Nice Podcast. Yann explained everything incredibly. As a CS student from IIIT Bangalore, I still feel that a decent understanding of Computer Science and ML is needed to understand the talk in detail. Understanding it in detail helps aspiring entrepreneurs to make better decisions.
What an incredible speaker Yann is. He explained many of the concepts I always wanted to learn, and he explained in a simple, relatable terms. Also, he is patient, and willing to explain something he has already explained again, and trying different way so that the other person can understand. Qualities of a great teacher. I would love to take his course.
This is by far the most intelligent and informative podcast I have watched. Being working in AI field and reading bunch of technical papers & articles, watching Yann explain is just treat. Thank you @Nikhil for conducting such a brilliant podcast.
That modesty and self reflection in the end "Trying to simplify concepts is an art and I am trying to be good at it" was the highlight for me. To be honest - things got super deep and technical for most part of the interview even though Nikhil was trying super hard to make a mental mind map which the viewers can latch on to. There is a huge opportunity for educators who have the knack of simplifying concepts, using analogies/examples and try to deliver the intuition behind deep tech like AI.
I never coded a single line of code in my life and I don't really feel overwhelmed. I kept up pretty well. I think he also ich kept up, but needed to write those notes to put or edit stuff on the video later on or something
I want to come back to this interview again and again. Not because of the technical knowledge, but to witness the humbleness of this great man and probably learn from that.
Love the fact how Yann emphasized on having solid knowledge via research based MA/PhD programs and publishing papers, rather than being swayed away by monetary incentives. Says a lot about how we approach innovation in our country.
Tu batayegaa ab kisko bulana h aur kisko nahi , jiska channel usko sochne do kyun idhar udhar ki baate kar rahe ho ustaad . X-MEN ko Ya deadpool ko bhi bulana chahiye
One of the Godfathers of AI in a podcast of an Indian Enterpreneur !! The magtitude of this feat by itself is just amazing! Thanks @nikhil.kamath for such quality podcasts
Wow, Nikhil Kamath, your the OG of podcasts , I believe your podcast will have a very bright future, Casually getting Head of AI from META to talk about AI, I am from AI and I know what it's like to get Yann LeCun for a 1.5 hour podcast like .. Your podcast is hands down the best on the internet rn..
Doesn’t matter what you think. He is building his personal brand in the name of educational podcasts, so that he can get idiots like you to write comments like this, which allows him to wank himself to sleep looking at these comments.
from 11:17 to 1:33:00 that's called transformation! started and ended with 'what is intelligence really?' 🔥 also team WTF, please check and edit the banner blurred at 53:33 which was supposed to hold some information I guesss
this video tells you that ai is not that easy that you think its totally different world altogether. Nikhil is doing what it mean to be done since a long back.
Nikhil seriously you create good podcast but for the very first time being an AI engineer I enjoyed any content every second please bring more people to talk upon AI
The biggest takeaway in terms of education is that LLMs (Large Language Models) are discrete. Video models, when fully developed, will have a deeper understanding of the world. If you want to ship an AI product, you should have a deep understanding of the AI field, as opposed to merely hiring someone and integrating AI into your product. good stuff!
A Student of philosophy relates to the Blind man and Elephant a lot more. Ksanikvada theory or the Theory of Causation, Pratityasamutpada has a lot to offer. I am a Science student and I find it fascinating.
Been waiting since the teaser release wish there more frequent and quick. 😅 I know that this isn't Nikhil's profession but he's really good at it !........
1:11:12 @Nikhil @WTF a correction in the visualisation, the remainder video goes through the "same encoder" as the first one and not "semi encoder". great content other wise.
Nikhil, for this talk with Prof LeCunn ,could have brought Varun Mayya to accompany him. Prof Cunn is in academic mode while Nikhil is in big picture mode so that we the layman could understand what Prof is trying to say. Varun would have been a good catalyst here.
I, respectfully, disagree. It takes a newbie to ask real newbie questions. Mayya has what people call 'the curse of knowledge.' I think most people watching the episode are also in the big-picture mode. They don't want to understand academic nuances but want to understand the basics of AI and the direction in which AI is heading from an authority figure/real academic like Prof LeCunn.
The interesting thing is that LeCun hesitated for a moment when asked: What will everyone else do when AI does everything for them? This speaks volumes about how much he has considered the issue.
Nikhil Best episode in your podcast till date and I've been following you since day 0 This particular episode is very heavy. So I think it would make sense to add more animations and visuals midway where a major theory is being explained. Like when yaan explains reinforcement learning or anything for that matter. We are not talking about a zomato or a swiggy here. We're talking about how biological beings think. And for a noob like me it's not very easy to explain. Please add more visuals. I'd love to connect and talk about how we can democratize information and learning more cause I'm sure a lot of us want to learn about AI but get daunted by the sheer scale of it.
Nikhil do a video on your take on day to day mundane things like your day , breakfast, lunch, dinner, your take on house, leisure, doing nothing, of course your mother tongue Kannada !!
Everyone's going ga-ga on this, Nikhil and I do appreciate the effort and the background prep. I'm also likely to get a lot of pushback for this too - but, oh well. Here goes -- I think a large part of the struggle and interruptions in the first half was simply because you were asking him to "describe" a hierarchical diagram with you following and drawing it "in the air", so to speak, whereas he could have just as easily drawn it himself box by box and explained it beautifully (50s we had this, then there were these 2 branches, this one was dominant till the 90s and so on..). Kind of like someone having you describe a ticker tape's progression to an OHLC chart to a point-and-figure and they drawing it while they are also asking you about Level 2 quotes and 0 DTE option chains and expectancy!
Thanks for finally getting a real AI expert, Yann is legendary. Though your questions are still too basic - maybe you wanted to cater to the broader audience. Appreciate you bringing real AI professionals, please consider bringing in geoffrey hinton or demis hassabis next if you can.
@gunaysoni6792 it's trained on the internet what do you want? Agentic systems have access to search engine. Also differentiability in memory is not new. Neural turing machine was a good demonstration of that.
i can definitely sense that towards the middle of the conversation, it was hard to keep track of all the nuances. while i admire the approach that Yann took, it seems that it still remains tricky to easily explain the mechanics behind the current AI trends. despite working in the field, i completely see that while trying to understand new concepts. i imagine it to be much more challenging for those from the outside. nonetheless it has been very inspiring for how to progress in this field and how to talk about it with others.
there is years of scientific effort gone behind identifying the ingredients of what is finally working a little bit today. Give respect to that. Not everything can be simplified in an hour long video.
Always find interviews of great people very tiring and frustrating when the interviewer interrupts. You have this great person that starts with a simple thought we actually know, and wants to end in something probably amazing. And we are curious to learn. In this interview, the interviewer has great questions, but actually destroy this greatness by always interrupting Yann at the worst time.. Let him talk man!
At 13:30 the elephant analogy is the famous concept in jainsism teaching anekantwad and syadwad. In short every one speaks and correct from his viewpoint, and truth is non absolute
Thanks for the forecast! I have a quick question: I have a SafePal wallet with USDT, and I have the seed phrase. (alarm fetch churn bridge exercise tape speak race clerk couch crater letter). How can I transfer them to Binance?
Please can we have Rajiv Bajaj, MD Bajaj Auto. He speaks brilliantly, would love to see Nikhil and Rajiv together. Also add Royal Enfield guy Sidhartha Lal along with Motor Inc Schumi or Ather's Tarun Mehta. A good episode on ICE to EV, bikes, brands etc. From product perspective, ICE bikes are India's best export yet. Or Rajiv Bajaj along with his brother Sanjiv and Nikhil along with his brother. 3rd generation vs 1st generation business owners.
Nikhil / Viewers, If you want to truly visualize how Neural Networks, Backpropagation, Gradient Descent, and Cost Functions work then see Grant Sanderson’s (3Blue1Brown) brilliant five-part series. It’s like a PhD in Deep Learning made visual and easy to grasp. A big thanks to Nikhil and Yann for this masterpiece!
I see so much of incompatibility on both the ends coz of not having the requisite knowledge. One thing Yann could've done better is to explain things with simpler examples coz Nikhil is a pure non tech guy (and a big chunk of audience watching this video). For example, its easy to explain what Reinforcement Learning is, by explaining "how to train dogs" and Supervised Learning can be explained using an example like "I show a bunch of fruits to a kid (training), and then I check the kid's knowledge by asking questions about those fruits (testing)" Something along those lines would've made a better conversation between the two. PS: I have huge respect towards Yann, he's a legend in the field of AI. Not taking away any of that from him
I'm curious to see how far the JEP-Architectures progress is already. There is nothing said about it, it's like so secretive, I don't know if it's gonna be open sourced (the future models based on jepa)
So basically , if i get it right, the next 5 years vision states that there would be alot of companies forming just on a configuration of an open source AI platform like lama and that configuration and changes they will making on that model to design the need will be their sole USP, So to summarise they aren't making a new invention but using the invention to build something useful to people and monetary fullfilling to the Companies?
As a core researcher in ML, I feel many people don’t fully grasp the immense influence Yann has in this community and on modern AI as a whole. He has literally built the foundation of everything we see in modern AI, the very AI that is now recognized by the general public. For someone so technically profound, engaging with him on a podcast or being a related practitioner demands exceptional acumen. He shouldn't have come here at all :/
23:14 Seeing this image of C & D I remember my computer teacher ask the class to write a code which will display C & D on simulation, almost every one wrote code of 50-60 line & i wrote it in ~10 lines, I got punished & bullied my teacher for being lazy & short-cut person.
The humbleness , The modesty of "Godfather of Ai " is evident . Something to learn
Bang onn..... recommendation
It is Geoffrey Hinton
@@RohitKumar-uc4fw is considered one of those big guns in the area as well
He is not god after of AI
Nice Podcast. Yann explained everything incredibly.
As a CS student from IIIT Bangalore, I still feel that a decent understanding of Computer Science and ML is needed to understand the talk in detail. Understanding it in detail helps aspiring entrepreneurs to make better decisions.
Agreed
I feel understanding is required to understand what is being said
agreed...india lacks deep tech podcasts
@@MitochondriaispowerhouseofceII lets start one then.
@@chanpreetsingh007 that's something i always wanted to do but pehele jee 2025 dedu🥲
What an incredible speaker Yann is. He explained many of the concepts I always wanted to learn, and he explained in a simple, relatable terms. Also, he is patient, and willing to explain something he has already explained again, and trying different way so that the other person can understand. Qualities of a great teacher. I would love to take his course.
This is by far the most intelligent and informative podcast I have watched. Being working in AI field and reading bunch of technical papers & articles, watching Yann explain is just treat. Thank you @Nikhil for conducting such a brilliant podcast.
That modesty and self reflection in the end "Trying to simplify concepts is an art and I am trying to be good at it" was the highlight for me. To be honest - things got super deep and technical for most part of the interview even though Nikhil was trying super hard to make a mental mind map which the viewers can latch on to.
There is a huge opportunity for educators who have the knack of simplifying concepts, using analogies/examples and try to deliver the intuition behind deep tech like AI.
Funny to see Nikhil struggling to keep up with info bombarded😂
Too much jargon for a non engineer
I never coded a single line of code in my life and I don't really feel overwhelmed. I kept up pretty well. I think he also ich kept up, but needed to write those notes to put or edit stuff on the video later on or something
I want to come back to this interview again and again. Not because of the technical knowledge, but to witness the humbleness of this great man and probably learn from that.
Yann is very patient and friendly to explain these simple terms. Really appreciate that, great minds come with great humbleness.
Love the fact how Yann emphasized on having solid knowledge via research based MA/PhD programs and publishing papers, rather than being swayed away by monetary incentives. Says a lot about how we approach innovation in our country.
Maybe the next guest should be Aswath Damodharan someone from Finance and Valuations😊
Tu batayegaa ab kisko bulana h aur kisko nahi , jiska channel usko sochne do kyun idhar udhar ki baate kar rahe ho ustaad . X-MEN ko Ya deadpool ko bhi bulana chahiye
Go Home and something like that i don't have anything , rapper , P. diddy
God bless you
You are blessed . Simulations
he will value zerodha in live xd
One of the Godfathers of AI in a podcast of an Indian Enterpreneur !! The magtitude of this feat by itself is just amazing! Thanks @nikhil.kamath for such quality podcasts
Excited for this release since you uploaded the teaser
Wow, Nikhil Kamath, your the OG of podcasts , I believe your podcast will have a very bright future, Casually getting Head of AI from META to talk about AI, I am from AI and I know what it's like to get Yann LeCun for a 1.5 hour podcast like .. Your podcast is hands down the best on the internet rn..
Doesn’t matter what you think. He is building his personal brand in the name of educational podcasts, so that he can get idiots like you to write comments like this, which allows him to wank himself to sleep looking at these comments.
Finally Indian podcast getting a quality guest who delivers real life technical knowledge especially for a low tier BE undergrad like me
from 11:17 to 1:33:00 that's called transformation! started and ended with 'what is intelligence really?' 🔥
also team WTF, please check and edit the banner blurred at 53:33 which was supposed to hold some information I guesss
this video tells you that ai is not that easy that you think
its totally different world altogether.
Nikhil is doing what it mean to be done since a long back.
Although this video doesn't have a ton of views, the content and the way it's presented are absolutely exceptional. Great work! 🎉👏
Those who explain complex things in simple words are quite genius ...yann is one of them.
One of the best podcast watched as an AI enthusiast particularly in India. Got to learn great insights. Thank you🎉
I want to thank you for this series. As a beginner in the world of business, I find these give me a great groundwork to begin with.
Watching this Podcast just after my Deep Learning Exam
Wonderful interaction, Loved it!
As an ai/ml student i love this podcast. Thaky you for bringing Yann ❤❤
God father of AI for a reason. Thanks Nikhil for asking questions that are simple and learning drive….awesome podcast..
Thankyou for bringing him...please bring more guys with deep understanding of tech especially ai
Knowledge is simplicity. Yann epitomises the simplicity. A teacher like him can make all the difference in your learning process.
Nikhil seriously you create good podcast but for the very first time being an AI engineer I enjoyed any content every second please bring more people to talk upon AI
Thank you so much Nikhil.. for providing great interview videos
Good Watch, it's criminal that it has such less views. Tech folks in India should watch this without a doubt
For the aspiring Data scientist , it is a must watch video. Even for a noob like me, it delivers great insights.
Wowww… he explained in such detail yet easier for a layman. Benefits of getting a guy who knows stuff.
Thanks for uploading this! Always love to hear from Yann LeCun. So much to learn yet.
The biggest takeaway in terms of education is that LLMs (Large Language Models) are discrete. Video models, when fully developed, will have a deeper understanding of the world. If you want to ship an AI product, you should have a deep understanding of the AI field, as opposed to merely hiring someone and integrating AI into your product. good stuff!
Extremely informative podcast , it's a delight to hear such conversations cheers Nikhil
Thank You Nikhil & Yann sir for making this ❤. Special thanks to Kailash Nadh sir for inspiring all us techies towards Open Source.
One of the best podcast ever he was very humble and passionate
A Student of philosophy relates to the Blind man and Elephant a lot more. Ksanikvada theory or the Theory of Causation, Pratityasamutpada has a lot to offer.
I am a Science student and I find it fascinating.
Sir please, Do a podcast with Andrew Ng.
What a combo is this? great! Would like to see more great people in your show
Been waiting since the teaser release wish there more frequent and quick. 😅 I know that this isn't Nikhil's profession but he's really good at it !........
Nice ❤
Quality podcast.
8:10 Sam is sick. Nikhil cooked here 😂
Altman ?
amazing and it is great to unfold the maths behind the AI
1:11:12 @Nikhil @WTF a correction in the visualisation, the remainder video goes through the "same encoder" as the first one and not "semi encoder". great content other wise.
This is incredible stuff. Loved the podcast.
Nikhil, for this talk with Prof LeCunn ,could have brought Varun Mayya to accompany him. Prof Cunn is in academic mode while Nikhil is in big picture mode so that we the layman could understand what Prof is trying to say. Varun would have been a good catalyst here.
I, respectfully, disagree. It takes a newbie to ask real newbie questions. Mayya has what people call 'the curse of knowledge.' I think most people watching the episode are also in the big-picture mode. They don't want to understand academic nuances but want to understand the basics of AI and the direction in which AI is heading from an authority figure/real academic like Prof LeCunn.
I actually liked both of you's POV
1:02:48 editior needs an appreciation ❤
The interesting thing is that LeCun hesitated for a moment when asked: What will everyone else do when AI does everything for them? This speaks volumes about how much he has considered the issue.
I don't know why but as an engineer would love to see K doing this podcast along with Nikhil 😊
Nikhil
Best episode in your podcast till date and I've been following you since day 0
This particular episode is very heavy. So I think it would make sense to add more animations and visuals midway where a major theory is being explained.
Like when yaan explains reinforcement learning or anything for that matter.
We are not talking about a zomato or a swiggy here. We're talking about how biological beings think.
And for a noob like me it's not very easy to explain.
Please add more visuals.
I'd love to connect and talk about how we can democratize information and learning more cause I'm sure a lot of us want to learn about AI but get daunted by the sheer scale of it.
it was really insightful and valuable podcasts for me sir keep coming with this type of tech podcast i really loved it a lot by heart
Being a data scientist I can say that this was a whole revision of ML and DL.
"Tell me if you don't understand it, because i can explain it in different ways" - Only true experts speak this language.
Been waiting for this!
Waiting since long time 🙇🏻♂️
Superb , just great episode ❤❤❤
Thanks for this fantastic educating Podcast
Thank you sir , a very educating podcasts!
This indian guy is going to rule the world one day.
Nikhil do a video on your take on day to day mundane things like your day , breakfast, lunch, dinner, your take on house, leisure, doing nothing, of course your mother tongue Kannada !!
Can we get Nikhil Kamath's Notes from this Conversation. Would be great summary of the entire discussion.
Everyone's going ga-ga on this, Nikhil and I do appreciate the effort and the background prep. I'm also likely to get a lot of pushback for this too - but, oh well. Here goes --
I think a large part of the struggle and interruptions in the first half was simply because you were asking him to "describe" a hierarchical diagram with you following and drawing it "in the air", so to speak, whereas he could have just as easily drawn it himself box by box and explained it beautifully (50s we had this, then there were these 2 branches, this one was dominant till the 90s and so on..). Kind of like someone having you describe a ticker tape's progression to an OHLC chart to a point-and-figure and they drawing it while they are also asking you about Level 2 quotes and 0 DTE option chains and expectancy!
Man this is too computer science heavy,
As a commerce student,
i love it ❤
Lol
Nikhil randomly dropping bangers at 10:30 on a Wednesday night
Thanks for finally getting a real AI expert, Yann is legendary. Though your questions are still too basic - maybe you wanted to cater to the broader audience. Appreciate you bringing real AI professionals, please consider bringing in geoffrey hinton or demis hassabis next if you can.
hey nikhil thanks for the interview.
Nikhil was right. Current LLM systems have persistent memory. Example is RAG systems ( Retrieval Augmented Generation systems)
RAG is not part of the LLM, you are doing retrieval separately and giving it to the LLM as context.
@gunaysoni6792 it's trained on the internet what do you want? Agentic systems have access to search engine. Also differentiability in memory is not new. Neural turing machine was a good demonstration of that.
@@TheCuriousCurator-Hindi Yann explained this with the book example. Parametric memory is not the same as persistent memory.
@gunaysoni6792 can you share the link?
@TheCuriousCurator-Hindi 1:03:05
i can definitely sense that towards the middle of the conversation, it was hard to keep track of all the nuances.
while i admire the approach that Yann took, it seems that it still remains tricky to easily explain the mechanics behind the current AI trends. despite working in the field, i completely see that while trying to understand new concepts. i imagine it to be much more challenging for those from the outside.
nonetheless it has been very inspiring for how to progress in this field and how to talk about it with others.
there is years of scientific effort gone behind identifying the ingredients of what is finally working a little bit today. Give respect to that. Not everything can be simplified in an hour long video.
Best podcast ever❤️
Always find interviews of great people very tiring and frustrating when the interviewer interrupts.
You have this great person that starts with a simple thought we actually know, and wants to end in something probably amazing. And we are curious to learn.
In this interview, the interviewer has great questions, but actually destroy this greatness by always interrupting Yann at the worst time..
Let him talk man!
At 13:30 the elephant analogy is the famous concept in jainsism teaching anekantwad and syadwad. In short every one speaks and correct from his viewpoint, and truth is non absolute
Ksanikvada and Pratitsamutpada as well.
Hello Fellow Philosophy geek 😂❤
18:10 bringing up chess in front of Nikhil was wild !! 😂
Thanks for the forecast! I have a quick question: I have a SafePal wallet with USDT, and I have the seed phrase. (alarm fetch churn bridge exercise tape speak race clerk couch crater letter). How can I transfer them to Binance?
Good work as always ❤
Thank you for this!
Thank you for enlightening India
Please can we have Rajiv Bajaj, MD Bajaj Auto. He speaks brilliantly, would love to see Nikhil and Rajiv together. Also add Royal Enfield guy Sidhartha Lal along with Motor Inc Schumi or Ather's Tarun Mehta. A good episode on ICE to EV, bikes, brands etc.
From product perspective, ICE bikes are India's best export yet.
Or Rajiv Bajaj along with his brother Sanjiv and Nikhil along with his brother. 3rd generation vs 1st generation business owners.
Questions could have been different, for these information one can watch many lectures on youtube
Nikhil / Viewers, If you want to truly visualize how Neural Networks, Backpropagation, Gradient Descent, and Cost Functions work then see Grant Sanderson’s (3Blue1Brown) brilliant five-part series. It’s like a PhD in Deep Learning made visual and easy to grasp. A big thanks to Nikhil and Yann for this masterpiece!
Thank you nikhil ❤
Much needed 🙌
it is quite disturbing this interviewer kept on interrupting Yann, otherwise, it would be a perfect interview. Thanks Yann.
finally an actual expert on AI :) who is not just regurgitating buzzwords
Oh yeah this, got notified but little late let’s watch
Legendary
I see so much of incompatibility on both the ends coz of not having the requisite knowledge.
One thing Yann could've done better is to explain things with simpler examples coz Nikhil is a pure non tech guy (and a big chunk of audience watching this video). For example, its easy to explain what Reinforcement Learning is, by explaining "how to train dogs" and Supervised Learning can be explained using an example like "I show a bunch of fruits to a kid (training), and then I check the kid's knowledge by asking questions about those fruits (testing)"
Something along those lines would've made a better conversation between the two.
PS: I have huge respect towards Yann, he's a legend in the field of AI. Not taking away any of that from him
Yarr Le Caaaunnn ko lekar agya bhai mera is baar..
*stupid me Google* - Heinnnm 😮😮😮😮😮
Nikhil your questions are real good for non engineerr
More engineer like..😊
Amaizing!
Information bombardment by nikhil Kamath sir 😂
Thanks Nikhil ❤
I'm curious to see how far the JEP-Architectures progress is already. There is nothing said about it, it's like so secretive, I don't know if it's gonna be open sourced (the future models based on jepa)
How do you become so eloquent as Nikhil? Would love to hear..
Amazing questions
He deserves a Nobel prize 😊
So basically , if i get it right, the next 5 years vision states that there would be alot of companies forming just on a configuration of an open source AI platform like lama and that configuration and changes they will making on that model to design the need will be their sole USP,
So to summarise they aren't making a new invention but using the invention to build something useful to people and monetary fullfilling to the Companies?
Dude u have a great network ❤
As a core researcher in ML, I feel many people don’t fully grasp the immense influence Yann has in this community and on modern AI as a whole. He has literally built the foundation of everything we see in modern AI, the very AI that is now recognized by the general public. For someone so technically profound, engaging with him on a podcast or being a related practitioner demands exceptional acumen. He shouldn't have come here at all :/
Is it me only or Yann is actually rhyming every word he's speaking 🤯
Very humble men a ai godfather
23:14 Seeing this image of C & D I remember my computer teacher ask the class to write a code which will display C & D on simulation, almost every one wrote code of 50-60 line & i wrote it in ~10 lines, I got punished & bullied my teacher for being lazy & short-cut person.