AI In Education: Shaping The Future Of Classrooms | Prof. Bharat N. Anand At #IndiaTodayConclave2025
Resumo
TLDRThe talk discusses the intersection of generative AI and education, highlighting the transformative potential of AI technologies like ChatGPT. It explores the rapid evolution of technology and the implications for educational institutions like Harvard. Key points include the need for educators to reconsider traditional methods, embrace AI for efficiency, and ensure equal access to technology. The speaker emphasizes the importance of critical thinking, human judgment, and maintaining the role of educators in fostering social and emotional skills, rather than simply relying on AI capabilities. Furthermore, the discussion touches on the risks of misinformation and the potential widening of educational disparities due to differences in access.
Conclusões
- 🌟 Transformative potential of AI in education is significant.
- 🕒 Generative AI can save time and automate mundane tasks.
- ⚠️ Caution is needed due to hallucinations and misinformation from AI.
- 🧠 Skills like critical thinking and empathy are essential for future generations.
- 👩🏫 Role of teachers will shift towards fostering creative and social skills.
- 📚 Lifelong learning and curiosity should be prioritized in education.
- 💻 Access to technology is crucial for leveling the educational playing field.
- 🚀 AI can streamline administrative duties in educational institutions.
- 💡 Students today are using AI for deeper learning, not just surface-level tasks.
- 📈 Online learning and AI tools have transformed educational accessibility.
Linha do tempo
- 00:00:00 - 00:05:00
The speaker introduces the topic of generative AI and its intersection with education, referencing their role at Harvard and posing questions about the anticipated impact of AI on educational institutions. A chart is presented showing the adoption rates of various technologies, suggesting that generative AI could have a transformative effect similar to past technological advancements.
- 00:05:00 - 00:10:00
The speaker argues that the transformative potential of generative AI relates more to improving accessibility rather than the intelligence of the output. They discuss the evolution of human communication and how AI has only recently become accessible through user-friendly interfaces, highlighting that everyone can now engage with this technology.
- 00:10:00 - 00:15:00
Moving forward, the speaker questions the ongoing reliance on advanced software expertise, suggesting that generative AI could democratize creative and design work and enable easier communication in professional fields, exemplified by the simplification of medical record-keeping through AI.
- 00:15:00 - 00:20:00
The speaker presents a two-dimensional framework for assessing the utility of AI in organizational tasks, emphasizing the importance of both error costs and the type of data input into AI systems, and discusses potential applications in education, such as assisting with queries and drafting educational materials.
- 00:20:00 - 00:25:00
A call for organizations to consider the benefits of generative AI irrespective of current intelligence nuances is made. The speaker uses low-cost services like Ryanair as a parallel to automation in organizations adopting generative AI for efficiency, despite the existing prediction errors.
- 00:25:00 - 00:30:00
In education, large language models can automate mundane tasks effectively while enhancing efficiency. However, the speaker warns about the smaller role human involvement will play as automation increases, prompting a reflection on organizational structures.
- 00:30:00 - 00:35:00
The speaker highlights concerns that generative AI may widen the knowledge divide as those with existing expertise can leverage AI more effectively, which could lead to an upsurge in inequality in educational success. They emphasize the need to prepare students for the digital divide related to generative AI's emerging influence.
- 00:35:00 - 00:44:20
Finally, the speaker encourages thinking critically about the role of education and emphasizes the importance of creativity and empathy as foundational skills for students, stating that a strategic conversation about educational content and delivery methods is necessary in the face of evolving technologies.
Mapa mental
Vídeo de perguntas e respostas
What is generative AI?
Generative AI refers to AI models that can generate text, images, or other content, such as ChatGPT.
How will generative AI affect education in the future?
It will likely transform educational practices by facilitating access to knowledge and automating mundane tasks.
What are some risks associated with generative AI in education?
Risks include lazy learning, misinformation, and potential reliance on AI for critical tasks.
Why should educators be cautious about AI-generated content?
AI can produce hallucinations or incorrect information, necessitating human review.
What skills are important for children to learn in the age of AI?
Skills like creativity, empathy, and critical thinking are essential as they are less likely to be automated.
How can AI be used effectively in classrooms?
AI can automate administrative tasks, assist in responding to common queries, and support personalized learning.
What should be the focus of future education?
Teaching children how to think critically and fostering lifelong learning and curiosity.
How does generative AI compare to traditional teaching methods?
AI can provide faster information retrieval but may lessen engagement and deeper understanding if relied on excessively.
What roles might teachers play in an AI-focused classroom?
Teachers could focus on fostering critical thinking, creativity, and socio-emotional skills, transitioning away from rote learning.
What challenges exist for low-income students regarding AI access?
There's a concern that disparities in technological access may widen the gap in educational opportunities.
Ver mais resumos de vídeos
- 00:00:00I need some more energy good
- 00:00:01morning it's a pleasure to be here with
- 00:00:04all of you uh today to talk about gen
- 00:00:07and
- 00:00:08education um for those who don't know
- 00:00:11what gen is imagine a
- 00:00:15person who's often wrong but never in
- 00:00:20doubt now be honest with me how many of
- 00:00:22you thought about your
- 00:00:24spouse I did not okay but that's
- 00:00:28gen um and what I I want to talk about
- 00:00:31is um what happens when we have large
- 00:00:34language models like chpt and generative
- 00:00:37AI intersect with institutions like
- 00:00:39Harvard where I sit and I've been there
- 00:00:41for the last 27 years currently
- 00:00:43overseeing teaching and learning for the
- 00:00:46University let me just um ask you a
- 00:00:49question how many of you think in the
- 00:00:51next 5 to 10 years generative AI will
- 00:00:54have a very large impact on education
- 00:00:57just raise your
- 00:00:58hands how many would say a moderate
- 00:01:01impact so we have a few how many would
- 00:01:04say little to no
- 00:01:07impact pretty much none okay uh let me
- 00:01:11come back to this here's a chart showing
- 00:01:14the rise of Technologies and the time it
- 00:01:17took for different Technologies to reach
- 00:01:1950% penetration in the US economy so if
- 00:01:23you look at computers it actually took
- 00:01:2620 years to reach about 30% penetration
- 00:01:31radio it took about 20 years to reach
- 00:01:33half the
- 00:01:35population uh TV about 12 years
- 00:01:38smartphones about 7 years smart speakers
- 00:01:41about uh 4
- 00:01:43years and chatbots about 2 and a half
- 00:01:46years this is part of the reason we're
- 00:01:48talking about this
- 00:01:50today here's what we know so far about
- 00:01:53gen and
- 00:01:55Education First the transformative
- 00:01:58potential St from its intelligence
- 00:02:01that's the eye in AI okay
- 00:02:06secondly as prudent Educators we should
- 00:02:10wait until the
- 00:02:12output is smart enough and gets better
- 00:02:15and it's less prone to hallucinations or
- 00:02:17wrong
- 00:02:19answers third given the state of where
- 00:02:22bot tutors are it's unlikely I think
- 00:02:25many believe that it's going to be
- 00:02:27ultimately as good as the best AC
- 00:02:30learning teachers who have refined their
- 00:02:31craft over many many years and
- 00:02:34decades fourth and Sal Khan talks about
- 00:02:37this this is likely to ultimately level
- 00:02:39the playing field in
- 00:02:40education and finally the best thing we
- 00:02:43can do is to make sure that we secure
- 00:02:46access to everyone and let them
- 00:02:48experiment before you take a screenshot
- 00:02:50of
- 00:02:51this
- 00:02:53don't because I'm going to argue all of
- 00:02:55this is
- 00:02:58wrong now that I hopefully have have
- 00:03:00your attention I'm going to spend the
- 00:03:01next 10 minutes arguing
- 00:03:03why uh let's actually start with the
- 00:03:05first one which is the transformative
- 00:03:07potential stems from how intelligence
- 00:03:09the output
- 00:03:10is I would argue and in fact we just
- 00:03:13heard this from the previous speaker
- 00:03:14we've been actually
- 00:03:16experiencing AI for 70 years machine
- 00:03:19learning for upwards of 50 years deep
- 00:03:21learning for 30 years Transformers for 7
- 00:03:23to 8 years this has been an improvement
- 00:03:26gradually over time there were some
- 00:03:28discrete changes recently but the
- 00:03:30fundamental reason why this is taken off
- 00:03:32I would argue has less to do with the
- 00:03:35discrete improvements in intelligence 2
- 00:03:36years ago as opposed to the Improvement
- 00:03:39in Access or the interface that we have
- 00:03:42with the intelligence what do I mean by
- 00:03:44that I'm going to give you the one
- 00:03:45minute history of human
- 00:03:47communication so we started out sitting
- 00:03:49around campfires talking to each other
- 00:03:52from there we started writing pictures
- 00:03:55on the walls that was Graphics from
- 00:03:58there we start writing scroll s and
- 00:04:00books that was formal text and finally
- 00:04:03the Pinnacle of human human
- 00:04:04communication which was ones and zeros
- 00:04:07and that's
- 00:04:08mathematics that's the evolution of
- 00:04:10human to human communication the
- 00:04:12evolution of human to computer
- 00:04:13communication has gone exactly in the
- 00:04:15opposite direction which is 60 70 years
- 00:04:18ago starting with Punch Cards ones and
- 00:04:20zeros for those of you old enough might
- 00:04:22remember that then we move to things
- 00:04:24like dos prompts commands that we had to
- 00:04:27input by the way and this is the
- 00:04:29fundamental thing the big difference
- 00:04:31between Windows 1.0 and windows
- 00:04:343.0 functionally they were almost
- 00:04:36identical the big difference was the
- 00:04:38interface meaning we moved to a
- 00:04:40graphical user interface and suddenly
- 00:04:427-year-old kids could be using computers
- 00:04:45that I think is more similar to the
- 00:04:46revolution we're seeing now which is AI
- 00:04:49for a long time was the province of
- 00:04:52computer programmers software Engineers
- 00:04:54Tech experts with chat GPT it basically
- 00:04:57became available to every one of us in
- 00:04:59the planet through a simple search bar
- 00:05:01that's basically the reason for the
- 00:05:02revolution where is this going probably
- 00:05:05towards just audio and I don't know if
- 00:05:08anyone can guess what's the next
- 00:05:09evolution of this in terms of
- 00:05:14communication neural reading
- 00:05:16emotions you might argue basically us
- 00:05:19grunting and shaking our arms formally
- 00:05:22that would be called the Apple Vision
- 00:05:24Pro uh you could argue we are regressing
- 00:05:28as a species on the other hand you could
- 00:05:30argue that in fact what's happening is
- 00:05:33that the distance between humans and
- 00:05:35computers is fundamentally
- 00:05:36shrinking so that's the first thing I
- 00:05:39just want to say which is fundamentally
- 00:05:40this is about access what does this mean
- 00:05:44it means that does anyone know what this
- 00:05:48is this is
- 00:05:50Photoshop there's a lot of people who
- 00:05:52spend one year 2 years four years trying
- 00:05:54to master this Graphics
- 00:05:57design arguably we don't need this kind
- 00:06:00of expertise anymore we can simply get
- 00:06:02it by communicating directly in natural
- 00:06:04language with computers now this for
- 00:06:07those of you who don't know is epic it's
- 00:06:09a medical software record my wife who's
- 00:06:11a cardiologist does not like this she
- 00:06:14spends 2 hours every single day filling
- 00:06:16in notes on these software records You
- 00:06:19could argue sometime in the near future
- 00:06:21that communication will become much
- 00:06:23simpler by the way one of the things to
- 00:06:25keep in mind is for every one of you
- 00:06:28sitting in organization ations and by
- 00:06:30the way this is a happy organization to
- 00:06:33think about what this is likely to do to
- 00:06:35the or
- 00:06:36structure if you think about the bottom
- 00:06:38of this organization there's people who
- 00:06:40have expertise in different kinds of
- 00:06:42software okay some expertise in
- 00:06:45Photoshop some in
- 00:06:47concur uh some in different kinds of
- 00:06:50software You could argue there's going
- 00:06:52to be consolidation within those
- 00:06:53functions the middle managers who used
- 00:06:56to oversee all these software experts
- 00:06:58it's likely we're going to see shrinkage
- 00:07:00there in fact you could argue all the
- 00:07:02way that the person at the top could in
- 00:07:05fact do sales Graphics design design
- 00:07:07marketing everything by just interacting
- 00:07:09directly with the computer it's not a
- 00:07:12stretch to say and some people predict
- 00:07:13this that the first 1 person billion
- 00:07:16dollar company is going to be likely to
- 00:07:18be born pretty soon okay and people are
- 00:07:20already working on this I would urge you
- 00:07:22to think about this question which is
- 00:07:24what does this mean for your expertise
- 00:07:26and organizations or the organizations
- 00:07:28you run
- 00:07:30because that's going to have big
- 00:07:31implications for how you run these
- 00:07:33organizations all right so that's the
- 00:07:35first point which is fundamentally this
- 00:07:36is not about intelligence about how it's
- 00:07:39accessed the implication of this is more
- 00:07:41people will be able to use more
- 00:07:43computers for specialized purposes but
- 00:07:45it doesn't necessarily mean it's likely
- 00:07:47to be the same people okay that's the
- 00:07:49first
- 00:07:50thing
- 00:07:52second I think we all look at these
- 00:07:54hallucinations and we say let's wait
- 00:07:57let's wait till it gets better
- 00:08:00by the way that begs the question that
- 00:08:01hallucinations are a fundamental
- 00:08:03intrinsic property of generative AI
- 00:08:05because they are probabilistic models
- 00:08:07but I would go further and say even when
- 00:08:10AI capabilities fall far short and
- 00:08:13impair the human value proposition
- 00:08:15there's still a reason to adopt it why
- 00:08:18do I say that I'm a strategist as
- 00:08:21strategist we think of two sides of the
- 00:08:23equation one is the benefit side what
- 00:08:26are customers willing to pay the other
- 00:08:28is the cost or the time
- 00:08:30side even if there's no improvement in
- 00:08:33intelligence simply because of cost and
- 00:08:36Time Savings there might be massive
- 00:08:38benefits to trying to adopt this so the
- 00:08:41metaphor I want you to think
- 00:08:42about is the following
- 00:08:47company has anyone flown Ryan
- 00:08:50Air uh what is the experience like
- 00:08:54isan basic efficient basic efficient by
- 00:08:57the way when I ask my students this they
- 00:08:59often say I hate it every single time I
- 00:09:01fly it and of course it begs the
- 00:09:04question why are you repeatedly flying
- 00:09:06it this is an airline like most lowcost
- 00:09:09Airlines it doesn't offer any food on
- 00:09:12board no seat selection you've got to
- 00:09:14walk to the tarmac you got to pay extra
- 00:09:16for bags no frequent flyers no lounges
- 00:09:19and this is the most profitable airline
- 00:09:20in Europe for the last 30 years
- 00:09:22running why it's not providing a better
- 00:09:26product it's saving cost
- 00:09:30that's the metaphor I would love for you
- 00:09:31to keep in mind when you think about
- 00:09:33generative Ai and its potential so let
- 00:09:35me just walk through this and sorry as a
- 00:09:37strategist I have to put up a 2 x two
- 00:09:39Matrix at some point there's two
- 00:09:42Dimensions here I'd love for you to
- 00:09:43think about the first is what is the
- 00:09:45data that we're inputting into these
- 00:09:47large language models and the data could
- 00:09:50be explicit in the form of files like
- 00:09:52text files numbers Etc that's explicit
- 00:09:56data or it could be tacit knowledge
- 00:09:59meaning
- 00:10:00creative judgment etc etc okay but the
- 00:10:04second dimension is as important which
- 00:10:06is what's the cost of making an error
- 00:10:10from the output not the prediction error
- 00:10:13what's the cost of something going wrong
- 00:10:16in some cases it could be low in some
- 00:10:18cases it could be high so let's actually
- 00:10:20talk through some
- 00:10:21examples first is explicit data low cost
- 00:10:25of Errors that's high volume customer
- 00:10:27support for the last 30 years this thing
- 00:10:29is being automated by the way that
- 00:10:31trajectory is likely to continue why do
- 00:10:34I say that it is virtually impossible
- 00:10:36for any company to have people manning
- 00:10:39the phones to talk to 100,000 customers
- 00:10:42this is the direction where it's going
- 00:10:44even if we have two or three or 4%
- 00:10:46errors it's okay it's simply much more
- 00:10:49efficient to respond to customers in
- 00:10:51this way okay so that's one
- 00:10:53dimension second dimension is drafting
- 00:10:56legal agreements for all the lawyers in
- 00:10:59the room just watch out it's going to be
- 00:11:01much much easier it already is to draft
- 00:11:03legal agreements but we can't rely on
- 00:11:07generative AI to Simply give us this
- 00:11:09thing without checking it some of you
- 00:11:11may have heard of that lawyer who did
- 00:11:13that a couple years ago basically didn't
- 00:11:16review the agreements there were some
- 00:11:17errors he got fired so we might have
- 00:11:20human in the loop you don't want to
- 00:11:22basically take the output at face value
- 00:11:25okay because the cost of making an error
- 00:11:26is simply too
- 00:11:27high third on the top right is creative
- 00:11:32skills design marketing copywriting
- 00:11:36these are things where it's hard to
- 00:11:37evaluate what's truly better or worse
- 00:11:41and so in some sense the design outputs
- 00:11:43we get the social media content we get
- 00:11:46as suggestions from generative AI pretty
- 00:11:48good the cost of making an error there
- 00:11:50not that high and finally we get to the
- 00:11:53top right where we want to be very very
- 00:11:56careful because this is like large LGE
- 00:11:59enterprise software integration you
- 00:12:01don't want to go there pretty soon okay
- 00:12:03or designing an aircraft now what does
- 00:12:06it mean for Education let's actually
- 00:12:08play this out I'm going to use our
- 00:12:10example as an illustration if I'm
- 00:12:13sitting at
- 00:12:15Harvard basically we get when we open up
- 00:12:17the
- 00:12:18website about 10,000 applications in the
- 00:12:21first couple months for admission maybe
- 00:12:2430,000 people who look at the website by
- 00:12:26the way they have questions it's
- 00:12:28impossible
- 00:12:29to speak personally and individually to
- 00:12:31everyone who has a question this is
- 00:12:34beautiful for chat chat Bots to be able
- 00:12:37to Simply respond again if there's an
- 00:12:40error in the response it's okay I mean
- 00:12:43these are people who are simply thinking
- 00:12:45about applying and they might find
- 00:12:47information in other ways secondly legal
- 00:12:50contracts with food contractors we want
- 00:12:52to be careful about human in the loop
- 00:12:54thirdly designing social media content
- 00:12:56when we go to the top left this is
- 00:12:58something we can do far more efficiently
- 00:13:00today with generative Ai and finally I
- 00:13:03can assure you we're not going to be
- 00:13:04using this anytime soon for hiring
- 00:13:06faculty or disciplinary actions against
- 00:13:09students by the way think about this not
- 00:13:11just for your organization think about
- 00:13:13it for you individually so if I was to
- 00:13:15do that responding to emails I get a lot
- 00:13:19of emails every
- 00:13:21day most of these emails are things that
- 00:13:24are very standard Professor when are
- 00:13:26your office ours where's the syllabus
- 00:13:28posted
- 00:13:29by the way even in other cases where
- 00:13:31students ask questions like Professor I
- 00:13:34have two offers one from McKenzie one
- 00:13:36from B Boston Consulting
- 00:13:38Group the cost of an error is not that
- 00:13:41high in my response you'll be okay or
- 00:13:44I'm trying to decide whether to go to
- 00:13:46Microsoft or Amazon you'll be okay okay
- 00:13:48I'm just kidding by the way I can assure
- 00:13:50you I respond to all those emails
- 00:13:51individually but you get the
- 00:13:54point writing a case study it takes us 9
- 00:13:57months to write these famous har
- 00:13:59business school case studies the head of
- 00:14:01the MBA program last year said I want to
- 00:14:03teach a case on Silicon Valley Bank
- 00:14:06tomorrow what he did was go to Chachi PT
- 00:14:09said write a case like Harvard Business
- 00:14:11School with these five sections
- 00:14:14financial information competitor
- 00:14:15information regulatory information it
- 00:14:17spits it out he then said please tweak
- 00:14:21the information give me this data on the
- 00:14:23financials talk about these competitors
- 00:14:25he iterated it kept spitting out
- 00:14:27information from beginning to to end he
- 00:14:29had a case study complete in 71
- 00:14:33minutes um if you're not scared by the
- 00:14:36way we are about what the potential here
- 00:14:39is brainstorming a slide for teaching
- 00:14:42there's a couple slides in this talk
- 00:14:44where I took some pictures and I started
- 00:14:45trying to resize it PowerPoint designer
- 00:14:48simply threw up some suggestions saying
- 00:14:50here's how you might want to do it in 1
- 00:14:52second it didn't take me 10 15 minutes
- 00:14:55to try and redesign these slides a
- 00:14:57beautiful application for using this and
- 00:14:59finally thinking about exactly how I
- 00:15:01teach in the classroom or my research
- 00:15:03Direction I'm not going there anywhere
- 00:15:05soon I'd love you to think about a
- 00:15:08couple things from this simple framework
- 00:15:11number one we are obsessed with talking
- 00:15:14about prediction errors from large
- 00:15:16language models I think the more
- 00:15:18relevant question is the cost of making
- 00:15:20these errors meaning in some cases the
- 00:15:24prediction error might be 30% but if the
- 00:15:27cost of error is zero it's okay to adopt
- 00:15:29it in other cases prediction errors
- 00:15:33might be only 1% but the cost of failure
- 00:15:36is very high you want to stay away so
- 00:15:39stop thinking about prediction errors
- 00:15:41let's start thinking about the cost of
- 00:15:42Errors for
- 00:15:43organizations secondly if you notice
- 00:15:45what I've done I've broken down the
- 00:15:47analysis from thinking about Industries
- 00:15:50what's the impact of AI on banking or
- 00:15:52education or retail into jobs and in
- 00:15:56fact gone a step further and broken it
- 00:15:58down into into tasks so don't ask the
- 00:16:01question of what is AI going to do to me
- 00:16:04ask the question which are the tasks
- 00:16:06that I can actually automate and which
- 00:16:08are the tasks I don't want to touch and
- 00:16:10the third is I don't know about you in
- 00:16:12my LinkedIn feed every single day I get
- 00:16:15new information about the latest AI
- 00:16:17models and where the intelligence
- 00:16:20trajectory is going getting better and
- 00:16:22better that's basically about the top
- 00:16:24right
- 00:16:25cell I would say that's a red herring
- 00:16:27for most organizations
- 00:16:29because basically there's three other
- 00:16:31cells where you can adopt it right now
- 00:16:33and today with human in the loop okay so
- 00:16:36that's just something I'd love you to
- 00:16:38think about by the way we did this with
- 00:16:40Harvard
- 00:16:41faculty where we interviewed 35 Harvard
- 00:16:45faculty who were using gen deeply in
- 00:16:47their
- 00:16:48classrooms those videos are up on the
- 00:16:50web if you just type in Google
- 00:16:52generative AI faculty voices Harvard you
- 00:16:54see all these videos here are some
- 00:16:56examples of what they were doing a
- 00:16:59faculty co-pilot chatbot it's almost
- 00:17:01like a teaching assistant that simulates
- 00:17:04the faculty that answers simple
- 00:17:06questions and is available to you
- 00:17:0924/7 secondly one of the things that we
- 00:17:13as faculty spend a lot of time thinking
- 00:17:15about is designing the tests and the
- 00:17:17quizzes and the assessments every year
- 00:17:20and we've got to make it fresh because
- 00:17:22we know our students probably have
- 00:17:24access to last year's
- 00:17:26quizzes large language models are
- 00:17:28basically spitting this out in a couple
- 00:17:30minutes and of course as individuals we
- 00:17:32would refine it we're not going to just
- 00:17:34take it at face value we refine it we
- 00:17:36look at it but it's saving a lot of time
- 00:17:39third when we're giving
- 00:17:41lectures students often have questions
- 00:17:44which they're too scared to ask live in
- 00:17:46front of 300 students oh it's beautiful
- 00:17:49if they can simply type in the questions
- 00:17:51have gen summarized the questions and
- 00:17:53put it up on a board The Faculty know
- 00:17:55exactly what the sentiment is in the
- 00:17:57classroom and where students are getting
- 00:17:59confused by the way notice one thing
- 00:18:01about all these
- 00:18:03examples every single one of them is
- 00:18:06about automating the mundane it's not
- 00:18:09about saying let's rely on the
- 00:18:11intelligence that's getting better and
- 00:18:13better it's the left column of that
- 00:18:15framework I was talking about so these
- 00:18:18are ways that it's being used nowadays
- 00:18:20in our
- 00:18:21classrooms the third thing this premise
- 00:18:25that bot tutors are unlikely to be as
- 00:18:27good as the best instructors
- 00:18:29we had a few colleagues at Harvard who
- 00:18:32tested this for a course called physical
- 00:18:34sciences 2 this is one of the most
- 00:18:36popular courses and by the way the
- 00:18:38instructors are very good in that course
- 00:18:40they've been refining Active Learning
- 00:18:41teaching methods for many years what
- 00:18:44they did as an experiment was say for
- 00:18:46half the students every week we'll give
- 00:18:49them access to the human tutors for the
- 00:18:51other half give them access to an AI bot
- 00:18:54and by the way the nice thing about the
- 00:18:56experiment is they flip that every
- 00:18:57single week so some people always had
- 00:19:00access to the humans some people had
- 00:19:01access to the AI for that week but then
- 00:19:03they'd flipped the next week every
- 00:19:06single week they tested your Mastery of
- 00:19:09the content during that
- 00:19:10week and what was interesting
- 00:19:13was the scores of the students using the
- 00:19:17AI Bots were higher than with the human
- 00:19:20tutors and these are tutors who've been
- 00:19:22refining their craft year in and year
- 00:19:25out what was even more surprising is
- 00:19:27engagement was higher
- 00:19:29by the way this is a first experiment
- 00:19:32the only point is we better take this
- 00:19:35seriously uh next will it level the
- 00:19:38playing field in education part of the
- 00:19:40premises because everyone has access any
- 00:19:44individual in a village a low-income
- 00:19:47area is basically going to have access
- 00:19:49to the same technology as those who are
- 00:19:51in Elite universities and this is going
- 00:19:53to level
- 00:19:55everything there's a possibility it
- 00:19:58might go exactly the other way which is
- 00:20:01the benefits might ACR
- 00:20:02disproportionately to those who already
- 00:20:04have domain expertise why do I say this
- 00:20:08think about a simple example when you
- 00:20:10have knowledge of a subject and you
- 00:20:12start using generative AI or chat PT the
- 00:20:16way you interact with it asking it
- 00:20:18prompts follow on prompts you're
- 00:20:21basically using your judgment to filter
- 00:20:23out what's useful and what's not useful
- 00:20:25if I didn't know anything about the
- 00:20:27subject I basically don't know what I
- 00:20:29don't know so in some sense the prompts
- 00:20:31are garbage in garbage out by the way
- 00:20:34this is being shown in different studies
- 00:20:36there was a metaanalysis summarized by
- 00:20:38The Economist a couple of weeks ago
- 00:20:41where they basically talk about
- 00:20:42different kinds of studies that are
- 00:20:43showing for certain domains and
- 00:20:46expertise the gap between high
- 00:20:50performance High knowledge workers and
- 00:20:51no knowledge workers is actually
- 00:20:53increasing we better take this seriously
- 00:20:56why and this is not the first time this
- 00:20:57has happened
- 00:20:5912 years ago there was a big revolution
- 00:21:01in online education Harvard and MIT got
- 00:21:04together created a platform called edex
- 00:21:07where we offered free online courses to
- 00:21:10anyone in the world by the way they
- 00:21:12still exist if you want to take a course
- 00:21:14from Harvard for
- 00:21:15free pay $100 for a certificate you can
- 00:21:19get it on virtually every subject what
- 00:21:21happened as a result edex reached 35
- 00:21:24million Learners as did corera and
- 00:21:26Udacity and other platforms
- 00:21:29what was beautiful is roughly free 3,000
- 00:21:32courses the challenge was completion
- 00:21:36rates less than 5% Why by the way if
- 00:21:39you're used to a boring lecture in the
- 00:21:40classroom the boring lecture online is
- 00:21:4310 times worse so there's virtually no
- 00:21:45engagement people take a long time to
- 00:21:47complete or may not complete but here's
- 00:21:49what's
- 00:21:50interesting the vast majority 75% of
- 00:21:53those who actually completed these
- 00:21:55courses already had college degrees
- 00:21:59meaning the educated rich were getting
- 00:22:01richer now think about that that's very
- 00:22:04sobering why is that because those are
- 00:22:07people used to curiosity intrinsic
- 00:22:10motivation by the way they're used to
- 00:22:11boring lectures they've gone to college
- 00:22:13but this has big implications for how we
- 00:22:15think about the digital divide so I just
- 00:22:17want to keep that in your mind and the
- 00:22:19last thing I just want to say is rather
- 00:22:22than going out and trying to create
- 00:22:23tutor Bots for as many courses as
- 00:22:25possible I think what we really need to
- 00:22:27do is have a strategic conversation
- 00:22:29about what's the role and purpose of
- 00:22:32teachers given the way the technology is
- 00:22:35proceeding the one thing I will say here
- 00:22:38is that when we think about what we
- 00:22:40learned in school okay think back think
- 00:22:43back many many
- 00:22:44years we learned many
- 00:22:47things tell me honestly how many of you
- 00:22:51have used geometry proofs since you
- 00:22:53graduated from high
- 00:22:56school three people
- 00:22:59why did we learn state capitals and
- 00:23:01world capitals of every single
- 00:23:05country okay uh foreign languages and by
- 00:23:09the way this is Italian Davy is not a
- 00:23:11goddess Davy in Italian says You must
- 00:23:15okay they have
- 00:23:17similarities um why did we learn foreign
- 00:23:19languages when we think about business
- 00:23:21Concepts in our curriculum I often get
- 00:23:23my students who come back 10 years later
- 00:23:25and say those two years were the most
- 00:23:27transformative years of my life life I
- 00:23:29often asked them what were the three
- 00:23:31most important Concepts you learned they
- 00:23:33said we have no idea I'm like no no okay
- 00:23:35give me one no no we have no idea I'm
- 00:23:38like so why do you say this was
- 00:23:39transformative the point simply being
- 00:23:41they're saying this was transformative
- 00:23:44not because of the particular content
- 00:23:46but because of the way we were learning
- 00:23:48we were forced to make decisions in real
- 00:23:49time we were listening to others we were
- 00:23:52communicating what are they saying
- 00:23:54they're saying that the real purpose of
- 00:23:56case method was listening and
- 00:23:58communication the real purpose of proofs
- 00:24:01was understanding Logic the real purpose
- 00:24:04of memorizing state capitals was
- 00:24:07refining your memory by the way that
- 00:24:08example there is the poem If by rard
- 00:24:10Kipling some of you might remember this
- 00:24:12from school it goes something like this
- 00:24:15if you can keep your head when all about
- 00:24:16you are losing theirs and blaming it on
- 00:24:17you uh I have PTSD because my nephew
- 00:24:20when he was reciting this to me
- 00:24:21preparing for his 10th grade exams I was
- 00:24:23like what the heck are you doing but it
- 00:24:25was basically refining memory skills and
- 00:24:28for for languages it was just learning
- 00:24:29cultures and syntax when we go deep down
- 00:24:32and think about what we were actually
- 00:24:35teaching I think that probably gives us
- 00:24:37a little more hope because it means it
- 00:24:40doesn't matter if some of these things
- 00:24:41are probably accessible through
- 00:24:43gen when calculators came along we
- 00:24:46thought it's going to destroy math
- 00:24:47skills we're still teaching math
- 00:24:49thankfully 50 years later and it's
- 00:24:51pretty good so this is something that I
- 00:24:53think is going to be an important
- 00:24:54strategic
- 00:24:55conversation this is the slide I'd love
- 00:24:57for you to keep in mind which is
- 00:24:59basically everything I've just said if
- 00:25:01you want to take a screenshot this is
- 00:25:02the slide to take a screenshot thank you
- 00:25:04all so much um and I hope to be in
- 00:25:09touch and keep this
- 00:25:12here at HBS I took Professor anan's
- 00:25:15class on economics for managers
- 00:25:16listening to him feels like being back
- 00:25:18in class fortunately he didn't call call
- 00:25:19anyone which is terrific so thank you
- 00:25:21for that now I have a few questions
- 00:25:24we've got young children and you've got
- 00:25:27so much of knowledge available now on
- 00:25:30chat prompts what's your advice to
- 00:25:33everyone who's got young children and
- 00:25:34are wondering about what should they be
- 00:25:36teaching their children so that when
- 00:25:38they grow up and when we don't know what
- 00:25:40the actual capabilities of these
- 00:25:41machines are that what they've learned
- 00:25:43is still
- 00:25:44useful how old are your kids rul so my
- 00:25:47son is nine and my daughter is five what
- 00:25:49are you telling them right now now I
- 00:25:51want to learn from you and I know we
- 00:25:53telling them a lot of stuff with a good
- 00:25:54bad ugly I don't I'm trying to refine
- 00:25:56that and give them a framework of what
- 00:25:57we should be telling them so there's two
- 00:25:59things so I think first of all this is
- 00:26:01probably one of the most common
- 00:26:02questions I get uh by the way it's
- 00:26:05really interesting that the tech experts
- 00:26:08and there was an article in the Wall
- 00:26:09Street Journal about this 10 days ago
- 00:26:12are basically telling their kids don't
- 00:26:13learn computer
- 00:26:16science that skill at least basic
- 00:26:18computer programming is
- 00:26:20gone Advanced Computer Science Advanced
- 00:26:23Data analysis if you want to do that
- 00:26:25that's going to be fine what are they
- 00:26:27telling their kids to learn they're
- 00:26:28telling their kids to learn how to teach
- 00:26:30dance they're telling their kids to
- 00:26:32learn how to do plumbing they're telling
- 00:26:35their kids to learn about the
- 00:26:37humanities why are they saying that
- 00:26:40implicitly they're saying what are those
- 00:26:42skill sets that are robust to machine
- 00:26:46intelligence now I will say it is
- 00:26:49virtually impossible to predict that
- 00:26:50given the pace at which this Improvement
- 00:26:52is
- 00:26:53occurring I probably have a slightly
- 00:26:55different kind of answer by the way my
- 00:26:57daughter's majoring in Psychology
- 00:26:59without me telling her anything so the
- 00:27:01kids I think know basically where this
- 00:27:02is going but the one thing I'll say
- 00:27:04Rahul is I don't know when you started
- 00:27:06out College what were you measuring in
- 00:27:08journalism journalism you started out
- 00:27:10with journalism okay that's enlightened
- 00:27:13I started out doing
- 00:27:15chemistry and then the reason I switched
- 00:27:17to economics was probably like many of
- 00:27:20you there was one teacher who inspired
- 00:27:23me and that's what made me
- 00:27:25switch and I would say to kids follow
- 00:27:29the teachers who inspire you and the
- 00:27:32reason is if you can get inspired and
- 00:27:34passionate about a subject that's going
- 00:27:36to build something that's going to be a
- 00:27:38skill that would last all your life
- 00:27:40which is
- 00:27:41curiosity which is intrinsic motivation
- 00:27:44we talked about in the last session this
- 00:27:45is no longer about learning episodically
- 00:27:49it's about learning lifelong and that's
- 00:27:51I think going to be the most important
- 00:27:52St in the way that Indian families
- 00:27:54operate and as do so many Asian families
- 00:27:56too parents want to equip their children
- 00:27:58with the skills that are likely to be
- 00:28:00most useful when they grow up so it used
- 00:28:03to be say engineering and doctors back
- 00:28:06in the day then uh it a few years ago so
- 00:28:10if you were looking ahead what do you
- 00:28:12think the children should be learning so
- 00:28:14they acquire skills which are useful in
- 00:28:16the job market years down yeah I think
- 00:28:18that's honestly being too
- 00:28:20instrumental as I said 10 years ago a
- 00:28:22lot of my students were talking to me
- 00:28:24and saying what should I major in I
- 00:28:25never told them computer science if I
- 00:28:27told them that that I would have
- 00:28:28regretted it but I genuinely mean this
- 00:28:31that's looking at it things too narrowly
- 00:28:33what I would say is think about things
- 00:28:35like creativity judgment human emotion
- 00:28:38empathy psychology those are things that
- 00:28:41are going to be fundamentally important
- 00:28:43regardless of where computers are going
- 00:28:45by the way you can get those skills
- 00:28:47through various subjects it doesn't
- 00:28:49matter it's not a one-o-one mapping
- 00:28:51between those skills and a particular
- 00:28:53topic or disciplinary ERA this is partly
- 00:28:55what I'm saying really think about where
- 00:28:57their passion is how do we teach our
- 00:28:58children how to think because
- 00:29:00everything's available on Google
- 00:29:02co-pilot chat GPT you can just chat GPT
- 00:29:05it so joining the dots giving them a
- 00:29:08framework to be able to interpret
- 00:29:10analyze and think how do you tell them
- 00:29:12that when the easiest thing is go yeah
- 00:29:16so uh it's a good question just two
- 00:29:18things on that the first is there was an
- 00:29:22interesting study done by colleagues at
- 00:29:23MIT recently where they had groups of
- 00:29:26students and they were asked to
- 00:29:29undertake a particular task or learn
- 00:29:31about a topic some students were given
- 00:29:34AI chat Bots some students were only
- 00:29:36given Google search with no
- 00:29:39AI what they found is the students with
- 00:29:42access to AI
- 00:29:44intelligence learned the material much
- 00:29:47faster but when it came time to apply it
- 00:29:50on a separate test which was different
- 00:29:52from the first one they found it much
- 00:29:55harder the students who learned the
- 00:29:57material through Google search with no
- 00:29:59other access took
- 00:30:01longer but they did much better on those
- 00:30:04tests why is that part of the issue is
- 00:30:07learning is not simple it takes effort
- 00:30:11okay and so part of the issue is you
- 00:30:13can't compress that
- 00:30:15effort um the harder it is to learn
- 00:30:19something the more likely you'll
- 00:30:21remember it for longer periods of time
- 00:30:25and so I think for me the big
- 00:30:26implication is when I tell my students
- 00:30:28look all these Technologies a are
- 00:30:30available it depends on how you use it
- 00:30:34my basic approach to them is just saying
- 00:30:38study because if you get domain
- 00:30:40expertise you will be able to use these
- 00:30:42tools in a much more powerful way later
- 00:30:45on uh so in some sense this goes back to
- 00:30:48the notion of agency it's like we can be
- 00:30:51lazy with tools and Technologies or we
- 00:30:53can be smart it's all entirely up to you
- 00:30:57but this is my advice you know some of
- 00:30:59my friends in Silicon Valley have the
- 00:31:02toughest controls on their children when
- 00:31:03it comes to devices you know we look at
- 00:31:06how much time our children can spend on
- 00:31:07their iPads or TV we're far more lenient
- 00:31:10and they're the guys who are actually in
- 00:31:11the middle of the devices and they're
- 00:31:13developing them and they know the
- 00:31:15dangerous side effects now those devices
- 00:31:17are also the repository of knowledge
- 00:31:18which is where you can learn so much
- 00:31:20from so as an educ every parent has his
- 00:31:22own take and how much time children can
- 00:31:24spend but as an educator how do you look
- 00:31:26at this device addiction just spending
- 00:31:28far more time picking up some knowledge
- 00:31:30but also wasting a lot of time yeah I
- 00:31:31think I mean there's a Nuance here which
- 00:31:34is basically what they're doing is not
- 00:31:36saying don't use devices they're saying
- 00:31:39don't use social media and this goes
- 00:31:41back again to one of the things we were
- 00:31:43talking about
- 00:31:44earlier uh we have gone through a decade
- 00:31:48where things like misinformation
- 00:31:50disinformation and so on there is no
- 00:31:52good solution as far as we know
- 00:31:54today there's also various other kinds
- 00:31:56of habits and so on that are going to
- 00:31:57improved that's partly what they're
- 00:31:59saying stay away from they're not saying
- 00:32:00stay away from computers we can't do
- 00:32:02that and in fact you don't want to do
- 00:32:04that but there's a Nuance in terms of
- 00:32:05how we interact with with tools and
- 00:32:07computers that we just want to keep in
- 00:32:09mind when we think about guardrails
- 00:32:11right are you seeing your students
- 00:32:13getting more and more obsessed with
- 00:32:14their devices and how does that impact
- 00:32:16what are you trying to do to get them to
- 00:32:19socialize more you know to spend more
- 00:32:21time with each other and not be stuck on
- 00:32:23their phones or that yeah it's a very
- 00:32:24interesting question so in some sense
- 00:32:26last year we had a conference at Harvard
- 00:32:28we had 400 people from our community
- 00:32:30attend the conference and some of our
- 00:32:33colleagues were saying we should have a
- 00:32:35policy of laptops down no laptops and
- 00:32:37class take out
- 00:32:39devices I was coming in for a session
- 00:32:41right afterwards but part of the reason
- 00:32:44I wanted them to take out their mobile
- 00:32:45phones was I had two or three polls
- 00:32:48during my lecture where I wanted them to
- 00:32:50give me their input so I said mobile
- 00:32:52phone's out okay and this was sort of
- 00:32:55crazy but the story illustrates
- 00:32:57something interesting which is these
- 00:32:59devices for certain things can be really
- 00:33:01powerful it can turn a passive learning
- 00:33:03modality into an active learning
- 00:33:05modality where every single person is
- 00:33:07participating we don't want to take that
- 00:33:09away what we want to try and deal with
- 00:33:11is people playing games while you're
- 00:33:14lecturing now by the way me personally I
- 00:33:16just put it on
- 00:33:17myself if I'm not exciting enough or
- 00:33:19energizing enough for my students to be
- 00:33:21engaged use your Mobile phones that's on
- 00:33:23me okay but that's partly what chall Eng
- 00:33:28show fans how many felt engaged during
- 00:33:30the session and how
- 00:33:33many okay no so uh that that that which
- 00:33:36is why Agent take Ai and chat Bots can
- 00:33:40never do what professors can right so uh
- 00:33:42we I'll take some questions KH has a
- 00:33:44question KH go
- 00:33:45ahead hi professor uh you mentioned that
- 00:33:48one of the things that we should work on
- 00:33:50to teach our children is empathy how do
- 00:33:54you actually teach empathy in know
- 00:33:56formal educ Ed ation system or does this
- 00:33:59just go back to then parents and
- 00:34:03family it's a it's a hard it's a hard
- 00:34:06question um in fact this is by the way
- 00:34:09one of the most important issues we're
- 00:34:11facing today on campuses it's related in
- 00:34:15part even in higher education not just
- 00:34:17younger kids when we talk about
- 00:34:19difficult conversations on
- 00:34:21campus part of the reason we're facing
- 00:34:24those issues is because people are
- 00:34:28intransigent it's like I don't care what
- 00:34:30you say I'm not going to change my mind
- 00:34:33one of the things we introduced a couple
- 00:34:35years ago on the Harvard application for
- 00:34:37undergraduate is a question that says
- 00:34:40have you ever changed your mind when
- 00:34:42discussing something with anyone else
- 00:34:44okay or something to that effect but
- 00:34:46that's basically saying how open-minded
- 00:34:48are we that's one version of empathy
- 00:34:50there's many other
- 00:34:51dimensions I think part of the challenge
- 00:34:53is that we don't teach that in schools
- 00:34:58right we don't teach that formally in
- 00:34:59schools which is partly why there's this
- 00:35:01whole wave now of schools not just in
- 00:35:04other countries in India which has
- 00:35:06stared to talk about how do we teach the
- 00:35:08second curriculum the hidden curriculum
- 00:35:10how do we teach those social and
- 00:35:11emotional skills The Book of Life so to
- 00:35:14speak and I think I mean it's not rocket
- 00:35:18science to say this it starts at home
- 00:35:20right like that's basically what we do
- 00:35:22with our kids every single day um but
- 00:35:25that's something that's I think going to
- 00:35:26become fundamentally more important
- 00:35:28partly because of the reasons are what
- 00:35:30what I talked about Dr sanjie B has a
- 00:35:32question okay I see lots of hands going
- 00:35:37up yes Dr wonderful wonderful listening
- 00:35:40to you um just with regards Ai and
- 00:35:43Technology I've always said that uh Ai
- 00:35:45and digital technology is not an
- 00:35:48expenditure It's actually an investment
- 00:35:51so very quickly if you allow me just 60
- 00:35:53seconds in healthcare it gives you
- 00:35:55better clinical outcomes it has has
- 00:35:57decreased from number one cause of death
- 00:36:00as Hospital acquired infections in many
- 00:36:03Hospital chains as practically less than
- 00:36:051% so it gives you a safer outcome it
- 00:36:08gives you a better patient experience
- 00:36:10the turnaround of the bed strength is a
- 00:36:12lot quicker and more importantly is it
- 00:36:14gives you better operational excellence
- 00:36:17so all the hospitals as far as medical
- 00:36:19facilities are concerned who have not
- 00:36:20embraced it as yet will find it
- 00:36:23difficult to operate in the present
- 00:36:24environment the what Ai and digital
- 00:36:27technology has made us learn as doctors
- 00:36:30is that data is the new gold if you
- 00:36:32don't analyze data if you don't see what
- 00:36:34your results are if you don't see where
- 00:36:36your clinical outcomes are then you
- 00:36:37can't go forward so AI is what is in the
- 00:36:40future for us all of us thank you that's
- 00:36:42more in the form of an let me just
- 00:36:44elaborate on that in two ways one is I
- 00:36:47think I would just go back and useful to
- 00:36:50contextualize AI right like right now we
- 00:36:53we often get obsessed by the latest
- 00:36:55technology when we think about
- 00:36:56upskilling reskilling in education
- 00:36:59there's a revolution that started a
- 00:37:01decade ago as I alluded to there's
- 00:37:04basically 3,000 courses available to all
- 00:37:06of you today on any subject so the
- 00:37:09notion of let's wait for AI no no no
- 00:37:11it's already there my father-in-law
- 00:37:13who's 92 years old during covid he said
- 00:37:15bat what should I do I said we have all
- 00:37:17these these courses from Harvard
- 00:37:19available in the last two years or three
- 00:37:21years he's completed 35 courses wow okay
- 00:37:26at the age of 92 wow wow by the way he's
- 00:37:29paid $0 for that because he said I don't
- 00:37:31need a certificate and so I told him
- 00:37:33you're the reason we have a business
- 00:37:34model problem okay but that's one
- 00:37:37aspect the the second aspect is sort of
- 00:37:40thinking about where you're going I
- 00:37:42think you're exactly right sanj which is
- 00:37:44every organization is going to have lwh
- 00:37:46hanging fruit the one thing I just
- 00:37:48caution is there's going to be a paradox
- 00:37:50of access meaning if every organization
- 00:37:54every one of your peers has access to
- 00:37:56the same technology as you
- 00:37:58it's going to be harder for you to
- 00:37:59maintain competitive advantage that's a
- 00:38:02fundamental question okay this is just a
- 00:38:04basic observation so I just want to sort
- 00:38:07of mention that but you're absolutely
- 00:38:09right about the lwh hanging fruit in
- 00:38:10medicine and Healthcare okay Toby Walsh
- 00:38:12has a question or an observation and
- 00:38:14then we I there lots of hands up okay I
- 00:38:16don't frankly know what to do because
- 00:38:18we're also out of time so let this just
- 00:38:19be where we conclude one of the greater
- 00:38:21challenges especially in higher
- 00:38:22education is the cost has gone through
- 00:38:24the roof are you optimistic that AI is
- 00:38:27going to be able to turn that around so
- 00:38:31again I'll just go back uh to what's
- 00:38:34happened in the last decade as I said
- 00:38:37you can now get access to credentials
- 00:38:39and certificates at a minimal cost
- 00:38:42compared to the cost of getting a degree
- 00:38:44okay just to put it in perspective we
- 00:38:46have 177,000 degree students every year
- 00:38:49who come to Harvard they are paying a
- 00:38:51lot of money those who need financial
- 00:38:53aid get financial aid by the way can
- 00:38:55anyone guess how many students we have
- 00:38:57touched over the last
- 00:39:00decade 10 times 100 times that it's
- 00:39:03about 15
- 00:39:04million that is not a story We publicize
- 00:39:07but that's a story about the number of
- 00:39:09students who've actually taken a Harvard
- 00:39:11course or enrolled in a Harvard course
- 00:39:13so in some sense I think where we are
- 00:39:14today is the marginal cost of providing
- 00:39:17education is very very low what we need
- 00:39:20for that is not incremental Improvement
- 00:39:23on the existing model we need to
- 00:39:25basically break it apart and say how do
- 00:39:28we put it back together again in a way
- 00:39:30that makes sense for everyone um there's
- 00:39:33an organization that we just started at
- 00:39:35Harvard called axim jointly with MIT
- 00:39:37with the endowment from the sale of the
- 00:39:38ax platform whose only function is to
- 00:39:41increase access and equity in education
- 00:39:43and by the way their focus is on 40
- 00:39:45million people in America Who start
- 00:39:47college but never completed not just
- 00:39:49because of cost for many other reasons
- 00:39:51right in some the potential to reduce
- 00:39:54the cost is massive but it's going to
- 00:39:57require leadership and strategy this
- 00:39:59gentleman here has a question can
- 00:40:01someone just take the mic to him
- 00:40:08please uh so uh earlier it was okay use
- 00:40:12Ai and it will summarize and help you in
- 00:40:14productivity but with the latest open AI
- 00:40:16models like o03 minia and all that they
- 00:40:19are doing reasoning which is much better
- 00:40:21than humans so the people who are not
- 00:40:24using it are at a dis disadvantage
- 00:40:28so isn't it right that the students use
- 00:40:30Ai and uh be familiar with it and uh be
- 00:40:35be up to speed with that rather than not
- 00:40:37using it and be at a disadvantage to
- 00:40:39other students yeah absolutely there's
- 00:40:41no question about that by the way I sit
- 00:40:44at Harvard overseeing the generative AI
- 00:40:46task force for teaching and learning and
- 00:40:48we have 17 faculty the most interesting
- 00:40:50conversations I've had about adoption
- 00:40:53are with our
- 00:40:54students now there's when we understand
- 00:40:57that Behavior it just throws up things
- 00:40:59that we wouldn't even have thought about
- 00:41:01I'll ask you one question we had a
- 00:41:03Sandbox that we created for the entire
- 00:41:04Harvard Community which was a safe and
- 00:41:06secure sandbox giving them access to
- 00:41:08large language models as opposed to
- 00:41:10using public open AI the adoption rate
- 00:41:12amongst our faculty was about 30 35% in
- 00:41:15the first year what do you think the
- 00:41:17adoption rate was amongst our
- 00:41:21students it was about
- 00:41:235% so we were surprised when we went to
- 00:41:26them we said what's going on are you
- 00:41:28familiar with the sandbox they said yeah
- 00:41:30we are we said are you using it they
- 00:41:32said no we said are you using AI in any
- 00:41:34way yeah yeah we have access to chat GPT
- 00:41:36we have our own private accounts there
- 00:41:38so we're like wait a wait why are you
- 00:41:40not using the secure Harvard
- 00:41:42sandbox what do you think their answer
- 00:41:46was they said why would we use something
- 00:41:49where you can see what we're
- 00:41:51inputting now by the way as faculty
- 00:41:54members if the number one question we
- 00:41:56talk about with generative of AI is oh
- 00:41:58we're worried about cheating and
- 00:41:59assessments our students are listening
- 00:42:01to us they're like oh if that's what
- 00:42:03you're worried about we're not coming
- 00:42:04anywhere close to you okay so part of
- 00:42:06the point is the students are far ahead
- 00:42:08of us in terms of using this they're
- 00:42:10using it to save time they're using it
- 00:42:11for engaging in deep learning we better
- 00:42:14understand that ourselves to figure out
- 00:42:15what we can do
- 00:42:17join brilliant presentation just wanted
- 00:42:19to understand one side of the spectrum
- 00:42:21you have all the you know the positives
- 00:42:24what's on the other side what risk do
- 00:42:26you think is there on the other side it
- 00:42:28starts coding on its own gets out of
- 00:42:30hand is that a possibility what's the
- 00:42:32possibility so so the risks are the
- 00:42:33things I talked about towards the end
- 00:42:35okay which is number one we put our head
- 00:42:38in the sand as institutions and we don't
- 00:42:41take this seriously that's the first
- 00:42:43risk the second risk is lazy learning
- 00:42:46the way I would call it now again that's
- 00:42:48agency it partly depends on you as a
- 00:42:51student do I want to be lazy do I not
- 00:42:53want to be lazy the third risk is
- 00:42:56everything we were talking about in the
- 00:42:57previous session with respect to
- 00:42:58misinformation
- 00:42:59disinformation the fourth big risk is
- 00:43:02asking the fundamental question what's
- 00:43:04our role as teachers and I'll just share
- 00:43:06one anecdote in closing there's a
- 00:43:08colleague at another school who called
- 00:43:09me and said my students have stopped
- 00:43:12reading the cases they're basically
- 00:43:15inputting the assignment questions into
- 00:43:16generative Ai and by the way they're so
- 00:43:18smart they're saying give me a quirky
- 00:43:20answer I can use in class okay the
- 00:43:23assessments are compromised and get this
- 00:43:25The Faculty have stopped reading cases
- 00:43:27they're inputting the cases and
- 00:43:29basically saying give me the teaching
- 00:43:32plan that's the
- 00:43:34downside you know we we met on a flight
- 00:43:37from Delhi to Mumbai and we had a long
- 00:43:39conversation about the future of
- 00:43:40Education we've been able to in the past
- 00:43:4245 minutes recreate the magic of that
- 00:43:44conversation here on stage can we have a
- 00:43:46very warm Round of Applause for the
- 00:43:47professor for making the effort of
- 00:43:49coming here and for joining us and for
- 00:43:52delivering this master class thank you
- 00:43:54absolute pleasure thank you so much
- 00:43:56thank you
- 00:43:59if you like the video do like comment
- 00:44:01share and
- 00:44:05[Music]
- 00:44:09subscribe if you like the video do like
- 00:44:12comment share and subscribe
- Generative AI
- Education
- Harvard
- Technology
- Learning
- ChatGPT
- Future of Education
- Critical Thinking
- Empathy
- Emotional Intelligence