4M views · Oct 29, 2022 · 3:28:48 · 3 HN recommenders · 11 mentions · discussed 2022–2025
Lex Fridman's long conversation with Andrej Karpathy on Tesla's self-driving approach, Optimus, neural networks, transformers, AGI and more. It also touches on his OpenAI work and working with Elon Musk.
Why Hacker News points people here: Cited for Karpathy's first-hand explanations of camera-only self-driving, transformers and ML trends.
“Anyone seriously interested in this question should listen to the Lex Friedman interview of Andrej Karpathy”HN comment, 2023
“Anyone seriously interested in this question should listen to the Lex Friedman interview of Andrej Karpathy”
Essential listening on self-driving
Anyone seriously interested in this question should listen to the Lex Friedman interview of Andrej Karpathy: this video
“Great listen overall IMHO”
Karpathy expands on comparing neural nets to brain, finding physics 0-days
Andrej Karpathy was recently on Lex Fridman's podcast and covered this to some extent. He has the same perspective on this topic and expanded on it quite a bit. Great listen overall IMHO - this video I like his idea of finding 0-days in physics. :)
“this was the takeaway I got from this explanation (starts at 36:21)”
Explanation of transformer layers as message passing
My understanding is the transformer layer in the LLM is basically doing something akin to message passing, it’s like a mini computer. In predicting the next word, it has to understand a lot about a lot of different kinds of topics My understanding is kinda fuzzy because I haven’t coded it up myself, but this was the takeaway I got from this explanation (starts at 36:21) this video
“There's absolutely scope to build a better search engine based on these tools.”
Karpathy on scope for a better search engine
At 1:03:30 - this video > I think there's definitely scope in building a better search engine today. Google have all the tools, all the people, (...) all the data, it's just not obvious that they're capable as an organization to innovate on their search engine right now, and if they don't, someone else will. There's absolutely scope to build a better search engine based on these tools.
“Andrej Karpathy talked a bit about working with Elon recently”
Engineer's account of working with Elon
If you want to get an idea, why not listen to some engineers that have worked with him? There are plenty of those around, for example, Andrej Karpathy talked a bit about working with Elon recently - this video
“it seems machine learning takes the same path as physics already did.”
Karpathy discusses ML following physics' path to big collaborations
As pointed out in [1], it seems machine learning takes the same path as physics already did. In the mid-20th century there was a "break" in physics, before individuals were making ground breaking discoveries in their private/personal labs (think Newton, Maxwell, Curie, Roentgen, Planck, Einstein, and many others) later huge collaborations (LHC/CERN, Icecube, EHT, et al.) are required, since the machinery, simulations, models are so complex, that groups of people are needed to create, comprehend and use them. 1. this video Lex Fridman podcast with Andrej Karpathy P.S. To counteract that (unintentionally actually, likely because of a simple optimization of instruments' duty cycle) in astronomy people come up with a concept of "observatory" (Like Hubble, JWST) instead of "experiment" (like LHC, HESS telescopes) where outside people can submit their proposals, and if selected get observational time. Along with raw data authors of the proposals get required expertise from the collaboration to process and analyze that data.
“Karpathy talks about the OpenAI project”
Karpathy discusses OpenAI's World of Bits project
Yeah, check out the Lex Friedman podcast episode #333 around minute 52 where Andrew Karpathy talks about the OpenAI project "World of Bits" that did this. this video
“Humor was one of the tests Karpathy would use as a benchmark for AI”
Karpathy on humor as an AI benchmark
>your boyfriend’s only in a wheelchair because he doesn’t want to kneel five times a day Within the roast context, this is a good line. Humor was one of the tests Karpathy would use as a benchmark for AI: this video
“Karpathy explains the reasoning on Lex's podcast”
Karpathy explains the camera-only reasoning
Karpathy explains the reasoning on Lex's podcast[1]. This was after he left Tesla, but of course he's hardly impartial. [1] this video
“There's some talk about it in the Lex Friedman podcast(s)”
Discussion of Tesla AI data collection goals
This, obviously, is one of the ongoing efforts and goals of Tesla AI, and the reason they collect so much data. There's some talk about it in the Lex Friedman podcast(s)[1]. [1] this video
“Here is a link to Karpathy discussing the trade-offs”
Karpathy discussing camera versus lidar trade-offs
The part that I always found difficult to square was not that Elon Musk towed this “no need for LiDAR” line so hard, but that Andrej Karpathy, who I generally consider a very reliable voice in this space, was also in strong agreement that cameras were all that was needed. Does anyone know if he still believes cameras is all you need? Edit: Here is a link to Karpathy discussing the trade-offs: this video
