Latest / Trojan Talks / Ep. 8: President Beong-Soo Kim with Prem Natarajan
Transcript
- 0:02Welcome, Prem.
- 0:04You really don't need much introduction,
- 0:05but let me just share a little bit
- 0:07about your background and demonstrate
- 0:10why you're really the perfect interlocutor
- 0:13to have a conversation about university-private industry engagement.
- 0:17So, I believe you started
- 0:19your career at BBN,
- 0:21now Raytheon, and became the executive vice
- 0:25president of research around language.
- 0:30And
- 0:32well, it was language
- 0:34and speech,
- 0:36language, and computer vision.
- 0:38And then you move from there to USC,
- 0:41where you were the executive director
- 0:43of the Information Sciences Institute
- 0:46and the senior vice dean
- 0:48for research at Viterbi.
- 0:50We have the Viterbi dean here in the audience, of course.
- 0:53Dean Yannis Yortsos.
- 0:56Then we, at USC,
- 0:59you began to work at Amazon
- 1:02working on Alexa AI.
- 1:05And a few years ago,
- 1:07you had an amazing opportunity to join Capital One
- 1:10as their executive vice president over enterprise
- 1:13AI and as their chief scientist.
- 1:15So you have really gone back
- 1:18and forth between university,
- 1:20USC in particular, and private industry,
- 1:22and I was wondering if you could just start by sharing
- 1:27your observations as to
- 1:29how the relationship between universities
- 1:31like USC and private industry
- 1:33have evolved over the years.
- 1:37Yeah.
- 1:37I'd say maybe we'll start somewhere in between,
- 1:41maybe in the late ‘90s.
- 1:43I mean, there's a long history here, especially with USC,
- 1:47with the aerospace industry and all of that,
- 1:48but that's maybe in the ’60s, ‘70s, and so on.
- 1:51But I think
- 1:54there are two or three factors
- 1:55that have brought industry,
- 1:58some elements of industry and academia
- 2:01together over time, and that has broadened.
- 2:04And one of those factors
- 2:06actually is DARPA, in my mind,
- 2:08because DARPA’s model,
- 2:13initially
- 2:14they just funded principal investigators.
- 2:16They gave them several hundred thousand dollars,
- 2:18which in the ‘70s was a lot of money,
- 2:20but they didn't really care
- 2:22about organizational affiliation.
- 2:24And as DARPA’s,
- 2:27like, track
- 2:27record grew and they were making real technological advances
- 2:30that could be translated into real-world
- 2:33things that could make a difference, they started realizing
- 2:36that they needed industrial-scale entities
- 2:39engaged in the DARPA mission.
- 2:42Space itself was the start of it, etc., but,
- 2:44but in many of the other areas.
- 2:47And so we started seeing
- 2:50these large companies come about.
- 2:52But those companies still needed
- 2:55the innovation engines
- 2:57to drive the core part of DARPA’s mission.
- 2:59And so they were heavily
- 3:03reliant on academia for it for the longest time.
- 3:05So I think that's been one trajectory
- 3:08for like seeding
- 3:10in a very systematic and sustained way.
- 3:13And that then started
- 3:15expanding into other areas.
- 3:17DARPA also because, in a sense,
- 3:20DARPA is the largest venture capitalist in the world.
- 3:23They get, like, $4 billion every year.
- 3:25They have to invest it in new ideas, and it gets replenished.
- 3:28They don't even have to go
- 3:30raise money from investors.
- 3:32The society gives it to them.
- 3:35So it's actually a massive venture capitalist thing.
- 3:38It's also created its own ecosystem
- 3:40of companies that kind of
- 3:42were born in the intersection
- 3:45of what was classical industry and academia, so
- 3:48a lot of jobs came out of it, etc., etc.
- 3:51And so that's been one, one trend.
- 3:54The, the other is,
- 3:56I think, we saw with the startup ecosystem,
- 4:00a lot of faculty members and especially graduate students
- 4:03starting to do startups.
- 4:05And so
- 4:07the industries
- 4:08of the future, sometime I would say
- 4:10starting late ‘80s,
- 4:12but really starting in the mid
- 4:14to late ’90s,
- 4:16most of today's companies
- 4:18were literally born on university campuses.
- 4:21So they're very rooted in academia.
- 4:24So when we look back at these 30 years,
- 4:26there are at least those two very strong forces
- 4:29that have created this connection.
- 4:31And then as some of them started going
- 4:34up, people started finding systematic values.
- 4:36There are now venture capital firms that specialize
- 4:39in finding faculty members with good advances and engaging with them
- 4:42and bringing ...
- 4:43like that's their whole business model.
- 4:44In fact, now most top universities
- 4:46have a venture capital fund
- 4:48that will call themselves the MIT Alumni Fund
- 4:51or this and that, and, like, they're doing it.
- 4:53So that has happened.
- 4:55With the advent of AI,
- 4:57we saw, like I'll say,
- 4:59these are the first two exponentials.
- 5:01I have to use the word exponential because Yannis loves it.
- 5:03Right?
- 5:04So, you know, he's
- 5:06the one who first taught me to hug the exponential, you know?
- 5:09And if you want to know what that is, please ask him.
- 5:11Right?
- 5:11And so, but the third exponential
- 5:15in my mind was unleashed by AI.
- 5:19What happened was AI kept developing,
- 5:23kind of incubating in academic centers
- 5:26for a very long time, and in dribs and drabs.
- 5:31It made its way to industry.
- 5:32And there was a lot of difference
- 5:34it made in the form of machine learning models, automating this,
- 5:37automating that,
- 5:38or scaling up this decision, making that decision, making ...
- 5:42But with the advent of deep learning,
- 5:45so 2010, 2011,
- 5:47thereabouts, people started seeing
- 5:49real potential for transformation.
- 5:51We're now on the other side of it.
- 5:52We are actually seeing how it's manifesting itself.
- 5:55But back
- 5:55then people didn't know how it was going to happen,
- 5:57but they started seeing the potential.
- 5:59So if you trace back to it, you will see
- 6:01a lot of top academicians
- 6:04actually went into industry.
- 6:07In fact,
- 6:08the exodus was so sudden and so large
- 6:12that academia responded by accommodating
- 6:15rather than lose this many,
- 6:17this many people at once.
- 6:19And so those people got to keep their academic
- 6:22affiliations, whatever keeping meant.
- 6:24But, and then they were industry.
- 6:26And so you still have that large cohort in there.
- 6:30I mean, even now if you look around, you know, people who have spent
- 6:33good chunks of time at Amazon or Google, etc.,
- 6:36just looking at those two gentlemen in the back and, and so ...
- 6:39but that has started.
- 6:40So I think all of this connection ...
- 6:43what that unleashed,
- 6:44this exponential, what made it an exponential
- 6:47is that companies that couldn't
- 6:49get these top academics to come to them,
- 6:51like if you're a Google or Microsoft or even Amazon,
- 6:55it's easy enough to attract academics to come to you.
- 6:58And Dropbox did.
- 6:59They got Sergio and others from, like, UC Irvine.
- 7:04Is that
- 7:05a whole host of other companies that are in technology said,
- 7:09“I don't know if I can have credibility
- 7:11if I'm not able to attract more academics.”
- 7:14And so it started becoming a trophy game.
- 7:17And in pure capitalist terms,
- 7:19when you have so many companies, so much money,
- 7:21they started going after, it's
- 7:23not academia that I care about,
- 7:25it's the academics and the credibility that they have
- 7:28that I care about.
- 7:28So that unleashed a thing.
- 7:31In between, there,
- 7:32there are some some of us who said, “Gee,
- 7:35the seed crop, we're eating the seed crop.
- 7:39We still have a lot of hunger for the,
- 7:41the graduate students and everyone they're producing.”
- 7:44So how do we make this
- 7:47kind of a flywheel
- 7:48rather than just an expeller,
- 7:50right, that somebody had fed.
- 7:52And so there's that one track of it that's coming.
- 7:55Then there's a bunch of other companies that just want the affiliation
- 7:59with the universities for various purposes, like,
- 8:02you know, whether it's their reputation or things like that.
- 8:05And so I think this is kind of the fertile landscape
- 8:09right now. Thing I notice is
- 8:12I think there's tremendous latent hunger
- 8:15on the corporate side for partnership.
- 8:19There are
- 8:19the quarterly earnings cycles
- 8:22that people fall victim to and then pay attention to it.
- 8:25And they forget that hunger that they have.
- 8:27And on the university side,
- 8:29I see a lack of understanding
- 8:32of how to structure your go-to-market,
- 8:36to kind of find the pockets of those hunger
- 8:39and then start to grow them.
- 8:41But I think like any efficient
- 8:44capital market, this one will clear itself in time.
- 8:46People will figure out on both sides how to make it happen.
- 8:50Let me ask a question.
- 8:51So traditionally, there have been concerns within academia
- 8:54about getting too close to industry,
- 8:56whether due to conflict reasons or, or other reasons.
- 8:59And I was wondering if you could talk
- 9:01a little bit about that, just to weave it
- 9:04into the historical context,
- 9:06and would love to get your thoughts.
- 9:09Is that, is that still a concern, or, or
- 9:12how do people think about it now
- 9:14at this present stage?
- 9:16I think it's still a concern, especially if you look into areas
- 9:20like medicine, etc., where
- 9:24where the interpretation of results
- 9:27is often subjective,
- 9:28like even though you might have lots of quantitative
- 9:31data, how,
- 9:33what to make of it and how to judge it
- 9:35and how to judge its impact on the people that ...
- 9:38from whom the data has been collected is still a subjective art, right?
- 9:41The data doesn't answer the question by itself.
- 9:44And so people have to express an opinion.
- 9:46This is a safe treatment,
- 9:48or this is a safe protocol, or this is ready for market.
- 9:52And of course there are lots of standards, etc.,
- 9:54but there is still a fair bit of subjectivity.
- 9:57In those areas, I do think there is issue,
- 10:01or ongoing concerns,
- 10:03but similarly in the social sciences, what is the impact of technology?
- 10:08What should policy be
- 10:10to regulate these technologies?
- 10:11Those areas, of course, there's an inherent,
- 10:13inherent conflict.
- 10:15But in some sense, we are also ...
- 10:18I'll just say this.
- 10:18I feel we are in a kind of ...
- 10:26We have a
- 10:28different sense of how to deal
- 10:29with ethics now in, in society.
- 10:31But I’ll just say, conflict ...
- 10:33those areas of conflict.
- 10:35I've always thought engineering is,
- 10:37like the pure sciences and engineering
- 10:41are kind of unique in the sense
- 10:43that the data speak for themselves.
- 10:45The results speak for themselves.
- 10:47They have to work in the real world.
- 10:49You can't say this car is going at 60 miles per hour
- 10:51unless you can measure it's going at 60 miles per hour.
- 10:55You can't say this car is self-driving itself
- 10:58if there is still a driver sitting behind
- 10:59the steering wheel and touching it every now and then.
- 11:02So the level of objectivity
- 11:04with which you can call
- 11:06outcomes is very high.
- 11:08And, and, and, and these outcomes are, like,
- 11:11objectively verifiable.
- 11:13And so I think there's a ...
- 11:15yes, you could foll-, you could consider ...
- 11:17And so if you look at it, I think people who are rooted
- 11:21in these other places where the, the conflict is inherent
- 11:25will approach it from the perspective
- 11:26that, “My God, if you're getting so much industry money
- 11:29in these schools, you must be compromised in some way.”
- 11:32But the, the, the nature of engineering
- 11:35and, and, and hard sciences is such
- 11:39that I don't think conflicts stand
- 11:41and you can see where it intersects with policy.
- 11:44It becomes hard, like climate, right?
- 11:46Moment, you can't measure.
- 11:48If the outcomes are not like immediate or within months,
- 11:51if the outcomes are over decades and they are out there,
- 11:54then there is the possibility to have controversy.
- 11:57And then ... yeah.
- 11:59You mentioned earlier, Prem,
- 12:00that there's a lot of latent desire for deeper,
- 12:04more long-term relationships
- 12:06between university and industry,
- 12:08and that we may be only in the early innings still,
- 12:11as to how this relationship can continue
- 12:15to evolve and be productive for society.
- 12:18What does that future potentially look like?
- 12:21And although you mentioned that market
- 12:24dynamics will, will get us there one way or another,
- 12:28what do you think the opportunities are
- 12:30for universities and university leadership
- 12:33to accelerate their own opportunities
- 12:37to take advantage of, of that?
- 12:39So, you know, I think the startup
- 12:43ecosystem provides a valuable insight here.
- 12:46If you look at many of these startups that are growing at rapid scale
- 12:49now or in the last 20 years,
- 12:51they’ll all start with a hypothesis.
- 12:53They'll find some product market fit with what they're doing, etc.,
- 12:56but they're actually not able to scale just with that.
- 12:59Once in a while, you do.
- 13:01You're a Dropbox, and the solution is just online,
- 13:03and it happens.
- 13:05But with a lot of things,
- 13:07especially with this new wave of AI, what you will see
- 13:10is that these startups will say,
- 13:12“Hey, you know my technology is so hard,
- 13:14and there is so much you need to know in how to match it
- 13:17to your business problem.
- 13:18I don't know your business problem.
- 13:20You know your business problem,
- 13:22but you don't know my technology either.
- 13:24And it's not expressible in a way
- 13:26that's easy for you to self-serve.
- 13:28So I'm going to put out these things that I'll call
- 13:30‘forward deployed engineers.’” By the way,
- 13:32you shouldn't use those terms because there was just an article
- 13:35in New York Times or Wall Street Journal just a couple of days ago,
- 13:38where the most glamorous job
- 13:41in AI is one nobody wants.
- 13:43And it's called a forward deployed engineer, right?
- 13:45And they used Palantir as an example.
- 13:48Not sure why, but they, they, they created this model in a way.
- 13:52But what happens is those folks
- 13:54go sit in that company for six months, eight months.
- 13:57They work in it.
- 13:58Through it, they understand the inside processes of the company,
- 14:01and then they start solving for it.
- 14:03So I think one of the ways
- 14:05in which this market can clear itself
- 14:08is if universities can create
- 14:11opportunities
- 14:13where the match between
- 14:16faculty and industry
- 14:19is not opportunistic and driven just by the company.
- 14:22But universities start programs where
- 14:25there could be rotation opportunities
- 14:27structured that allow faculty to go in.
- 14:30And oftentimes what I notice
- 14:32when I speak with some faculty is, you've been there for 20 years.
- 14:36You kind of are fatigued.
- 14:38It's also true that when you've been in industry for 20 years,
- 14:40you're also fatigued in a slightly different way.
- 14:42And so I do think there's like
- 14:44that opportunity bidirectionally,
- 14:47but really the, the,
- 14:51the challenges,
- 14:52the expertise in the universities
- 14:55is resident in individuals
- 14:57who are not naturally connected
- 15:00in an organi-.
- 15:01I mean, academia is not a team game, right?
- 15:05It's the Wild West.
- 15:06It's the individual faculty who is creating their own brand,
- 15:09and they're going out there and establishing their brand and, like,
- 15:12developing their own reputation.
- 15:14Companies are entirely team games.
- 15:16The company wins or you all lose.
- 15:19There is no way for an individual to win.
- 15:22So there is that impedance
- 15:24mismatch in a way.
- 15:27And so I'd say, like, that ...
- 15:31universities that can solve that impedance mismatch,
- 15:36and there are ...
- 15:37we can talk about it later, but I think solving that impedance
- 15:41mismatch will make universities
- 15:46kind of part of the deal flow.
- 15:48Right now, the deal flow is just
- 15:50between companies and faculty members for the most part,
- 15:53because universities are not putting themselves
- 15:55out there in a way to enable it.
- 15:57You mentioned the rotational opportunities for faculty, in particular.
- 16:00What about for students?
- 16:02Yeah, for students, I mean,
- 16:04this gets to be a slightly more
- 16:06or less controversial, whichever
- 16:08you think about it, because the notion
- 16:09of a cooperative program is not a new one.
- 16:12But I do think right now
- 16:14in an environment where,
- 16:17by the way, that recording
- 16:18is throwing me off my thing.
- 16:20I'm finding it distracting.
- 16:21So just, you know, so
- 16:24the for, for, for, for
- 16:28undergraduate students, especially in engineering,
- 16:31especially in computer science and AI, etc.,
- 16:34what's happening
- 16:35is the pace of advances is so fast.
- 16:37I don't know if you know this, like some of the startups right now
- 16:40have actually created their own programs
- 16:43that, like, resemble undergrad programs, right?
- 16:45Like, come join us as an intern, spend a year with us,
- 16:48and we will train you.
- 16:50Now, any ...
- 16:51there will be a bunch of people who take that on.
- 16:54OK.
- 16:54So that tells you that there is a marketplace for it,
- 16:58but there's massive numbers
- 17:00of students who want it, but who are anchored
- 17:03to the traditional type of education, who feel like,
- 17:06“If I go out there on a limb, I might, I lose out.”
- 17:09But universities that are able to blend those two opportunities for
- 17:12their students are going to make their students
- 17:15the winners 20 years from now.
- 17:18Right?
- 17:18And so I think,
- 17:20you know, the university
- 17:22that is going to win 20 years
- 17:24from now, which means their students have won 20 years from now, needs
- 17:27to think about what that destination looks like
- 17:29and work back from there.
- 17:31To me, it looks like
- 17:33just like on the research side, you know,
- 17:36last week ...
- 17:37no, earlier this week, we had the,
- 17:39the first annual Capital One
- 17:42AI symposium in McLean, Virginia.
- 17:45Gaurav was there.
- 17:46A couple of other faculty members were there and presented.
- 17:50But in that ...
- 17:51in, in writing about it online,
- 17:53I talked about how we kind of view this.
- 17:56The first day was structured as academic summit.
- 17:59The second day was structured as the frontier forum.
- 18:02So the first day was really academic.
- 18:05Academics and academic thought leaders talking about it.
- 18:07The second day was the frontier forum,
- 18:09where there was industry and startups
- 18:11in there and, and,
- 18:14but the way we viewed it and the way it worked out also
- 18:17that we did not see this as two sides of a coin, but we saw this ...
- 18:21We see this as a continuous loop of collaboration
- 18:24from how science goes into production and how production
- 18:28requirements inform the new science problems to be solved,
- 18:31and that's the conversation we're starting.
- 18:34That is moving at different speeds
- 18:36and different areas of engineering.
- 18:38But I think in most areas of engineering
- 18:40that has accelerated to a point
- 18:43where I think students
- 18:45spending four years
- 18:47at a university at one stretch,
- 18:50I'm no longer convinced,
- 18:53you know, even though
- 18:54my three daughters will do exactly that.
- 18:56And so you could ask me
- 18:58to have the courage of my conviction, you know,
- 19:01to actually tell them, don't do it.
- 19:03Or you could ask, do they trust me
- 19:05enough to listen to me if I tell them that?
- 19:07But that aside, I do wonder
- 19:10whether we, you know,
- 19:12people in the academic profession,
- 19:14are we say, you know,
- 19:16anything that hasn't changed for 60 or 70 years
- 19:19as a product, when the world has changed so much,
- 19:23it almost begs the question, really?
- 19:26I mean, our product market
- 19:28fit is like humans and oxygen?
- 19:30Like this is enduring and never going to go away.
- 19:34Well, it can't be because it's not biological.
- 19:36It's not organic.
- 19:36So the product market fit cannot be that enduring.
- 19:39And what can we do?
- 19:41And of course,
- 19:42there are other things that people learn in the university
- 19:45other than specific skills, right?
- 19:47A lot of socialization.
- 19:49The networking, yeah.
- 19:51Let me ask you this.
- 19:53So having had the benefit now
- 19:56of time in academia, time in the private sector.
- 20:00In fact, you're still a research professor at USC.
- 20:02I should have noted.
- 20:07We'll see what he decides after this panel.
- 20:09So, so,
- 20:12the question I had was,
- 20:14it sounds like there are a lot of opportunities for universities
- 20:17to lean into these future opportunities.
- 20:20Is industry ready for that?
- 20:22Are they, are they primed
- 20:24at this point to explore new models
- 20:27that are not just individual faculty based,
- 20:30but are longer term and more institutional?
- 20:34I think so.
- 20:35I mean, there's just so many examples.
- 20:37I’ll take USC, for example, right?
- 20:40I think, when I joined,
- 20:43the big talk used to be
- 20:44about the Chevron Center
- 20:46and how it had been established many years ago, etc.
- 20:49Then when I went to Amazon about like,
- 20:53I think less than a year and a half, created
- 20:56this program called the Amazon University
- 20:58Hubs Program.
- 21:00First one was Columbia.
- 21:01Second one was USC.
- 21:03Now it's got, like, a dozen or more.
- 21:05Right?
- 21:07The explicit design behind those
- 21:10was that we wanted to scale
- 21:12beyond faculty to the full talent
- 21:16that is resident at a university, which includes the students, etc.
- 21:19Faculty are still center.
- 21:21I mean, there is no university
- 21:23really without the faculty,
- 21:25but there's also no university without the students.
- 21:26And so we wanted to tap into both of those.
- 21:29And so we, we created that.
- 21:31As part of it, we also said, you know,
- 21:33but the university needs certain
- 21:35kinds of things to sustain itself.
- 21:37And so we wrote into that program
- 21:41resources that would be discretionary a little bit at the,
- 21:46you know,
- 21:46whether it's with the dean or the right sources.
- 21:49I, I see somewhat similar
- 21:53steps happening with,
- 21:56you know, other tech companies
- 21:58that have started
- 22:01these centers, etc.
- 22:03As I said, the latent hunger
- 22:05in a lot of companies is
- 22:07we are getting into a world
- 22:08which is so tech-driven, almost ...
- 22:11like, there are companies that are tech companies
- 22:15and there are companies that are going to be tech companies.
- 22:18I think that's the only dichotomy I see.
- 22:21Like, you know, almost every business has become a tech business.
- 22:24You either know you're in a tech business or you don't.
- 22:27And the ones that discover it become enduring companies.
- 22:30This is why I feel like Capital One is the first fintech in a way.
- 22:33If you go back to the history of Capital One,
- 22:35and we just focus on data and analytics
- 22:38from it, that completely transformed the industry we’re in.
- 22:41So I think there's
- 22:43a lot of hunger across companies to start
- 22:47connecting with universities, getting new thoughts.
- 22:50There is an issue there.
- 22:52They don't know how to convince themselves
- 22:54about the return on investment, how to think about
- 22:56what is the return they should worry about, how should they engage.
- 23:00And so there is something to be done
- 23:02in terms of a go-to-market plan, if you want to call it that,
- 23:05that says, “How do I scale this up?” Right?
- 23:09Like, I think oftentimes
- 23:11universities are set up with saying,
- 23:13“I put so much effort into development
- 23:16that if it doesn't result in something that several hundred
- 23:18thousand per year for many years, like million,
- 23:21the payoff is not there.” So I think you have to reinvent yourself
- 23:25and say, “What is the model in which I can see in five years
- 23:28this could become a multimillion dollar
- 23:30initiative, but it's going to start with a $50
- 23:32to $100K engagement?” But we think of it.
- 23:36And so what is the life cyc-, so people who can imagine
- 23:39what is the life cycle of that growth,
- 23:42there will be places like us.
- 23:45And why do we have any confidence in this, is because I can tell them
- 23:48I did it at Amazon.
- 23:49And why did they have any confidence?
- 23:50Because I said, I want to do
- 23:51it, and Amazon is underrepresented in this thing.
- 23:53And so we did it.
- 23:54But other places, they may not have ...
- 23:57I mean, if you have an internal advocate somewhere,
- 23:59that's the best way to start.
- 24:01But if you don't, then you need to figure out
- 24:04what's the way in which I'm going to gain the confidence.
- 24:06Because even the people who are there who want to do it,
- 24:09they don't know how to gain the confidence
- 24:12of whoever it is that might be releasing
- 24:14the money for them. Right?
- 24:16And so you can, if,
- 24:19if you are the recipient of the funding
- 24:22that they're going to release, that's no longer their problem.
- 24:25That is kind of, you know, your problem.
- 24:27And so it's worth investing in how to solve that problem
- 24:31and what are the pathways you can provide.
- 24:34And there's a bunch of things to explore.
- 24:36And I think it will take a bunch of experimentation,
- 24:39really.
- 24:40The way universities are,
- 24:41whoever cracks
- 24:42it will share the formula and then other universities will replicate.
- 24:45So you have to be OK with that.
- 24:46But it's OK.
- 24:47That's, that's the mission of universities.
- 24:50One of the things that we've heard is that sometimes
- 24:54these ideas of more intensive
- 24:57institutional relationship
- 24:59don't even get off the ground because there can't ...
- 25:04an agreement can't be reached with respect
- 25:06to revenue sharing and tech-transfer provisions
- 25:10that aren't even relevant at the time,
- 25:13but could become relevant sometime in the future, and I was wondering
- 25:15if you could talk about how companies
- 25:18view those challenges, and how universities
- 25:21might be able to differentiate themselves.
- 25:24Yeah, I find that the hardest part
- 25:26where universities ...
- 25:28the hardest part of the deal is
- 25:33when it's a matter of principles,
- 25:35because universities are built
- 25:37around the notion, or university cultures are built around the notion
- 25:40that the principle is the most enduring thing.
- 25:43Right?
- 25:44Anything else you do is a transactional convenience.
- 25:47And, and so if something
- 25:50conflicts with the principle,
- 25:51then the right thing,
- 25:53the university is considered to walk away,
- 25:56but increasingly that's not the case.
- 25:59And so I think there are,
- 26:02there are places that have models like Stanford,
- 26:05for example, that takes a very different view
- 26:07with some of these IP issues
- 26:09and they've had success.
- 26:12I think talking about experiences
- 26:14here, I'd say that there have been times when I've noticed,
- 26:18like this thing, you know, I think universities have to ask
- 26:21themselves, “Do I want 25%
- 26:26of, like, two pies,
- 26:29or can I be OK with 2% of, like, 200 pies?”
- 26:32Because I really think the scale is that big,
- 26:35because in having experienced that process,
- 26:39both vicariously through people here who wanted to start things,
- 26:42and not just here, at other universities
- 26:44as well, and, and myself
- 26:47having one or two rounds
- 26:49of interactions with it.
- 26:51It's, it's, it would be very hard to argue
- 26:54that we have,
- 26:58that the Office of Entrepreneurship
- 27:02is actually an advocacy
- 27:03office for entrepreneurship.
- 27:05You know, it's, it's like, it's like
- 27:08extreme regulation
- 27:09and like a preemptive ...
- 27:11and this worry that I might give away the gold mine.
- 27:15So I wonder whether there are like reasonable agreements
- 27:18to be made where people say, “Look,
- 27:20you can go do this.” And then, right,
- 27:22if things like, if,
- 27:24if it crosses this massive threshold ...
- 27:27I know there's a bunch of ways in which one
- 27:29could expansively problem solve for this,
- 27:31but the way it is right now, it's extremely hard.
- 27:35And that often also forces people to take,
- 27:38like, a year or two off and then go do it by themselves.
- 27:41I do see some universities, by the way, like ASU,
- 27:45I saw a bunch of people who really do
- 27:47get partnership from the thing.
- 27:49So there are some people breaking the mold a bit,
- 27:52but I think it has to take ...
- 27:54get traction on, on, on that front.
- 27:58Just to kind of recap some of what we've talked about.
- 28:01You've talked in the past about the four R’s
- 28:04as a way of kind of understanding
- 28:08the benefits
- 28:09for universities and private sector collaboration.
- 28:13I was just wondering if you could explain that. Yeah.
- 28:14So in fact, this came about because as we were
- 28:16scaling the model at Amazon,
- 28:19I was trying to get other businesses at Amazon to participate
- 28:24because I thought this would be a big benefit for everyone
- 28:26to have this connection.
- 28:28And I said, “But you're in a company and, you know,
- 28:31you can't measure the ROI of the money you're putting in using
- 28:34any notion of investment and return in dollars.”
- 28:37So I said, “What are the dimensions of this return?”
- 28:40And so I put forth first, it was a three R that
- 28:44I made it a four R model, but the four R’s are,
- 28:47the first one is reputation, right?
- 28:50Which means it becomes very selective in terms
- 28:52of the universities that you engage with.
- 28:55The second is recruitment.
- 28:58Third is retention.
- 29:00And the fourth is research.
- 29:03And I really think research is at the bottom.
- 29:06This gets at that IP issue.
- 29:08I kind of stack ranked it
- 29:09because the possibility of making breakthrough IP is so low.
- 29:12But that's the one that tanks most discussions, right?
- 29:16So in the case of Amazon
- 29:17and with Capital One, the way I structured
- 29:19it is, we're going to make it a gift.
- 29:23We have expectations.
- 29:24We'll make a commitment upfront
- 29:26that this gift will go on for five years,
- 29:29but annually renewable.
- 29:32And then we'll extend it, another five years,
- 29:34another five years beyond it.
- 29:36And the regulating aspect of it
- 29:38is the fact that performance during
- 29:40the year determines the renewal.
- 29:41And so we don't need to write in, like, performance
- 29:45things or spend time negotiating a statement of work, this and that.
- 29:48That took like a fair bit of time
- 29:51to get everybody aligned around it.
- 29:54Right?
- 29:56But I learned from that, and the second time
- 29:57it didn't take quite as much time.
- 29:59And so, but these four R’s are also framework.
- 30:02And I don't think every company will care
- 30:05about all the four R’s all the time.
- 30:07Right?
- 30:08And so if you find this
- 30:10framework useful, then you want to think about,
- 30:12when I'm targeting this person, what is it that they care most about?
- 30:16Sometimes there are companies
- 30:18that are kind of in a tech business,
- 30:20but they're lower down in the supply chain,
- 30:23who just want that affiliation
- 30:24with the university
- 30:25to be able to bring people, their customers, to the university,
- 30:28to host meetings at the university.
- 30:30Literally, the halo effect of having like,
- 30:32you have a really, you know, you know, you know,
- 30:37somebody, you know,
- 30:39well-endowed standing next
- 30:40to a cheesy product and then you don't think about the product,
- 30:44and it's this person with bursting muscles, you know?
- 30:47Things like that.
- 30:47And so that, that transference
- 30:49in advertising, that happens.
- 30:51But that happens in real life.
- 30:53They want the reputational glow.
- 30:56In those cases, I think the university should price it
- 30:59high enough, because literally what the university gets
- 31:01is just that money.
- 31:03There's no other fertilization
- 31:06of the university's mission that happens.
- 31:08In other cases, it's a much more complete ...
- 31:10in our case, with Capital One, what I found is
- 31:11all four R’s are coming to life.
- 31:13We do care about reputation, and we think it's a little bit
- 31:16bidirectional.
- 31:18We really care about recruitment.
- 31:20We're growing.
- 31:20We're hiring applied researchers
- 31:22in AI at tremendous rate.
- 31:26We could consume all the stu-, all the PhD students in AI
- 31:29who come out of here annually as long as they meet our bar.
- 31:32Not everybody does,
- 31:34but we have the capacity right now.
- 31:37Retention.
- 31:37We have a lot of people we have hired.
- 31:40Everybody is looking to do the next best thing.
- 31:43We want them to stay there because we want them to feel like
- 31:47their résumé is becoming more valuable by the day.
- 31:50How does it become?
- 31:51It's when they're learning more,
- 31:52they're doing more and they're learning more.
- 31:54These university engagements
- 31:56give them the sense, I'm at the cutting edge.
- 31:58I’m, I'm interacting with the top researchers in the world.
- 32:01So when, you know,
- 32:03if they get to do a joint research
- 32:04project with Gaurav or Shri or someone like, OK,
- 32:07that's awesome, I can talk about it with my colleagues outside.
- 32:11The last is research.
- 32:12The actual output of that thing.
- 32:15Once in a while, it’s something that's awesome.
- 32:17It's a very long once in a while.
- 32:19I mean, things, you know, so
- 32:21it might be different in drug discovery, etc.
- 32:24So I don't know how it works there.
- 32:25But, yeah ...
- 32:27You've mentioned
- 32:28this a number of times,
- 32:29and just in the last comment,
- 32:32that Capital One is, in a way, just a technology company
- 32:35with financial services that it offers.
- 32:38I was wondering if you could talk more, Prem,
- 32:41about how you see
- 32:42AI accelerating all of the opportunities,
- 32:46but also some of the urgency around what we've been discussing around
- 32:50university-private sector partnership.
- 32:53If, if, in other words, let's now
- 32:56talk about what if a university
- 32:58doesn't lean in
- 33:00and doesn't make an effort, is,
- 33:02is there a real threat of being left behind,
- 33:05given what we're seeing from AI?
- 33:11I think, yeah.
- 33:12I mean, even in your question,
- 33:14I hear the echoes of an answer, right,
- 33:17in the, in the sense I was trying to think about it,
- 33:20but I think,
- 33:24you know,
- 33:25I’ll raise the stakes a little bit more.
- 33:28I'll say there are universities that have already been left behind.
- 33:31They just don't know it yet.
- 33:33Because if you've not been thinking about this for a few years
- 33:36and you've not been preparing for this world
- 33:38and you don't have some things,
- 33:40there's just too much competition out there, right?
- 33:43So I think there are universities that are left behind, just like
- 33:47there are all kinds of organizations that are already left behind
- 33:51that don't know it, like people who are asking themselves, “Gee,
- 33:54how do I win the AI game?” I usually tell them,
- 33:56take six years back and, and,
- 33:58and ask yourself, “Was
- 34:00I trying to win the data and analytics game?”
- 34:03If you weren't,
- 34:05you're probably bare
- 34:06minimum six to eight years behind
- 34:08because if you weren’t laying the data
- 34:10and analytics foundations, because AI is an analytic,
- 34:13if you weren't using the analytics of the day,
- 34:16you're not certainly going to be able
- 34:17to use the most powerful analytic of belief.
- 34:20You never knew how to mobilize the power of data
- 34:22to feed the previous generations of analytics.
- 34:24Right?
- 34:25And so there is like
- 34:27a cascading and, like, thing here.
- 34:30So I think the same with university.
- 34:31I mean, a thing with USC, I'll say, like,
- 34:35you have a bachelor's in
- 34:36AI already, and then you have these other,
- 34:39you know, even before that
- 34:41there was the MS.
- 34:43I mean, there was this whole ...
- 34:44I think USC has been responding
- 34:46to the market in a very nice way,
- 34:48at least based on what, what I know.
- 34:51There was also the data science program that was joint
- 34:54across different schools here, etc.,
- 34:57but I think it's time to
- 35:00bring momentum to it.
- 35:02Like, look, most companies now
- 35:04see themselves as
- 35:06AI is the transformational power of ...
- 35:09AI is the core transformational power, right?
- 35:12And then in a university
- 35:14setting, it's hard to take one discipline
- 35:16and say, because there's just so many disciplines
- 35:18and so much richness and so much to add,
- 35:21but there is still an underlying truth
- 35:23to the fact that certain things are transformative
- 35:25in the years that they're transformative.
- 35:28And so, to me, the biggest challenge for the university
- 35:31is how to make sure all the constituencies are being carried along.
- 35:35So everybody continues to feel that sense of relevance.
- 35:38But at the same time, there's a shared acknowledgment
- 35:40that we are in the thrall of this transformative thing.
- 35:44And how do we make sure we all take advantage of it?
- 35:47And as it turns out, you know,
- 35:49if, if you're in,
- 35:51you know, the age of rocketry,
- 35:53then those are the people who are the experts.
- 35:55If you're in the age of AI,
- 35:57the experts, and that obviously need
- 35:59to be mobilized in the right way
- 36:02and leveraged so that the university
- 36:05takes itself to the next level and gets ready
- 36:07for the next wave, whatever that might be.
- 36:10Right? So I do think
- 36:12in every part
- 36:13of it, from curriculum to, to
- 36:17learning modes,
- 36:20to the, to the,
- 36:22to the dissemination of education itself, like we talked about.
- 36:26Is it for years at a stretch?
- 36:29How do you update curricula
- 36:32if the science ...
- 36:34Imagine a science, for whatever reason,
- 36:37that is advancing,
- 36:39making breakthrough advances
- 36:40every two or three months,
- 36:42but your curriculum
- 36:45committee process takes a year
- 36:47and a half to update the curriculum.
- 36:50How, how would you stay relevant?
- 36:52Right?
- 36:53And so there are these stresses that are coming out
- 36:55in the real world that, that ...
- 36:58So when I think of it as like,
- 36:59you know, education as an area where technology is not fully adopted,
- 37:03I think of it not as like in terms
- 37:05of disseminating grades or like online access, and this ...
- 37:07It's more in the terms of the fundamentals of,
- 37:11of, of the core business.
- 37:13Has, is that bringing in technology,
- 37:16how can we update this curricula faster?
- 37:18How do we do it?
- 37:19And part of it is you have to create the culture for it.
- 37:23Culture is simply the integration of the individual,
- 37:27the collective thought processes of the community,
- 37:29the history of the collective thought process of the community.
- 37:32The community has to evolve.
- 37:34This is why I think that rotation program is important.
- 37:36You need to bring in the people who are innovating fast
- 37:39so they can come and say, “But what are the enduring things?”
- 37:41And the people who are focused on the enduring things,
- 37:43they're going to those places.
- 37:44“Wait a minute.
- 37:45Are we doing this thoughtfully?
- 37:46Are we doing this responsibly?
- 37:48We shouldn't be doing this just because we can.
- 37:51It might have impacts down the road.” So that rotation,
- 37:53I think, is going to become more and more important.
- 37:58Prem Natarajan, thank you very much.