Harvard professor Bharat Anand explains how AI is changing education by improving access, not just intelligence. He explores its impact on jobs, learning, and the role of teachers.
Category
🗞
NewsTranscript
00:00Speaker 1 – We come now to what is, to my mind, one of the most important conversations
00:07of the moment.
00:10How do we educate our children at a time when machines can do what they can, when every
00:17single query you may have is available at a chat prompt, what skills do we equip our
00:25children with, what do we teach them?
00:27My wife is here.
00:28It's something that we talk about all the time.
00:29So what we did is, we decided to call in one of the best experts in the world, someone
00:34who is researching and thinking very deeply, in fact even probably writing a book on the
00:39future of education.
00:40He is one of the top professors of the Harvard Business School.
00:42He has made a lot of effort to fly down specially to the India Day Conclave to be over here.
00:47Ladies and gentlemen, can we have a very warm round of applause as I welcome Professor Bharat
00:51Anand.
00:52He is the Vice Provost for Advances in Learning, the Henry Byers Professor of Business Administration.
00:58He is an expert on digital strategy, media, entertainment strategy, corporate strategy
01:02and organizational change and at this moment, he is focusing very deeply on the future of
01:08education.
01:09If you got young kids, if you are middle aged yourself, wondering how to upskill yourself,
01:13this is a session you need to pay a lot of attention to.
01:15How we are going to do this is this, Professor Anand will first make a presentation and then
01:20I have lots of questions, I am sure so do you about what do we teach our children, how
01:25do we train them, how should we go about this, so just keep all those thoughts in your head.
01:29We will hand the stage to Professor Anand.
01:31This is a master class.
01:32He said, I don't need to sit anywhere and you can be cold call, okay.
01:35Cold call is, if you are not paying attention to him, he sees because it's around lunchtime
01:40and you seem to be looking at the door, looking around for food, he can cold call you and
01:44ask you some questions.
01:45So, be on standby for that.
01:46With that, Professor Anand, the stage is yours.
01:48Thank you, Rahul.
01:50Good morning.
01:51I need some more energy.
01:53Good morning.
01:55It's a pleasure to be here with all of you today to talk about Gen-AI and education.
02:03For those who don't know what Gen-AI is, imagine a person who is often wrong but never in doubt.
02:12Now, be honest with me, how many of you thought about your spouse?
02:18I did not, okay, but that's Gen-AI.
02:22And what I want to talk about is what happens when we have large language models like JATCPT
02:29and generative AI intersect with institutions like Harvard where I sit and I've been there
02:34for the last 27 years, currently overseeing teaching and learning for the university.
02:40Let me just ask you a question.
02:42How many of you think in the next five to 10 years, generative AI will have a very large
02:48impact on education?
02:50Just raise your hands.
02:52How many would say a moderate impact?
02:55So we have a few.
02:57How many would say little to no impact?
03:01Pretty much none, okay.
03:03Let me come back to this.
03:05Here's a chart showing the rise of technologies and the time it took for different technologies
03:11to reach 50% penetration in the U.S. economy.
03:16So if you look at computers, it actually took 20 years to reach about 30% penetration.
03:23Radio, it took about 20 years to reach half the population.
03:29TV about 12 years, smartphones about seven years, smart speakers about four years, and
03:37chat bots about two and a half years.
03:40This is part of the reason we're talking about this today.
03:44Here's what we know so far about gen AI in education.
03:49First, the transformative potential stems from its intelligence.
03:54That's the I in AI.
03:58Secondly, as prudent educators, we should wait until the output is smart enough and
04:07gets better and is less prone to hallucinations or wrong answers.
04:13Third, given the state of where bot tutors are, it's unlikely, I think many believe,
04:19that it's going to be ultimately as good as the best active learning teachers who have
04:23refined their craft over many, many years and decades.
04:27Fourth, and Sal Khan talks about this, this is likely to ultimately level the playing
04:32field in education.
04:34And finally, the best thing we can do is to make sure that we secure access to everyone
04:39and let them experiment.
04:42Before you take a screenshot of this, don't, because I'm going to argue all of this is
04:48wrong.
04:51Now that I hopefully have your attention, I'm going to spend the next 10 minutes arguing
04:54why.
04:55Let's actually start with the first one, which is the transformative potential stems from
05:01how intelligent the output is.
05:04I would argue, and in fact, we just heard this from the previous speaker, we've been
05:07actually experiencing AI for 70 years, machine learning for upwards of 50 years, deep learning
05:14for 30 years, transformers for seven to eight years.
05:17This has been an improvement gradually over time.
05:20There were some discrete changes recently, but the fundamental reason why this has taken
05:24off, I would argue, has less to do with the discrete improvements in intelligence two
05:29years ago, as opposed to the improvement in access or the interface that we have with
05:35the intelligence.
05:36What do I mean by that?
05:37I'm going to give you the one minute history of human communication.
05:41So we started out sitting around campfires, talking to each other.
05:45From there, we started writing pictures on the walls.
05:49That was graphics.
05:51From there, we start writing scrolls and books.
05:54That was formal text.
05:55And finally, the pinnacle of human communication, which was ones and zeros, and that's mathematics.
06:02That's the evolution of human to human communication.
06:05The evolution of human to computer communication has gone exactly in the opposite direction,
06:09which is 60, 70 years ago, starting with punch cards, ones and zeros.
06:13For those of you old enough, might remember that.
06:16Then we moved to things like DOS prompts, commands that we had to input.
06:21By the way, and this is the fundamental thing, the big difference between Windows 1.0 and
06:25Windows 3.0, functionally, they were almost identical.
06:30The big difference was the interface, meaning we moved to a graphical user interface and
06:35suddenly 7-year-old kids could be using computers.
06:38That I think is more similar to the revolution we're seeing now, which is AI for a long time
06:43was the province of computer programmers, software engineers, tech experts.
06:48With chat GPT, it basically became available to every one of us on the planet through a
06:52simple search bar.
06:54That's basically the reason for the revolution.
06:56Where is this going?
06:58Probably towards just audio.
07:00I don't know if anyone can guess what's the next evolution of this in terms of communication.
07:04Neural, reading emotions.
07:10You might argue basically us grunting and shaking our arms, formerly that would be called
07:15the Apple Vision Pro.
07:19You could argue we are regressing as a species.
07:22On the other hand, you could argue that in fact what's happening is that the distance
07:26between humans and computers is fundamentally shrinking.
07:30That's the first thing I just want to say, which is fundamentally this is about access.
07:35What does this mean?
07:37It means that, does anyone know what this is?
07:42This is Photoshop.
07:44There's a lot of people who spend one year, two years, four years trying to master this,
07:48graphics design.
07:50Arguably, we don't need this kind of expertise anymore.
07:53We can simply get it by communicating directly in natural language with computers now.
07:59This, for those of you who don't know, is Epic.
08:01It's a medical software record.
08:03My wife, who's a cardiologist, does not like this.
08:06She spends two hours every single day filling in notes on these software records.
08:12You could argue sometime in the near future that communication will become much simpler.
08:17By the way, one of the things to keep in mind is for every one of you sitting in organizations,
08:23by the way, this is a happy organization, to think about what this is likely to do to
08:27the org structure.
08:30If you think about the bottom of this organization, there's people who have expertise in different
08:34kinds of software.
08:36Some expertise in Photoshop, some in Concur, some in different kinds of software.
08:44You could argue there's going to be consolidation within those functions.
08:47The middle managers who used to oversee all these software experts, it's likely we're
08:52going to see shrinkage there.
08:54In fact, you could argue all the way that the person at the top could, in fact, do sales,
08:59graphics design, design, marketing, everything by just interacting directly with the computer.
09:04It's not a stretch to say, and some people predict this, that the first one-person, billion-dollar
09:08company is going to be likely to be born pretty soon.
09:12People are already working on this.
09:14I would urge you to think about this question, which is what does this mean for your expertise
09:19in organizations or the organizations you run?
09:22Because that's going to have big implications for how you run these organizations.
09:26All right.
09:27That's the first point, which is fundamentally this is not about intelligence, but it's how
09:30it's accessed.
09:32The implication of this is more people will be able to use more computers for specialized
09:36purposes, but it doesn't necessarily mean it's likely to be the same people.
09:42That's the first thing.
09:44Second, I think we all look at these hallucinations and we say, let's wait.
09:50Let's wait until it gets better.
09:52By the way, that begs the question that hallucinations are a fundamental intrinsic property of generative
09:57AI because they're probabilistic models.
10:00But I would go further and say even when AI capabilities fall far short and impair the
10:06human value proposition, there's still a reason to adopt it.
10:10Why do I say that?
10:12I'm a strategist.
10:13A strategist, we think of two sides of the equation.
10:17One is the benefit side.
10:19What are customers willing to pay?
10:20The other is the cost or the time side.
10:24Even if there's no improvement in intelligence, simply because of cost and time savings, there
10:30might be massive benefits to trying to adopt this.
10:33The metaphor I want you to think about is the following company.
10:40Has anyone flown Ryanair?
10:45What is the experience like, Ishan?
10:49Basic efficient.
10:50By the way, when I ask my students this, they often say, I hate it every single time I fly.
10:56And of course, it begs the question, why are you repeatedly flying it?
10:59This is an airline, like most low cost airlines.
11:03It doesn't offer any food on board, no seat selection.
11:06You've got to walk through the tarmac.
11:08You've got to pay extra for bags.
11:09No frequent flyers, no lounges.
11:11And this is the most profitable airline in Europe for the last 30 years running.
11:16Why?
11:17It's not providing a better product.
11:20It's saving cost.
11:23That's the metaphor I would love for you to keep in mind when you think about generative
11:26AI and its potential.
11:28So let me just walk through this.
11:29And sorry, as a strategist, I have to put up a two by two matrix at some point.
11:34There's two dimensions here I'd love for you to think about.
11:36The first is, what is the data that we're inputting into these large language models?
11:42And the data could be explicit in the form of files, like text files, numbers, et cetera.
11:48That's explicit data.
11:50Or it could be tacit knowledge, meaning creative judgment, et cetera, et cetera.
11:56But the second dimension is as important, which is, what's the cost of making an error
12:02from the output?
12:04Not the prediction error.
12:06What's the cost of something going wrong?
12:09In some cases, it could be low.
12:10In some cases, it could be high.
12:12So let's actually talk through some examples.
12:15First is, explicit data, low cost of errors, that's high volume customer support.
12:21For the last 30 years, this thing has been automated.
12:23By the way, that trajectory is likely to continue.
12:26Why do I say that?
12:28It is virtually impossible for any company to have people manning the phones to talk
12:32to 100,000 customers.
12:34This is the direction where it's going.
12:37Even if we have 2% or 3% or 4% errors, it's OK.
12:41It's simply much more efficient to respond to customers in this way.
12:45So that's one dimension.
12:47Second dimension is drafting legal agreements.
12:51For all the lawyers in the room, just watch out.
12:53It's going to be much, much easier.
12:55It already is to draft legal agreements.
12:57But we can't rely on generative AI to simply give us this thing without checking it.
13:04Some of you may have heard of that lawyer who did that a couple of years ago.
13:08Basically didn't review the agreements.
13:10There were some errors.
13:11He got fired.
13:12So we might have human in the loop.
13:14You don't want to basically take the output at face value, OK?
13:18Because the cost of making an error is simply too high.
13:21Third, on the top right, is creative skills, design, marketing, copywriting.
13:28These are things where it's hard to evaluate what's truly better or worse.
13:33And so in some sense, the design outputs we get, the social media content we get as
13:39suggestions from generative AI, pretty good.
13:42The cost of making an error there, not that high.
13:45And finally, we get to the top right, where we want to be very, very careful.
13:50Because this is like large enterprise software integration.
13:54You don't want to go there pretty soon, OK?
13:56Or designing an aircraft.
13:58Now, what does it mean for education?
14:00Let's actually play this out.
14:02I'm going to use our example as an illustration.
14:05If I'm sitting at Harvard, basically we get, when we open up the website, about 10,000
14:13applications in the first couple months for admission.
14:16Maybe 30,000 people who look at the website.
14:19By the way, they have questions.
14:20It's impossible to speak personally and individually to everyone who has a question.
14:26This is beautiful for chatbots to be able to simply respond.
14:31Again, if there's an error in the response, it's OK.
14:35I mean, these are people who are simply thinking about applying, and they might find information
14:40in other ways.
14:41Secondly, legal contracts with food contractors.
14:45We want to be careful about human in the loop.
14:47Thirdly, designing social media content, when we go to the top left.
14:51This is something we can do far more efficiently today with generative AI.
14:55And finally, I can assure you, we're not going to be using this anytime soon for hiring faculty
15:00or disciplinary actions against students.
15:02By the way, think about this not just for your organization.
15:05Think about it for you individually.
15:07So if I was to do that, responding to emails.
15:11I get a lot of emails every day.
15:15Most of these emails are things that are very standard.
15:18Professor, when are your office hours?
15:20Where is the syllabus posted?
15:22By the way, even in other cases where students ask questions, like, Professor, I have two
15:26offers.
15:27One from McKinsey, one from Boston Consulting Group.
15:32The cost of an error is not that high in my response.
15:35You'll be OK.
15:36Or I'm trying to decide whether to go to Microsoft or Amazon.
15:40You'll be OK.
15:41OK, I'm just kidding, by the way.
15:42I can assure you I respond to all those emails individually.
15:45But you get the point.
15:48Writing a case study.
15:49It takes us nine months to write these famous Harvard Business School case studies.
15:53The head of the MBA program last year said, I want to teach a case on Silicon Valley
15:57Bank tomorrow.
16:00What he did was go to Chachi P.T., said, write a case like Harvard Business School with these
16:05five sections, financial information, competitor information, regulatory information.
16:10It spits it out.
16:11He then said, please tweak the information.
16:14Give me this data on the financials.
16:16Talk about these competitors.
16:18He iterated.
16:19It kept spitting out information.
16:21From beginning to end, he had a case study complete in 71 minutes.
16:28If you're not scared, by the way, we are about what the potential here is.
16:33Brainstorming a slide for teaching.
16:34There's a couple slides in this talk where I took some pictures and I started trying
16:38to resize it.
16:40PowerPoint designers simply threw up some suggestions saying, here's how you might want
16:44to do it in one second.
16:46It didn't take me 10, 15 minutes to try and redesign these slides.
16:49A beautiful application for using this.
16:52And finally, thinking about exactly how I teach in the classroom or my research direction,
16:56I'm not going there anywhere soon.
16:59I'd love for you to think about a couple things from this simple framework.
17:04Number one, we are obsessed with talking about prediction errors from large language models.
17:10I think the more relevant question is the cost of making these errors.
17:14Meaning, in some cases, the prediction error might be 30%.
17:19But if the cost of error is zero, it's okay to adopt it.
17:23In other cases, prediction errors might be only 1%, but the cost of failure is very high.
17:30You want to stay away.
17:32So stop thinking about prediction errors.
17:33Let's start thinking about the cost of errors for organizations.
17:36Secondly, if you notice what I've done, I've broken down the analysis from thinking about
17:41industries.
17:43What's the impact of AI on banking or education or retail into jobs?
17:49And in fact, gone a step further and broken it down into tasks.
17:53So don't ask the question of what is AI going to do to me.
17:57Ask the question, which are the tasks that I can actually automate?
18:00And which are the tasks I don't want to touch?
18:03And the third is, I don't know about you, in my LinkedIn feed, every single day, I get
18:08new information about the latest AI models and where the intelligence trajectory is going,
18:14getting better and better.
18:15That's basically about the top right cell.
18:19I would say that's a red herring for most organizations, because basically there's three
18:23other cells where you can adopt it right now and today with human in the loop.
18:29So that's just something I'd love for you to think about.
18:32By the way, we did this with Harvard faculty, where we interviewed 35 Harvard faculty who
18:38were using Gen AI deeply in their classrooms.
18:42Those videos are up on the web.
18:43If you just type in Google generative AI faculty voices Harvard, you see all these videos.
18:49Here's some examples of what they were doing.
18:51A faculty copilot chatbot.
18:53It's almost like a teaching assistant that simulates the faculty that answers simple
18:59questions and is available to you 24-7.
19:02Secondly, one of the things that we as faculty spend a lot of time thinking about is designing
19:09the tests and the quizzes and the assessments every year.
19:13And we've got to make it fresh, because we know our students probably have access to
19:17last year's quizzes.
19:20Large language models are basically spitting this out in a couple minutes.
19:23And of course, as individuals, we would refine it.
19:26We're not going to just take it at face value.
19:28We refine it.
19:29We look at it.
19:30But it's saving a lot of time.
19:31Third, when we're giving lectures, students often have questions which they're too scared
19:37to ask live in front of 300 students.
19:40Oh, it's beautiful if we can simply type in the questions, have Gen AI summarize the questions
19:46and put it up on a board.
19:47The faculty know exactly what the sentiment is in the classroom and where students are
19:51getting confused.
19:52By the way, notice one thing about all these examples.
19:57Every single one of them is about automating the mundane.
20:02It's not about saying, let's rely on the intelligence that's getting better and better.
20:06It's the left column of the framework I was talking about.
20:10So these are ways that it's being used nowadays in our classrooms.
20:15The third thing, this premise that bot tutors are unlikely to be as good as the best instructors.
20:22We had a few colleagues at Harvard who tested this for a course called Physical Sciences
20:272.
20:28This is one of the most popular courses.
20:30And by the way, the instructors are very good in that course.
20:32They've been refining active learning teaching methods for many years.
20:36What they did as an experiment was say, for half the students every week, we'll give them
20:42access to the human tutors.
20:44For the other half, give them access to an AI bot.
20:47And by the way, the nice thing about the experiment is they flipped that every single week.
20:51So some people always had access to the humans.
20:54Some people had access to the AI for that week, but then they'd flip the next week.
20:58Every single week, they tested your mastery of the content during that week.
21:04And what was interesting was the scores of the students using the AI bots were higher
21:12than with the human tutors.
21:14And these are tutors who've been refining their craft year in and year out.
21:18What was even more surprising is engagement was higher.
21:22By the way, this is a first experiment.
21:24The only point is we better take this seriously.
21:28Next, will it level the playing field in education part of the premises?
21:34Because everyone has access.
21:36Any individual in a village, a low income area is basically going to have access to
21:41the same technology as those who are in elite universities.
21:46And this is going to level everything.
21:49There's a possibility it might go exactly the other way, which is the benefits might
21:54accrue disproportionately to those who already have domain expertise.
21:59Why do I say this?
22:00Think about a simple example.
22:02When you have knowledge of a subject, and you start using generative AI or chat GPT,
22:08the way you interact with it, asking it prompts, follow on prompts, you're basically using
22:14your judgment to filter out what's useful and what's not useful.
22:18If I didn't know anything about the subject, I basically don't know what I don't know.
22:23So in some sense, the prompts are garbage in, garbage out.
22:26By the way, this is being shown in different studies.
22:29There was a meta-analysis summarized by The Economist a couple of weeks ago, where they
22:34basically talk about different kinds of studies that are showing for certain domains and expertise,
22:40the gap between high-performance, high-knowledge workers and no-knowledge workers is actually
22:45increasing.
22:47We better take this seriously.
22:48Why?
22:49And this is not the first time this has happened.
22:52Twelve years ago, there was a big revolution in online education.
22:56Harvard and MIT got together, created a platform called edX, where we offered free online courses
23:02to anyone in the world.
23:04By the way, they still exist.
23:06If you want to take a course from Harvard for free, pay $100 for the certificate, you
23:11can get it on virtually every subject.
23:14What happened as a result?
23:16edX reached 35 million learners, as did Coursera and Udacity and other platforms.
23:21What was beautiful is roughly free 3,000 courses.
23:26The challenge was completion rates less than 5%.
23:30Why?
23:31By the way, if you're used to a boring lecture in the classroom, the boring lecture online
23:35is 10 times worse.
23:37So there's virtually no engagement.
23:39People take a long time to complete or may not complete.
23:41But here's what's interesting.
23:44The vast majority, 75% of those who actually completed these courses already had college
23:50degrees, meaning the educated rich were getting richer.
23:55Now think about that.
23:56That's very sobering.
23:57Why is that?
23:59Because those are people used to curiosity, intrinsic motivation.
24:03By the way, they're used to boring lectures.
24:04They've gone to college.
24:06But this has big implications for how we think about the digital divide.
24:09So I just want to keep that in your mind.
24:12And the last thing I just want to say is, rather than going out and trying to create
24:16tutor bots for as many courses as possible, I think what we really need to do is have
24:20a strategic conversation about what's the role and purpose of teachers, given the way
24:26the technology is proceeding.
24:28The one thing I will say here is that when we think about what we learned in school,
24:34okay, think back.
24:35Think back many, many years.
24:38We learned many things.
24:41Tell me honestly, how many of you have used geometry proofs since you graduated from high
24:47school?
24:49Three people.
24:52Why did we learn state capitals and world capitals of every single country?
25:00Foreign languages.
25:01And by the way, this is Italian.
25:03Devi is not a goddess.
25:05Devi in Italian says, you must.
25:08They have similarities.
25:11Why did we learn foreign languages?
25:13When we think about business concepts in our curriculum, I often get my students who come
25:17back 10 years later and say, those two years were the most transformative years of my life.
25:21I often ask them, what were the three most important concepts you learned?
25:26They said, we have no idea.
25:27I'm like, no, no.
25:28Okay, give me one.
25:29No, no.
25:30We have no idea.
25:31I'm like, so why do you say this was transformative?
25:33The point simply being they're saying this was transformative not because of the particular
25:37content, but because of the way we were learning.
25:40We were forced to make decisions in real time.
25:43We were listening to others.
25:44We were communicating.
25:46What are they saying?
25:47They're saying that the real purpose of case method was listening and communication.
25:52The real purpose of proofs was understanding logic.
25:56The real purpose of memorizing state capitals was refining your memory.
26:00By the way, that example there is the poem If by Rudyard Kipling.
26:04Some of you might remember this from school.
26:06It goes something like this.
26:07If you can keep your head when all about you are losing theirs and blaming it on you.
26:11I have PTSD because my nephew, when he was reciting this to me preparing for his 10th
26:15grade exams, I was like, what the heck are you doing?
26:17But it was basically refining memory skills.
26:20And for foreign languages, it was just learning cultures and syntax.
26:23When we go deep down and think about what we were actually teaching, I think that probably
26:29gives us a little more hope.
26:31Because it means it doesn't matter if some of these things are probably accessible through
26:35gen AI.
26:37When calculators came along, we thought it's going to destroy math skills.
26:41We're still teaching math, thankfully, 50 years later, and it's pretty good.
26:44So this is something that I think is going to be an important strategic conversation.
26:48This is the slide I'd love for you to keep in mind, which is basically everything I've
26:52just said.
26:53If you want to take a screenshot, this is the slide to take a screenshot.
26:57Thank you all so much.
26:59And I hope to be in touch.
27:06At HBS, I took Professor Anand's class on economics for managers listening to him feels
27:10like being back in class.
27:11Fortunately, he didn't cold call anyone, which is terrific.
27:14So thank you for that.
27:15I have a few questions.
27:17We've got young children, and you've got so much of knowledge available now on chat prompts.
27:25What's your advice to everyone who's got young children now wondering about what should they
27:28be teaching their children so that when they grow up and when we don't know what the actual
27:33capabilities of these machines are, that what they've learned is still useful?
27:38How old are your kids, Rahul?
27:39So my son is nine, and my daughter's five.
27:42What are you telling them right now?
27:43Now, I want to learn from you, and we're telling them a lot of stuff, whether good, bad, ugly,
27:47I don't know.
27:48I'm trying to refine that and give them a framework of what we should be telling them.
27:51So there's two things.
27:52So I think, first of all, this is probably one of the most common questions I get.
27:57By the way, it's really interesting that the tech experts, and there was an article in
28:02the Wall Street Journal about this 10 days ago, are basically telling their kids, don't
28:06learn computer science.
28:10That skill, at least basic computer programming, is gone.
28:14Advanced computer science, advanced data analysis, if you want to do that, that's going to be
28:18fine.
28:19What are they telling their kids to learn?
28:21They're telling their kids to learn how to teach dance.
28:24They're telling their kids to learn how to do plumbing.
28:27They're telling their kids to learn about the humanities.
28:31Why are they saying that?
28:32Implicitly, they're saying, what are those skill sets that are robust to machine intelligence?
28:40Now I will say it is virtually impossible to predict that, given the pace at which this
28:44improvement is occurring.
28:46I probably have a slightly different kind of answer.
28:48By the way, my daughter's majoring in psychology, without me telling her anything.
28:53So the kids, I think, know basically where this is going.
28:56But the one thing I'll say, Rahul, is I don't know, when you started out college, what were
28:59you majoring in?
29:00Journalism.
29:01Journalism.
29:02You started out with journalism.
29:04OK, that's enlightened.
29:06I started out doing chemistry.
29:09And then the reason I switched to economics was probably like many of you.
29:13There was one teacher who inspired me.
29:16And that's what made me switch.
29:19And I would say to kids, follow the teachers who inspire you.
29:24And the reason is, if you can get inspired and passionate about a subject, that's going
29:28to build something that's going to be a skill that will last all your life, which is curiosity,
29:35which is intrinsic motivation.
29:36We talked about it in the last session.
29:38This is no longer about learning episodically.
29:41It's about learning lifelong.
29:43And that's, I think, going to be the most important skill.
29:45But in the way that Indian families operate, and as do so many Asian families too, parents
29:50want to equip their children with the skills that are likely to be most useful when they
29:55grow up.
29:56So it used to be, say, engineering and doctors back in the day, then IT a few years ago.
30:03So if you were looking ahead, what do you think the children should be learning so they
30:07acquire skills which are useful in the job market years down?
30:11I think that's honestly being too instrumental.
30:14As I said, 10 years ago, a lot of my students were talking to me and saying, what should
30:17I major in?
30:18I never told them computer science.
30:20If I told them that, I would have regretted it.
30:22But I genuinely mean this.
30:24That's looking at things too narrowly.
30:26What I would say is think about things like creativity, judgment, human emotion, empathy,
30:32psychology.
30:33Those are things that are going to be fundamentally important regardless of where computers are
30:37going.
30:38By the way, you can get those skills through various subjects.
30:41It doesn't matter.
30:42It's not a one-to-one mapping between those skills and a particular topic or disciplinary
30:46area.
30:47This is partly what I'm saying.
30:48Really think about where their passion is.
30:50How do we teach our children how to think?
30:52Because everything's available on Google, Copilot, ChatGPT.
30:56You can just ChatGPT it.
30:58Joining the dots, giving them a framework to be able to interpret, analyze, and think,
31:04how do you tell them that when the easiest thing is, let's Google it?
31:10It's a good question.
31:11Just two things on that.
31:13The first is, there was an interesting study done by colleagues at MIT recently where they
31:18had groups of students and they were asked to undertake a particular task or learn about
31:23a topic.
31:25Some students were given AI chatbots.
31:28Some students were only given Google Search with no AI.
31:33What they found is the students with access to AI intelligence learned the material much
31:39faster, but when it came time to apply it on a separate test, which was different from
31:45the first one, they found it much harder.
31:48The students who learned the material through Google Search with no other access took longer,
31:55but they did much better on those tests.
31:57Why is that?
31:58Part of the issue is learning is not simple.
32:02It takes effort.
32:03Okay?
32:04And so part of the issue is you can't compress that effort.
32:10The harder it is to learn something, the more likely you'll remember it for longer periods
32:16of time.
32:18So I think for me, the big implication is when I tell my students, look, all these technologies
32:22are available, it depends on how you use it, my basic approach to them is just saying study.
32:31Because if you get domain expertise, you will be able to use these tools in a much more
32:36powerful way later on.
32:39So in some sense, this goes back to the notion of agency.
32:43It's like we can be lazy with tools and technologies or we can be smart.
32:47It's all entirely up to you.
32:50But this is my advice.
32:51You know, some of my friends in Silicon Valley have the toughest controls on their children
32:56when it comes to devices.
32:58You know, we look at how much time our children can spend on their iPads or TV, we're far
33:02more lenient.
33:03And they're the guys who are actually in the middle of the devices and they're developing
33:06them and they know the dangerous side effects.
33:08Now, those devices are also the repository of knowledge, which is where you can learn
33:12so much from.
33:13So as an educator, every parent has his own take on how much time children can spend.
33:17But as an educator, how do you look at this device addiction, just spending far more time
33:21picking up some knowledge but also wasting a lot of time?
33:24Yeah, I think, I mean, there's a nuance here, which is basically what they're doing is not
33:29saying don't use devices.
33:31They're saying don't use social media.
33:34And this goes back again to one of the things we were talking about earlier.
33:38We have gone through a decade where things like misinformation, disinformation, and
33:43so on, there is no good solution as far as we know today.
33:47There's also various other kinds of habits and so on that are getting approved.
33:51That's partly what they're saying stay away from.
33:53They're not saying stay away from computers.
33:54We can't do that.
33:55And in fact, you don't want to do that.
33:57But there's a nuance in terms of how we interact with tools and computers that we just want
34:01to keep in mind when we think about guardrails, right?
34:04Are you seeing your students getting more and more obsessed with their devices?
34:08And how does that impact?
34:09What are you trying to do to get them to socialize more?
34:13You know, to spend more time with each other and not be stuck on their phones?
34:16Yeah, it's a very interesting question.
34:17So in some sense, last year, we had a conference at Harvard.
34:21We had 400 people from our community attend the conference.
34:25And some of our colleagues were saying, we should have a policy of laptops down.
34:29No laptops in class, take out our devices.
34:33I was coming in for a session right afterwards.
34:35But part of the reason I wanted them to take out their mobile phones was I had two or three
34:40polls during my lecture where I wanted them to give me their input.
34:44So I said, mobile phones out.
34:47And this was sort of crazy.
34:49But the story illustrates something interesting, which is these devices for certain things
34:53can be really powerful.
34:54It can turn a passive learning modality into an active learning modality where every single
34:59person is participating.
35:01We don't want to take that away.
35:03What we want to try and deal with is people playing games while you're lecturing.
35:07Now, by the way, me personally, I just put it on myself.
35:11If I'm not exciting enough or energizing enough for my students to be engaged, use your mobile
35:15phones.
35:16That's on me.
35:18But that's partly what the challenge is.
35:19No, no.
35:20They're quite engaged.
35:21Show of hands.
35:22How many felt engaged during the session?
35:23And how many were like,
35:24I'm hungry.
35:25When's lunch?
35:26Tell us.
35:27OK.
35:28So that, which is why agentic AI and chatbots can never do what professors can, right?
35:34So I'll take some questions.
35:36Kali has a question.
35:37Kali, go ahead.
35:38Hi, professor.
35:40You mentioned that one of the things that we should work on to teach our children is
35:44empathy.
35:46How do you actually teach empathy in our formal education system?
35:51Or does this just go back to then parents and family?
35:57It's a hard question.
36:01In fact, this is, by the way, one of the most important issues we're facing today on campuses.
36:07It's related in part even in higher education, not just younger kids.
36:11When we talk about difficult conversations on campus, part of the reason we're facing
36:17those issues is because people are intransigent.
36:20It's like, I don't care what you say.
36:24I'm not going to change my mind.
36:26One of the things we introduced a couple of years ago on the Harvard application for undergraduate
36:31is a question that says, have you ever changed your mind when discussing something with anyone
36:36else?
36:37Something to that effect.
36:39That's basically saying how open-minded are we?
36:42That's one version of empathy.
36:43There's many other dimensions.
36:45I think part of the challenge is that we don't teach that in schools, right?
36:51We don't teach that formally in schools, which is partly why there's this whole wave now
36:55of schools, not just in other countries in India, which are starting to talk about how
37:00do we teach the second curriculum, the hidden curriculum?
37:03How do we teach those social and emotional skills, the book of life, so to speak?
37:07And I think, I mean, it's not rocket science to say this.
37:12It starts at home, right?
37:13That's basically what we do with our kids every single day.
37:18But that's something that's, I think, going to become fundamentally more important, partly
37:21because of the reasons of what I talked about.
37:23Dr. Sanjeev Baga, he has a question.
37:25Okay, I see lots of hands going up.
37:28Yes, Dr. Baga.
37:31Wonderful, wonderful listening to you.
37:33Just with regards to AI and technology, I've always said that AI and digital technology
37:39is not an expenditure.
37:41It's actually an investment.
37:43So, very quickly, if you'll allow me just 60 seconds, in healthcare, it gives you better
37:48clinical outcomes.
37:49It has decreased from number one cause of death as hospital-acquired infections in many
37:55hospital chains as practically less than 1%.
37:59So it gives you a safer outcome.
38:01It gives you a better patient experience.
38:03The turnaround of the bed strength is a lot quicker.
38:06And more importantly is it gives you better operational excellence.
38:09So all the hospitals, as far as medical facilities are concerned, who have not embraced it as
38:14yet, will find it difficult to operate in the present environment.
38:18What AI and digital technology has made us learn as doctors is that data is the new gold.
38:24If you don't analyze data, if you don't see what your results are, if you don't see where
38:28your clinical outcomes are, then you can't go forward.
38:31So AI is what is in the future for us, all of us.
38:35That's more in the form of an observation.
38:36Let me just elaborate on that in two ways.
38:38One is, I think I would just go back, and useful to contextualize AI, right?
38:44Like right now, we often get obsessed by the latest technology.
38:48When we think about upskilling, reskilling in education, there's a revolution that started
38:53a decade ago.
38:54As I alluded to, there's basically 3,000 courses available to all of you today on any subject.
39:01So the notion of let's wait for AI, no, no, no, it's already there.
39:05My father-in-law, who's 92 years old, during COVID, he said, Bharat, what should I do?
39:09I said, we have all these courses from Harvard available.
39:12In the last two years or three years, he's completed 35 courses.
39:17Wow.
39:18OK, at the age of 92.
39:19Wow.
39:20By the way, he's paid $0 for that because he said, I don't need a certificate.
39:25So I told him, you're the reason we have a business model problem.
39:28But that's one aspect.
39:31The second aspect is sort of thinking about where you're going.
39:34I think you're exactly right, Sanjeev, which is every organization is going to have low-hanging fruit.
39:39The one thing I just caution is there's going to be a paradox of access, meaning if every
39:45organization, every one of your peers has access to the same technology as you,
39:50it's going to be harder for you to maintain competitive advantage.
39:54That's a fundamental question.
39:56OK, this is just a basic observation.
39:58So I just want to sort of mention that.
40:00But you're absolutely right about the low-hanging fruit in medicine and health care.
40:04OK, Toby Walsh has a question or an observation, and then we either lots of hands up.
40:08OK, I don't frankly know what to do because we're also out of time.
40:11So let this just be where we conclude.
40:13One of the greater challenges, especially in higher education, is the cost has gone through the roof.
40:18Are you optimistic that AI is going to be able to turn that around?
40:23So again, I'll just go back to what's happened in the last decade.
40:28As I said, you can now get access to credentials and certificates at a minimal cost compared to the cost of getting a degree.
40:36OK, just to put it in perspective, we have 17,000 degree students every year who come to Harvard.
40:42They are paying a lot of money.
40:44Those who need financial aid get financial aid.
40:47By the way, can anyone guess how many students we have touched over the last decade?
40:5310 times, 100 times that.
40:55It's about 15 million.
40:57That is not a story we publicize.
40:59But that's a story about the number of students who have actually taken a Harvard course or enrolled in a Harvard course.
41:05So in some sense, I think where we are today is the marginal cost of providing education is very, very low.
41:11What we need for that is not incremental improvement on the existing model.
41:17We need to basically break it apart and say, how do we put it back together again in a way that makes sense for everyone?
41:24There's an organization that we just started at Harvard called Axum, jointly with MIT,
41:29with the endowment from the sale of the edX platform, whose only function is to increase access and equity in education.
41:35And by the way, their focus is on 40 million people in America who start college but never complete it,
41:41not just because of cost, but for many other reasons.
41:44In some sense, the potential to reduce the cost is massive, but it's going to require leadership and strategy.
41:51This gentleman here has a question.
41:53Can someone just take the mic to him, please?
42:02So earlier it was, OK, use AI and it will summarize and help you in productivity.
42:07But with the latest open AI models like O3 Mini and all that, they are doing reasoning which is much better than humans.
42:15So the people who are not using it are at a disadvantage.
42:20So isn't it right that the students use AI and be familiar with it and be up to speed with that
42:29rather than not using it and be at a disadvantage to other students?
42:32Yeah, absolutely. There's no question about that.
42:34By the way, I sit at Harvard overseeing the generative AI task force for teaching and learning,
42:40and we have 17 faculty.
42:42The most interesting conversations I've had about adoption are with our students.
42:48Now, when we understand their behavior, it just throws up things that we wouldn't even have thought about.
42:53I'll ask you one question.
42:55We had a sandbox that we created for the entire Harvard community, which was a safe and secure sandbox,
43:00giving them access to large language models as opposed to using public open AI.
43:04The adoption rate amongst our faculty was about 30%, 35% in the first year.
43:09What do you think the adoption rate was amongst our students?
43:14It was about 5%.
43:17So we were surprised.
43:18When we went to them, we said, what's going on?
43:20Are you familiar with the sandbox?
43:22They said, yeah, we are.
43:24We said, are you using it?
43:25They said, no.
43:26We said, are you using AI in any way?
43:28Yeah, yeah, we have access to ChatGPT.
43:30We have our own private accounts there.
43:32So we're like, wait a minute.
43:33Why are you not using the secure Harvard sandbox?
43:36What do you think their answer was?
43:40They said, why would we use something where you can see what we're inputting?
43:45Now, by the way, as faculty members, if the number one question we talk about with generative AI is,
43:51oh, we're worried about cheating in assessments,
43:53the students are listening to us.
43:54They're like, oh, if that's what you're worried about,
43:56we're not coming anywhere close to you.
43:58So part of the point is the students are far ahead of us in terms of using this.
44:02They're using it to save time.
44:03They're using it for engaging in deep learning.
44:05We better understand that ourselves to figure out what we can do.
44:09Join in, Lukaji.
44:10Brilliant presentation.
44:11Just wanted to understand one side of the spectrum.
44:14You have all the positives.
44:16What's on the other side?
44:18What risk do you think is there on the other side?
44:21It starts coding on its own, gets out of hand.
44:23Is that a possibility?
44:24So the risks are the things I talked about towards the end, which is, number one,
44:30we put our head in the sand as institutions, and we don't take this seriously.
44:35That's the first risk.
44:36The second risk is lazy learning, the way I would call it.
44:40Now, again, that's agency.
44:41It partly depends on you as a student.
44:44Do I want to be lazy?
44:45Do I not want to be lazy?
44:47The third risk is everything we were talking about in the previous session
44:50with respect to misinformation, disinformation.
44:53The fourth big risk is asking the fundamental question,
44:56what's our role as teachers?
44:58And I'll just share one anecdote in closing.
45:00There's a colleague at another school who called me and said,
45:04my students have stopped reading the cases.
45:07They're basically inputting the assignment questions into generative AI.
45:09And by the way, they're so smart, they're saying,
45:11give me a quirky answer I can use in class.
45:14Okay?
45:15The assessments are compromised.
45:17And get this, the faculty have stopped reading cases.
45:20They're inputting the cases and basically saying,
45:22give me the teaching plan.
45:25That's the downside.
45:28You know, we met on a flight from Delhi to Mumbai,
45:31and we had a long conversation about the future of education.
45:33You've been able to, in the past 45 minutes,
45:35recreate the magic of that conversation here on stage.
45:38Can we have a very warm round of applause for the professor?
45:41For making the effort of coming here and for joining us
45:44and for delivering this master class.
45:47Absolute pleasure.
45:48Thank you so much.