dollar of prize money.”
Dario Amodei says he joined the AI field in late 2014, working at Baidu with Andrew Ng on speech recognition.
“I first joined the AI world when I was, uh, working at Baidu with Andrew Ng in late 2014, which is almost exactly ten years ago now.”
Dario Amodei says that seeing GPT-1's results in 2017 convinced him language was the domain where scaling could work.
“I think it wasn't until 2017 when I first saw the results from GPT one that it clicked for me that language is probably the area in which we can do this.”
Hancock says the earliest known anatomically modern human skeletal remains come from Jebel Irhoud in Morocco and are about 310,000 years old.
“the earliest anatomically modern human skeletal remains are from Jebel Irhoud in Morocco and date to about 310,000 years ago”
Hancock says the Indus Valley civilization was unknown until the 1920s, when railway workers stumbled across its ruins.
“the Indus Valley civilization was a lost civilization, uh, until the nineteen twenties, when, uh, railway workers accidentally stumbled across some some ruins”
Lex Fridman said he admires historian William Shirer, author of The Rise and Fall of the Third Reich, as a model for his interviews.
“I admire William Shirer, the author of many books on Hitler, including The Rise and Fall of the Third Reich.”
Pavel Durov is the CEO of Telegram, a messaging app that offers an optional end-to-end encryption mode.
“Pavel Durov is CEO of Telegram, which is a messenger app that has end to end encryption mode.”
Pavel Durov was arrested in France in 2024 on charges related to criminal activity on Telegram.
“Pavel was arrested in France on a long list of charges related to quote unquote criminal activity carried out on the Telegram platform”
Lex Fridman said Brazilian Supreme Court justice Alexandre de Moraes ordered X to block accounts and threatened to arrest its local representatives when X refused.
“He ordered X to block certain accounts that are spreading quote unquote misinformation. Elon and X denied the request, then de Marías threatened to arrest X representatives in Brazil”
Lex Fridman said he recently moved from the VS Code editor to Cursor, an LLM-based editor, having earlier moved from Emacs to VS Code.
“I recently switched from VS Code to Cursor, and before that it was Emacs to VS Code switch.”
Lex Fridman said he earned his bachelor's, master's and doctorate degrees at Drexel University.
“I went to Drexel University for my bachelor's, master's, and doctorate degrees, and I took a while”
Lex Fridman said he trained and competed in judo and jujitsu through his twenties and earned a black belt.
“I trained and competed in judo and jujitsu for my entire twenties, got a, uh, black belt from it.”
Roman Yampolskiy argues that artificial general intelligence will almost certainly eventually destroy human civilization.
“He argues that there's almost one hundred chance that AGI will eventually destroy human civilization.”
Lex Fridman says engineers building state-of-the-art AI systems generally put the probability of AGI killing all humans at about 1 to 20 percent.
“I would say those folks put the infamous P doom or the probability of AGI killing all humans at around one to 20%.”
Yann LeCun, as quoted by Lex Fridman, argues that open research and open source are the best ways to understand and mitigate AI risks.
“open research and open source are the best ways to understand and mitigate the risks, and two, AI is not something that just happens.”
Lex Fridman recommends Pessimist Archive, a channel that documents historical fear-mongering about new technologies.
“There's a there's a really nice channel that I recommend people check out, Pessimist Archive, that documents all the fear mongering about technology that's happened throughout history.”
Louis Anslow, creator of Pessimist Archive, has written that people have feared robots and automation for over a hundred years.
“There's a blog post where Louis Anslo, who created Pessimist Archive, writes about the fact that we've been, uh, fear mongering about robots and automation for for over a hundred years.”
The paper 'Towards Guaranteed Safe AI' lists authors including Josh Tenenbaum, Yoshua Bengio, Stuart Russell and Max Tegmark.
“She just mentioned this paper towards guaranteed safe AI, a framework for ensuring robust and reliable AI systems. Like you mentioned, it's like a who's who. Josh Tenenbaum, Yoshua Bengio, Stuart Russell, Max Tegmark”
Yampolskiy's paper on hacking the simulation opens by noting that many researchers have conjectured that humankind and the physical universe are simulated.
“Many researchers have conjectured that the humankind is simulated along with the rest of the physical universe.”
Lex Fridman cited a controversy in which Google's Gemini image generation produced historically inaccurate images such as Black Nazis and Black founding fathers.
“The Gemini 1.5 came out recently. There's a lot of drama around it, speaking of theatrical things, and it generated, uh, Black Nazis and Black founding fathers.”
Lex Fridman says Jeff Bezos first wrote about 'day one' thinking in his 1997 letter to Amazon shareholders.
“I think you 1st wrote about it in 97 letters to shareholders.”
Lex Fridman travelled to Pittsburgh to be scanned for his Codec Avatar.
“I went to Pittsburgh and went through this whole scanning procedure, which has so much incredible, uh, technology, so software and hardware going on”
Lex Fridman said Sam Altman had asked him how open OpenAI should be about GPT-4 and whether he would open-source the code.
“He turned the question to me of how open should OpenAI be about GPT four? Would you open source the code?”
Lex Fridman states that GPT-3 and GPT-3.5 had 175 billion parameters.
“so GPT three, 3.5 had, uh, 175 billion parameters”
Lex Fridman says the viral claim that GPT-4 had 100 trillion parameters came from a presentation of his, given when GPT-3 was released, that was taken out of context.
“I said, like, GPT four, like the next as it progresses, what I should have said is GPT n or something.”
Microsoft announced a new multi-year investment in OpenAI that was reported to be worth $10 billion.
“Microsoft announced the new multi-year, multi-billion dollar reported to be $10 billion investment into OpenAI.”
Ray Kurzweil first predicted in his 1999 book The Age of Spiritual Machines that computers would pass the Turing test in 2029.
“I made this in 1999 in my book, The age of spiritual machines, and then you repeated the prediction in 2,000 in 2005.”
Lex Fridman described game six of the 2021 World Chess Championship as the longest game ever played in chess.
“if you look at game six of the previous world Championship, uh, the longest game ever played in chess”