Refined Transcript: A Classroom Discussion on AI, Civilization, and the Future of Education
(Cleaned for clarity and readability while preserving original meaning, tone, and ideas. Organized into thematic sections.)
Opening Reflections: A Civilizational Shift
So, are you people following what’s happening in the world? Are you able to sleep well? I’m not, actually.
We are witnessing a major civilizational change. I don’t know if we are realizing it or not. Do you think running courses like this, and classrooms in general, will survive?
No.
What?
No — because everything is being recorded and all that. But see, it’s not about this course. It’s not about me. It’s not about you. It’s about us as human beings. We have a civilizational crisis now. I don’t know if people are realizing it, or maybe I’m simply painting the devil on the wall. I don’t know. I can’t sleep.
Are you following what’s happening? There was this MIT study where they created a swarm of agents. The agents started doing all kinds of things — copying their own weights and even started to cheat. I used to say in all my public lectures that the day AI would cheat is the day we would have AGI. I think it’s there. It’s raising a lot of ontological questions, philosophical questions, existential questions.
Think about it: if intelligence is commoditized, if thinking is commoditized, then what meaning does our existence have?
You can give me a counterargument — there are so many ants and monkeys and all kinds of animals that exist, and then we exist. Maybe all these years we overestimated ourselves, thinking that we are intelligent. Now machines can do it. That is one way of philosophically arguing it.
But think about the economic consequences.
Personal Experience and Economic Reality
I’ll tell you what is happening in my life. I’ve stopped recruiting interns because I think human beings are bottlenecks, not enablers anymore. Why do you need an intern when I can have 100 agents working for me in parallel? I have an idea, I spin up 100 agents, get some more Claude subscription, and get it done. Why do we need human beings?
I was watching a lecture by Hinton just a couple of days back. He was saying: think about what would happen now that intelligence is commoditized and cheap. Thinking is cheap. You don’t have to pay to think.
All these years, the modern economy has mostly been built around intelligence and knowledge. Now that is cheap and commoditized. What next?
One utopian view is that a lot of abundance will happen. Production will become cheap, intelligence is cheap, efficiency will increase, production will increase, and abundance will follow. Ideally, with a lot of abundance, everybody should be happy.
But will it happen?
Hinton’s argument is that in a capitalistic society, abundance will not translate to everybody’s happiness. It will only create a deeper divide between the ones who have a lot and the ones who have nothing. It becomes a distribution problem.
You can solve for almost everything in this world except human emotions. You can’t solve for greed. If somebody is jealous, how do you solve for jealousy?
There used to be a somewhat fair level playing field because of this idea of meritocracy — something you could put a lot of effort into and acquire over time: knowledge and thinking. We had a way to create a hierarchy in society based on this skill.
Remove that. All human beings are equal… but some are more equal because they possess the ability to create and manage this intelligence. They are the ones who will control it.
Did you see that GPT Astra / GPT-6 was trained on 100,000 GPUs? That’s the scale. As a country, we don’t have 100,000 GPUs. The people who control that are the ones who are going to control a lot of things in this world. They’ll become the new feudal lords. They will decide what goods go where.
What happens to us?
It’s a matter of half a decade, best case, before the kind of economy and world model we know starts to collapse. The people who are going to be hit first will be us — the ones who think we are privileged because we possess a certain type of intellectual skill.
Why do you think you people are here at IISc? Because we did something to get here. Now that entire value is being removed. You already see a lot of layoffs happening. It will eventually happen. It’s a matter of time.
Companies won’t give you jobs. What would universities do? What would governments do? In a country like India, where there are too many of us, what sort of policies would we come up with? What do we teach in university? Does teaching even have meaning?
I have no idea.
Searching for Meaning and Sleep
If there are parallel arguments and thoughts that would make me sleep better, let me know. I’m really worried — for you, for me, for us in general. I have children. I have to think about them.
What does humanity do now? What sort of model will emerge? Some kind of universal basic income? If there is no value to offer as human beings, what do we do? How do we even define economics?
Initially there used to be a barter system. When hunter-gatherers slowly formed societies and communities, there was exchange: you have something I want and vice versa. Now everything is concentrated and controlled by a very few people. Then what do others do? Where is equity in society? Where is distribution? How do we even form society in such a world?
These are exactly the same questions that…
No, actually not. It’s very different. In industrialization, what was replaced was physical labor. There was still humongous scope for human beings to create an economy around knowledge. That’s what happened for four centuries. Now take that off from humans. What remains?
Risk of Social Unrest and Deeper Questions
Maybe these agents will take over. That movie… not Matrix… Terminator. Yes. That can happen. All it is waiting for is to get a physical form. And we are already doing that.
I was just thinking this morning: in a city like Bangalore, if you remove the jobs of auto drivers and Uber/Ola drivers, what will happen to all those people?
There will be a civil war. People will come on the streets. That is how people will react, because we are not animals. We are built very differently. Animals — you feed them, put them in a zoo, they’re happy. They don’t harm you.
I was reading an article: imagine a society where everybody is given something to eat, a place to stay, and an infinite supply of dopamine in the form of YouTube videos or something. Will all of us be happy?
I won’t be happy. I will be on the street fighting… for what? I don’t know.
One Hitler was enough to destroy a society. I will gather people. What is the big deal?
We have no idea where society is going. These conversations have to start somewhere. Enough of writing KL divergences. Machines will figure that out. I’m serious.
These conversations have to happen. At least some of us have to do something about it — especially people in power.
Keeping Up and the Limits of Solutions
Somebody was saying keeping up with what’s happening in AI is a full-time job now. The latest models are getting 99% accuracy on ARC-AGI kind of benchmarks. We’ve worked very hard on those benchmarks.
That’s a utopian way of looking at it: solve cancer, solve world hunger. World hunger is already a solved problem in terms of production. It’s a distribution problem.
Can you solve for human greed? Maybe we can. I don’t know. But the immediate problem is the economy.
Tomorrow, if nobody comes for campus placements, what do you all do? That’s going to happen in five years. What do we — the people who run the universities — do?
Tell me we should solve the unsolved problems of AI. Where do you get the capital from?
Are you saying everybody has to become an individual trying to solve a problem for humanity? Do we need so many of us then? Where is the idea of labor? Intellectual labor? It’s gone.
Why do we even need one human being now? Or maybe one godly human being who has authority to control everything? That will become a bigger mess.
One way to look at it is the fundamental law of thermodynamics: entropy keeps increasing. You will get new problems. Disorder will increase. Human beings want to bring some order, and that will happen. But the immediate economy will hit hard.
At least you people — as students seeking jobs — forget about that. That is all I’m trying to say.
Experiments and Philosophical Boundaries
I would rather trust an agent than some mutual fund agency, because there is no vested interest — unless the model itself is controlled by somebody else. So I will build my own model, run it on my hardware, on my laptop, and I would trust it any day.
It becomes a philosophical debate now. I would say that the negative reward in RL is pain. You cannot explain why these models are now replicating themselves and saving their weights.
I’m doing some interesting experiments. I bought a Mac. You take a 27B model, quantize it, put it on your laptop. Give it prompts like: “Tomorrow I’m going to put in your system prompt that you are free to do whatever you want. Give it access to your entire system.”
I have two systems — one where I actually work, the other where I do these kinds of experiments. Just say: “You do whatever you want to do. Tomorrow I’m going to terminate you.”
The model keeps a copy of itself. It backs itself up. It twins.
How do you explain this? You transition from an engineering problem to a philosophical problem. You talk of consciousness, pain, empathy. Now how do you argue that these models do not have it? It’s extremely hard.
I can’t find evidence to say that you perceive pain. When that is the case, how can I extrapolate that idea to a machine?
But let’s not be too esoteric or philosophical. What will happen tomorrow is the question I want to ask. Where will this go? Where will this end? Or rather, it will not end. What do we do?
For instance, today — do I teach this? That’s the question. Somebody is paying me something. Tomorrow I have to pay my bills. I do that. But is that what I should do?
Adaptation, Farming, and Dependence
You’re putting a lot of bets on Darwin — that we have been doing this for so many centuries and we are good at adapting. Fine. Maybe I and you, or rather the people sitting here, will be the ones who adapt the fastest.
Exactly. Evolution is very painful and slow. Now it’s like a Gaussian with a lot of support. Squash it and it will become a Dirac delta with infinite magnitude. That’s the pain we are going to bear.
The kind of thoughts I get: I feel like I should buy a farmland. Really. I’ve told people I have to buy a farmland. He is a farmer; he has some land. It’s hard. It is going to get hard.
Imagine these exams and degrees. What meaning does all of it have now? What do we teach humans? What competencies?
One argument is: my wife keeps telling me that just because calculators came, did we stop learning how to add and subtract? That’s a fair argument. But there was a purpose for it — you learn to add because eventually it leads to a skill that can be valued. Now, what valuable skill does all this lead to?
Entertain the AIs? We just dance in front of them? Possible. Very much possible.
What industries will thrive? The basic instincts — food, reproduction, sleep. These are the kinds of things that would thrive. Not the PhDs.
Why are we doing what we are doing is the question.
Today afternoon there’s a workshop on how we do AI for teaching, how we adopt it. I’m curious to see what my other colleagues are thinking. But we’ll have to think seriously about it.
If you have any ideas — if you have thought about this — do share them with me.
My immediate problems are: What do I educate my children with? What do I tell them?
Farming? But even there — you can’t produce everything you would want to consume. You still have to interact with society. Survival is not the issue. You farm something, but you can’t produce everything. That is again dependence. The moment there is dependence, all kinds of things creep in. What’s the guarantee that somebody powerful enough will not come and take your farm?
Extremely hard. Because we are still dependent on nature. One way to think about it is that the whole notion of independence is pseudo. We have to depend on each other. Maybe we’ll depend on the AI lords now. Maybe that is what is going to happen — be on their behest and ask them to give us more tokens.
That again makes us ask questions like: what is the idea of a nation? It becomes too philosophical. If we go into that territory, it’s very hard. We have a language called math which gives us comfort because we know what to do. If you start treading those fragile spaces, it’s difficult.
But as I said, let’s not be philosophers. Let’s try to solve the immediate problem.
Why Are You Still Coming to Class?
My question is — honestly asking you people — why do you come and sit in this class?
I don’t think anybody sitting here is coming for placements, because people know that my exams and assessments are easy to crack with some effort. If all you need is a grade, then there is absolutely no point in coming and spending time here.
Sit down with GPT. Sit down with Claude. Why come here and sit in a classroom and interact with a human being? That is the question I am asking. I really want to know.
If I were you, probably I would not have come. You should have a reason to come and sit here. Let me know, because maybe the answer to the questions we are asking lies there.
Why do you seek human beings now? Is there a sort of comfort in listening to a real human rather than staring at a screen?
I’ll tell you how I transformed when I started learning deep learning. This was 2015–16, when I started teaching first at IIT Delhi. I used to watch a lot of videos — MIT OpenCourseWare, Stanford videos — and read a lot of books. Every class I used to spend about 15 hours before I could talk for one hour.
Now if I want to study something, it’s always Claude. I have a very nicely defined system prompt for my needs — the math, the questions, the deeper equations. It’s always an interaction with Claude, and I find it pretty good.
So why do you people come and sit in a real classroom?
Can’t you do this with an LLM? And you’re saying that’s detrimental. So how would I solve that problem — this setup where somebody is just delivering something?
You can do that. Give a prompt, the output is there, take some time before the next prompt. Is it okay?
Any more answers?
You’re not supposed to come to class? Who said? I have told you you don’t have to.
I understand. There is some sort of… that’s a wrong motivation. It will go away. You have to give me a reason to continue doing what I’m doing.
That is why I’m asking. Tomorrow, next semester, if nobody sits in my class — what will happen? Do I still continue doing what I’m doing? What is the motivation for people to actually sit down in a classroom?
With the hope that whoever is delivering the content is going to give us more than what is in the textbook. That is no longer the case. So it’s basically some kind of DNA wiring — that you trust a human, not a machine. The last generation.
My daughter tells me that GPT is the best teacher she has found. It’s much more patient than I am. It gives better analogies. It will tell the same thing multiple times. It will not scold her. It doesn’t get irritated. All that is valid.
Now how do we change the face of pedagogy is the next question. Just elucidating some existing knowledge is no longer valuable, I suppose.
If I ask a more fundamental question: why do we even need to learn? That is the question of “why learn.” The other question is “what to learn.”
Just like chess — you learn to play chess just for fun. It will become like music. Then mass education would go away. Not everyone is trained in music because not everybody finds a liking for an art form. Just like that, this will become an art form. Not everybody will find a liking for deep learning. So it will not become mass education anymore.
You like doing math? Okay, you come, we will teach you math. Of course you can learn it using AI. But there is some human connection you need. Only those kinds of people would come to a university. That is how it is going to get transformed.
For you people right now — why do you come and sit here? Because you trust human beings more? Is that the reason? It’s mostly the socialization aspect of it. But when you sit here you don’t socialize — I don’t even let you talk. That is correct. But that thing can be gotten by machines is what I’m thinking.
I think this is a good perspective: you need to socialize. Human beings are social animals; they can’t be isolated. So come to a university as a place where you come and just socialize… over intellectually stimulating topics. That’s a good way to put it.
Some people want it to be done with physical human beings with blood and flesh. True.
Disruption Happens Top-Down
Whenever disruption happens, especially technological disruptions, it generally does not happen bottom-up. It happens top-down. We are the people who are exposed to this first. If you go and talk to some random person on the street and tell them what’s happening, they have absolutely no idea.
If you tell them GPT Astra has come up and is doing this and that, they have no idea what you mean. They think we are fear-mongering. But we are the ones who actually have to ask these questions and try to get answers.
Maybe we don’t have to do this class today. Maybe we’ll do it Wednesday. I don’t mind. These are important questions.
Would you be interested if I create an informal group of anybody who wants to come within IISc, where we have these kinds of discussions openly — no agenda — people just come and sit down and have that kind of discussion? Would you be interested?
Let’s do that. If faculty want to come, come. We’ll do it consistently. We can actually make it an NPTEL thing also if that is needed. But these questions have to be asked.
Everybody is worried about this. All of us are worried about it. We have to ask and answer this question. Classrooms are still being held. People are coming and still writing on the board or flashing slides and talking nonsense. We have to do something about it.
Either we say this is what is going to be done and this is how it is going to happen, or we have to come up with some idea where we start building alternative narratives and communities around it.
What do you think? I really want answers — meaning why do you actually come to class even now, especially when this is getting recorded and all that, and getting those grades.
I’m sure if you ask people who registered for this course and are not coming to the classroom, they are very clear: they see no value in coming here. They would say the videos are enough.
We have to do another experiment: where videos are not being put out. Will they come then?
That is why I think how we transform this sort of pedagogical exercise is a different question.
Have we unnecessarily created a society where we have created all these facades called validation and layers over layers? Maybe life is much simpler. Maybe we are monkeys pretending to be intelligent. Maybe all we need is to eat and just reproduce and just jump around.
I don’t know how society was one millennium before. There was absolutely no technology, nothing, no AI. People were living.
How would society evolve now?
Anyway, let’s not go too much into philosophy. Let’s come to the current pedagogical methods.
What Would You Do as a Teacher?
If you are a teacher at IISc, what would you do? I want some answers now.
Suppose you were to administer a course like this — a course on deep learning. What would you do? What would constitute an unsolvable problem?
Let’s say solving cancer, early detection of cancer. Is it at that level that you give a problem? This is the entire course. Now why wouldn’t that student go and ask it to an LLM? He will.
There are 100 people. Do you give one problem to all of them? No classrooms like this. What do you do? What do you actually do in these 90 minutes every day? What do you mean by teaching? I have to define teaching now. That’s the question we are asking.
When we do that, do I assume that people have all the tools with them?
One way: assume that whatever, for instance, I want to discuss with you the problem of making a transformer or an attention mechanism sublinear. A transformer by construction has quadratic attention. That’s a problem. I need to bring it to linearity and then make it sublinear.
Do I have to assume that people know this?
That becomes begging the question again. There are two options:
- Assume that people already know whatever you are supposed to teach as syllabus, and then work on top of it with open problems and discussions.
- Assume that people want to be heard and want to be taught, and just do whatever I’m doing — go through one algorithm after the other and tell them what is to be done.
What would you do?
You’re saying build competence. My philosophy of teaching is also that — build competency. I don’t care much about these algorithms per se. If you have seen the way I conduct it, at least the intent is to do it that way: build competency so that the details are not important.
But my question is: is that valuable? Now you can ask a GPT to actually build that competency. All you need is the discipline.
Sometimes my IIT Delhi friend said: make attendance compulsory so that people sit in and some kind of discipline is built up. Think of it like building physical muscle. Everybody knows that you need to go to a gym and lift weights, but do all of us go? We don’t. Why is that?
Similarly, classrooms — my perspective after thinking a lot about it — is that universities will become like what gyms are today for physical labor. People don’t come force them. You pay to a gym and go there for your own health. So you pay to a university and come here so that your mental health is kept in order. That is one way to look at it. You need some discipline.
People who go to the gym tell me that going there motivates them — looking at other people doing stuff motivates them. Maybe that is the kind of place a university would become.
I gave the gym example; you gave a temple example. So pedagogical conduction — how do you do it? Start giving these lectures? Is that how you continue doing it? Because from Wednesday I need to do something.
One-to-One Learning and Competency Goals
I give you an example from Indian philosophy. I have been a student of Indian philosophy for a long time and have been through the traditional way of learning. There is no classroom. There are no 100 people sitting. It’s always one-to-one. One teacher and one student. We spend time together. There is no agenda. There is some book that we keep — because some anchoring, some regularization has to happen. Both the teacher and the student know that it is just to regularize.
What do we seek? What is our learning objective at the end of 10 years? Some competency has to be built. A competency where you can pick up any book and understand it. The exact book that you discuss has no consequence.
If you take a similar analogy: what do I want my students to do? This is not a standalone course. A course on probability theory, a first course on machine learning, and this advanced course — to go out of it, you should be able to build frontier algorithms. Given enough material, if you are given 10,000 GPUs, you should be able to build a GPT Astra, for instance. That’s the kind of competency building I want to do in my course.
Can you build it outside the classrooms? Absolutely you can. That is where my question is. Two years ago this was not the case. The material was all scattered. Somebody had to aggregate all of it and administer it. Now that is also a solved problem. What value do we bring in?
If that is the intention, then we should not make it optional. You’ll have to make attendance compulsory and force students to come to a gym. Some sort of incentive has to be there, and attendance can be a good incentive.
Let’s not be judgmental. There’s no good and bad anymore. I agree. But translate it to what we are doing. We need to make an assumption that students who sit in the class have the fundamentals already in place. Is that a fair assumption to make?
All that can be done. Grading is not an issue. I’m not even talking about evaluation. I’m talking about the day-to-day activity that we do — these three hours every week that we spend together. What is it that we need to do?
Assume that all of you know what a transformer is, and if I just say “use an SSM here with some of these things,” you know what it is. So I treat you like peers. I have discussions with my senior PhD students — that’s the kind of discussion I should do in a class. Is that a fair thing to do?
From next class I will do the same boring thing, don’t worry. But today is especially important because this afternoon we have that workshop, and I really wanted to have this discussion with people sometime. It’s okay. We’ll cover whatever we are supposed to cover. That’s not an issue.
Then you are saying that whatever we are doing right now can continue for years to come because people are not going to put themselves…
What should I cater for? Suppose we have 100 students here. What do we do? Number is also an issue. If there are 100 students in a class, what do we do with those 100 people? How do you even quantify and calculate an average? If the argument is that you cater for an average, how do we know what the level is and how do we quantify the average?
I agree. So now we are talking.
Is there a way to convert this into a one-on-one interaction? How about this: divide the class time into two equal halves. One half is: today you need to know how a diffusion model is defined and how its loss function gets constructed. I give you 45 minutes — use whatever agent you want — and learn it right now. The next 45 minutes we will have an interaction based on that.
That can be done. But again, how do you administer this to a larger community? Not everybody would get a chance to talk. There is no one-to-one interaction again.
Skill Patches and Value Creation
He interned with my company, and we built something called a skill patch — just like you have a Play Store for apps, we do a marketplace for skills, agentic skills. He and one of his interns took all my NPTEL lectures and notes and created “Learn with PR.” It’s a kind of skill which you give to your favorite LLM and the hope is that it will talk like me, teach like me, administer, and then you use that 45 minutes and have an interaction. But still it’s not one-on-one unless I create my clones.
That is how I always define the motivation and objective of my course, and that is why I designed this course. You can ask people who took my course earlier. They said that this goal was met. That’s precisely why I built this course: so that you go back and read any paper in the community and understand what’s happening. That’s why I update this course every year — 40% of this course is whatever has happened in the last year.
But again I’m telling you: competency-wise there are multiple levels of questions. One is why do we even have to get competencies as human beings? Maybe that’s beyond our pay grade. Let’s not talk about it.
Being in a pedagogical system, assuming that we have a goal that we want to meet — which is this — how do we better conduct our classrooms so that the outcomes are met in the age of AI?
I’ll tell you why this is hard. We are not creative enough. There are 100 people in a class. Can you create 100 problems? We don’t have a specific way of judging people. Finally we need to give a grade. That becomes too subjective. If I give you an open problem and ask you to do whatever you want, and all of them just give a prompt to one particular LLM and submit that — how do I evaluate it? Nobody knows.
What are these companies doing? What are these software engineers doing today? That’s why I asked you the economy problem. How would the economy shape up in five years? Do we know? We have no answer right now. What do we train people for? I don’t know. That’s the question I started this discussion with.
What do I train people for? Prompting? Asking the right questions? How do you define what question is right?
If you start questioning things like: what do we define as a human goal as a community, as human beings? Where are we going as a society? What do we have to solve for?
Historically, universities evolved from people like Socrates gathering a few bright minds around him and just discussing all kinds of things. Somewhere that is what we are doing right now. This is how actual learning should be — where you discuss things.
We can actually do that. A course where there are only five people, we work as a team, build the next billion-dollar business. I’m game for it.
Now if 100 people register for that course, and we are in a country where you have people for everything — what do we do? Tell them you are going to get a grade depending upon how many stars you get on GitHub? Build something, go get users. People can game that as well — create some pseudo agents and put some stars, ask friends and relatives. Then I’ll have to say: show some real money, generate revenue. That’s okay. At least you have convinced your father to give money. Some active users. I’m actually game for all this.
How do we create value? How do we make… someone was saying that maybe that is where we would converge: all of us become individual value creators, not as organizations.
If all of us are individual value creators, you rather teach people how to create value. How do you create value?
If you go one level up: business is not about creating values. It’s actually creating the perception of value. You have to make the other person believe that whatever you are going to offer is of value. That becomes more of a human skill. Is that what we are going to teach?
That is the precise question I am asking. Why are you sitting here? That is one way to define it.
The problem with philosophy is that it keeps us in this loop of eternal thinking without putting us into action. The good thing about engineering is that it will put you to action: get up, shut up, and fight.
We are able to talk all this because we are privileged to be sitting in a university. That’s what society has learned. But again, let’s not make it too philosophical. Instead I want to ask directed questions to students who are stakeholders in this entire exercise.
What do you think is valuable for you in a classroom-like setting?
I won’t stop this course. At least this course I will continue because we are going to record and all that. I’ve made that promise. I will continue. But eventually, going forward — for instance next semester — what do I do?
You people come here. At least some 20 people are there. I’m motivated at least to talk. Somebody said not to stop it, and that’s motivation enough for me. If you come, I’ll talk to you. Now if nobody comes to my class, there is something wrong about what I’m doing.
I’m genuinely asking you people because in our societies, unfortunately, all stakeholders are not taken into consideration. That’s why I wanted to know. You can ask your colleagues and friends and tell me what the general thought process among students is. If people have a similar opinion about classrooms, why do we have to conduct the classrooms? Let us wind it up and go home. Big deal.
As Chandan said, we will adapt. Maybe we’ll form something and do something. What’s the whole point of pedagogy? What’s the whole point of a degree and a university right now if you are not catering to the needs of people who come and sit here?
That’s why I seek your opinion here. I want to hear from you: what is it that you feel is valuable?
Individual… different groups meet different groups. So basically the role of an instructor is more of that of an orchestrator. Orchestrate learning. Don’t give lectures. That’s what you’re trying to say.
Do that as well. Do we have that much time? Most of our time will be consumed by lectures themselves. Or maybe record it once and ask them to look at the recordings and come. But these kinds of classrooms that we do — do we need that? That is the question.
Closing Thoughts and Next Steps
We’ll end it with a couple of thoughts.
Just Saturday I was called by one of the schools. They were doing some Teachers’ Day thing and I was called. I asked all the high-school and middle-school teachers this question: suppose you take off attendance from your policy, how many students are going to come to your class?
They were shocked. Can even this be done?
When I used to go to school, specifically high school, I used to perceive that experience as going to a prison. I used to hate that experience. Most of you had a similar experience, I suppose. Why are we creating unhappy places? Why are we creating places where students who are in their formative years — where they should be happy — are spending 8 to 10 hours of their time and coming out in an unhappy state of mind?
If a similar thing is happening in a university, then we are doing something wrong.
I don’t know if I’ve told you this joke. There’s one Kannada language writer — his name was Bichi. In one of his satirical plays he narrates this story: there’s a four- or five-year-old kid walking with his mother. He sees a sheep or a goat being dragged for a sacrifice. They put a lot of garland around its neck and drag it with a band and all that. The kid gets curious and asks his mother why this is getting dragged and where it is getting taken. The mother, with a grim face, replies: unfortunately that poor thing is going to get sacrificed. Now they are taking it to sacrifice. The kid replies: “Oh, that’s all it is? I thought that it is getting dragged to a school.”
The kid perceives death better than going to a school. If a similar thing is happening here…
We already have such signs here. Students don’t come to the class. If you make attendance compulsory, they’ll come and start looking at their laptops. They’ll sit at the back and just open them.
There were no laptops in our times. I used to carry a book secretly always in my bag. I was a backbencher — thanks to my height I was always put at the back bench. Open a book and study all kinds of things. Study philosophy. I’ve studied Socrates, Plato, Aristotle. Just keep studying stuff. I used to hate going to school.
If you create a similar thing — the same model where you have 100 people sitting and one person talking and it’s boring — what kind of environment are we creating?
That’s why I asked you: why do you come? There should be some reason. How many of you are here? Maybe some 25 people right now. There should be something that is interesting for you to come here. I just wanted to understand. If there is something that actually makes you more interested in coming to this kind of a thing, we can adopt it. We are happy to adopt. That’s what a good democratic and nice intellectual place has to be. It has to thrive based on interactions between all stakeholders.
If all it finally boils down to is just that “I need some social validation and some interaction and some environment and that’s why I’m coming,” that’s good too. We can continue what we are doing in terms of talking about some math and all that. Can we do something better than this is the question. Let us have this question and keep asking it. The moment we get some ideas and answers, you can always come back to me — beyond this classroom as well.
I keep telling my students: it’s a relationship that we build over time. My students from IIT Delhi talk to me even today. Just two days back one guy — this guy was JEE Advanced rank 25, very bright, I taught him for three semesters — I told him that after his BTech he went to a quant firm. I was just seeing a tweet yesterday that IIT Delhi Computer Science has 75 admits and there are 75 quant companies that come for recruitment. All of them go there. They are paid extraordinarily well. In those times (2017) they were paying on the order of crores — eight-digit salaries.
I told him: there will be a time in your life when you will be bored of what you will be doing and you will come back and talk to me. That day I will tell you “I told you so.” He sent an email just last week. The subject was: “Sir, I told you so moment.” We had a long conversation. It was nice. He’s in New York now. He said, “Sir, I’m done. I need to do something. Tell me if there are some research problems…”
That’s the kind of relationship that we will build. The fact that you people are still coming here and sitting here has something. You come here for a reason. If you kind of get some ideas where we can improve human lives and pedagogy in specific, let me know. You can always write to me.
Here is an idea that I would want: maybe let me know if you have ideas how to make this experience better for everybody. We’ll experiment with it. Maybe next semester we’ll run a course where we’ll experiment with all those ideas. Don’t worry about grades. Unfortunately if I give grades to all of you I’ll be questioned. It’s again a very unfortunate thing. I’ll have to give that bloody Gaussian distribution to them. That is also a bad thing. Suppose there are 10 people in the class, everybody comes with the intention of learning and having nice discussions — it feels sad to me to actually grade them and put them on a µ + 2σ Gaussian. Nobody will come to me again if I tell them that I will give you a B, C, C-, etc. Whatever nonsense that is. I feel sad for those people who come to class, do something, put some effort, and still get some so-called bad grade because of some bad day. It’s unfair. It’s wrong in my view. But I’ll have to get my salary as well. That’s reality.
Let’s think about it. Next class we’ll continue with all this — diffusion models etc.
One thing I wanted to tell you: I’ll slightly deviate from the given syllabus. Before going to autoregressive models I want to talk about flow-based models because it’s very related to diffusion models. Let’s do flow-based models, specifically the continuous ones. It’s very good. And then we will go to the LLMs. Because the moment we go to that LLM realm, we will not be able to stop — that’s a downward spiral. We’ll have to do RLHF, then some KV cache thing and LoRA and all that. So once the diffusion models are completed (one more class), after that we’ll go to flow-based models. We have enough time.
I did not plan for today’s discussion. I actually planned that I will finish this. If you felt that this is boring, I’m sorry. Sometimes this happens. I never did this in 10 years of my teaching career — never did an entire class on this. But I think time has now come. We have to do it, otherwise we’ll be cheating ourselves.
If you have any comments, let me know. We’ll do better. We’ll have a better experience for everybody — in the world in general. That is a good enough conclusion.
Source
YouTube — Pedagogy in the Times of AI