Yann Lecun on Llama 3 open source model | Yann LeCun and Lex Fridman
Vložit
- čas přidán 10. 03. 2024
- Lex Fridman Podcast full episode: • Yann Lecun: Meta AI, O...
Please support this podcast by checking out our sponsors:
- HiddenLayer: hiddenlayer.com/lex
- LMNT: drinkLMNT.com/lex to get free sample pack
- Shopify: shopify.com/lex to get $1 per month trial
- AG1: drinkag1.com/lex to get 1 month supply of fish oil
GUEST BIO:
Yann LeCun is the Chief AI Scientist at Meta, professor at NYU, Turing Award winner, and one of the most influential researchers in the history of AI.
PODCAST INFO:
Podcast website: lexfridman.com/podcast
Apple Podcasts: apple.co/2lwqZIr
Spotify: spoti.fi/2nEwCF8
RSS: lexfridman.com/feed/podcast/
Full episodes playlist: • Lex Fridman Podcast
Clips playlist: • Lex Fridman Podcast Clips
SOCIAL:
- Twitter: / lexfridman
- LinkedIn: / lexfridman
- Facebook: / lexfridman
- Instagram: / lexfridman
- Medium: / lexfridman
- Reddit: / lexfridman
- Support on Patreon: / lexfridman - Věda a technologie
Full podcast episode: czcams.com/video/5t1vTLU7s40/video.html
Lex Fridman podcast channel: czcams.com/users/lexfridman
Guest bio: Yann LeCun is the Chief AI Scientist at Meta, professor at NYU, Turing Award winner, and one of the most influential researchers in the history of AI.
Yann Lecun makes interesting points not on this particular Clip but on one of the previous, through the interview, about languages and processed information like when we speak more than one language and someone says for instance the word 'sun', what pops in our mind would be the image of the sun and not the word spelled out written somewhere, but I don't know if that is the response of the human brain generally or only type of reasoning and others will actually when listening to the word 'sun' think of not only the image that it represents but also how the word is written in all of the languages the listener is familiar with.
Also interesting and something was one of the first things that popped to my mind when I first started following you Lex Fridman and the subject of AI, was the fact that AI couldn't reproduce the sense of smell and taste, but now a few months later in my new technology interest thinking pursuit, along with my previous general mind subject interest, I have come to question as to whether the neurological response of the human brain to the senses of taste and smell, could actually in fact indeed be reproduced by AI.
I do think though that Mr. Lecun seems to be stuck in the past with his ballpen analogy and a doomsday possibility that needs to be eliminated before it even starts but it wouldn't if we stuck with his analogy, that isn't keeping up with the progressed achieved so far in humanity and societies to the level of getting away from the doomsdays of the old ages where inquisitions and genocides were just around the corner, thankfully we have progressed past those and in our fight to preserve peace and progress we can't not adjust our current views of the pictures and carry on going by merely and only narratives of the past, we need to give ourselves some credit and carry on moving forward with more humility to the possibilities of betterment that we can't today even imagine and that is not gonna happen if everyone working with AI held only that type of view. AI is not ballpen, and it shouldn't be treated like a primitive tool as it isn't, and because of it 'open source' should be only basic and minimal. Ask NASA if they're gonna put all the source codes for the moon trajectory available to the public everywhere.. obviously not.
Cheers. Thanks Lex Fridman for the amazing work you do and thanks for sharing with us all the interviews you do with amazing folks, best of the best on every area. We need Yann LeCun ideas just not only his ideas, we need all ideas on the table, well, not the open public's table, but your tables, in the best capable hands, minds and hearts. Bless ya ❤
woah it sounds like hes talking to an ai
yann is the first AI person i've seen that actually has it right. too many people on lex and other podcasts have a financial incentive to talk up their LLM products. he's right. they are doomed. AGI might come someday, but this isn't it.
I’m a layman but I’m unsure about this argument that ai is inefficient compared to the human brain. If I could write a masterful essay and paint a beautiful picture in under a minute my brain would require a lot more energy.
It's the training that uses up a lot of power. Racks of GPUs maxed out for months. More than your brain would ever consume in a lifetime. But a single query doesn't use so much.
AGI is not possible using computers or any silicon based technologies.
LeCun is way too honest. He is single-handedly making investors lose interest in Meta by saying things like he doesn't have his JEPA systems working well yet. Also, he doesn't seem like a real nice guy, by saying that LLMs can never get to AGI just to create drama. You can see he is smiling when he is saying this.
Llms aren’t thinking… they are pattern matching through graphs. No, that’s not agi
Yeah, that's not drama. That's just a fact. I'm a data scientist, so I can say with confidence that even a sufficiently large LLM, *alone,* is never going to *poof* and AGI. It's not built like that.
@@AutitsicDysexlia I’m not arguing that LLMs will lead to AGI. I am saying that LeCun isn’t a chill dude. You misunderstood what I was saying.
Yeah, because he's the only one speaking and working in the area, nobody else.
What makes you makes you come to such conclusions, have you perhaps stopped to wonder for yourself? Just curious
You might be a "chill dude" but you probably are working in a 9-5 you don't want to work in & have very close to 40IQ @@greengoblin9567