Hello, and welcome to the NVIDIA AI Podcast.
I'm your host, Noah Kravitz. In July of 2020, the University of Florida and NVIDIA unveiled a plan to build the world's fastest AI supercomputer in academia.
The public-private partnership promised to make UF one of the leading AI universities in the country. advance academic research, and help address some of the state's most complex challenges.
A year later, in September of 2021, UF rose to number five in the U.S.
News and World Reports list of the best public colleges in the U.S., The rise was, in part, a nod to the university's vision for infusing its teaching and research with AI.
And just six months after that in March of 2022, the University of Florida's Academic Health Center, UF Health, teamed up with NVIDIA to develop Syngatortron, a neural network that generates synthetic clinical data researchers can use to train other AI models in healthcare.
It's been quite a two-year run for UF, putting their AI-powered vision into practice across all aspects of higher education.
Here to tell us about his school's efforts to put AI to work is Joe Glover.
Joe is Provost and Senior Vice President of Academic Affairs at the University of Florida.
Joe, thank you so much for taking the time to join the NVIDIA AI podcast.
Welcome. Thank you, Noah. It's a pleasure to be here.
Quite an introduction there, but I wanted to hit on these points because it's really remarkable in such a short time span, even as fast as...
AI and high performance computing move. It sounds like UF has really done a lot of work in a short amount of time.
Maybe we could start by you telling the story of how this vision came to be and what the past few years have been like for you.
Sure, I'd be happy to. When people ask me about this, I tell them it's a combination of of generous philanthropy, some good decisions, a little inspiration and a few miracles here and there along the way.
And the way this this really started out was I was in a building one day and down the hallway came our vice president for advancement.
And he had a guy with him that I had met once or twice before.
His name was Chris Malachowski. I didn't know Chris well, and as you know, he was one of the co-founders of NVIDIA.
And out of the blue, Chris came up to me and he asked the question, If I were to give you an AI supercomputer, what would you do with it?
And this really caught me flat footed. We should all be so lucky to be caught with that question.
Yes, I suppose. But I hope you're more prepared for it than I was.
And after a minute or so of thought, I said to him, what we would do is we would teach AI across the entire curriculum.
We wouldn't limit it to the Computer Science Department or the College of Engineering, but we would make it available to everyone, all the students, all the faculty.
And I think Chris liked that answer at the time.
And after giving that answer, we began to really think about the possibility.
I mean, this was a tremendous offer of a gift of a really advanced technological tool.
And we began to think about how this could change everything that the university did, how we deliver education, how we prepare ourselves for the future.
We know that AI is going to be, and in many ways already is, a ubiquitous technological tool.
I'm old enough to have lived through the PC revolution in the 1980s.
I was here before there were PCs. And I know the amazing transformation that PCs introduced to universities and indeed all of society, you know, with email, the invention of the web, et cetera.
Obviously, the same thing is going to happen with artificial intelligence.
And so we began to think, how does this really fundamentally change what we do?
And the more we thought about it, the more we realized that my quick answer about AI across the curriculum was something that we really had to put into practice and really make AI not only available to all of the students at the university, but an integral part of their education at the university.
And so, as I mentioned earlier, there are many parts to this story.
One of the miraculous parts was Chris and NVIDIA teamed up to offer us this machine And it was amazing, but we actually have a modern data center which was largely empty. and was prepared to accept the machine, right?
Did we have a place to put the machine? Well, miraculously we did.
We had to put in about $15 million to upgrade the power in the HVAC systems to truly receive it and make it functional.
But we did that. And within the space of several months, the machine was delivered and between our own IT staff and the NVIDIA specialists who came along, They assembled the machine during the pandemic, which was another one of the minor miracles that occur.
They did absolutely stupendous work in the course of two months, standing this machine up, vetting it, testing it and getting it to run.
And in fact, the first large research project that we did was Gatortron, which was the precursor of Syngatortron that you mentioned during your introduction.
So around the time we were beginning to think about AI across the curriculum, the other thing that we became aware of, which we hadn't been aware of before, was the National Security Commission on Artificial Intelligence, which stood up during the Trump administration and continued into the Biden administration.
And that commission was really charged with looking at all aspects of AI and how it related to the United States and its future.
And they had a couple of conclusions there that we thought were very important to our work.
They obviously concluded that AI was very important to the future of the United States.
And they concluded that it was a great national security and economic security concern to the country. because we, the United States, are not preparing a 21st century AI-enabled workforce at the same pace that many of our competitor nations are.
And what we realized as we began to develop our AI across the curriculum philosophy and implementation is that we really have the first working model to develop that 21st century workforce at scale.
We're making AI available to all of our students at the University of Florida.
We graduate about 10,000 students per year and they go out into all sectors of the economy.
So we already think we're beginning to supply at a reasonable scale the needs of the nation.
We also realized that we couldn't be the sole players in this game.
And so we very quickly extended the use of of HyperGator, which is what we call the AI supercomputer.
We extended the use of HyperGator to the entire state university system of Florida for educational purposes at no charge.
And we extended it to the universities and the SEC, the Southeast Consortium Athletic Consortium, because we believe that these universities, if they also adopt the philosophy of AI across the curriculum, will multiply our impact in sending tens of thousands of trained AI specialists out into the economy to help build this workforce.
So we actually think we stumbled onto something which is not only important for the curriculum at the University of Florida, But it's an important model for the United States to catch up and to compete with other nations.
So you mentioned being around during the PC revolution and kind of witnessing that.
And it makes me think of when... I was doing my undergraduate work.
The internet was a couple of years out from taking public hold, but we had vax terminals and there were some early Netscape browsers running on some Macs in the lab and that sort of thing.
At the time, And I know there were people far wiser than I was at my school who had better thoughts on these things than I did.
At the time, I, for one, had no idea how the internet might shape my own life, let alone the global economy and the workforce.
When you talk about training future members of the workforce and the economy to use AI. what what does that mean i guess you know either both from an educational perspective or even just what what that notion of somebody who is entering the workforce ready to use AI or understanding how AI will affect their jobs, job prospects.
What does that mean to somebody like you and your position?
Yeah. Well, I'm a mathematician by training.
That's why I do my research in mathematics.
And I remember before there were PCs and we were all anticipating the arrival of PCs, we all thought it would be a great revolution in computation.
I would finally have the computing power of an IBM 360 on my desk.
And I would argue that's not at all what happened.
The effect of the PC revolution wasn't so much in computation as it was in communication, right?
And I think the same thing to your point will happen with the AI revolutions.
I think we begin to see glimmers of how it's going to change what we do.
But I think that there are many surprises and unanticipated consequences of it.
And so we believe that our students need to go out into the workplace, into the world, with a firm grounding and a firm understanding of the fundamentals so that when opportunities arise, they can grasp them.
Now, what does that mean in practice? We believe that it means embedding AI and AI instruction within the disciplines.
So the needs and interests of a student majoring in business, I mean, you have business analytics and FinTech developing as all the financial services incorporate.
AI into their systems. That's going to be very different from the needs and interests of a student, for example, in agriculture.
In agriculture, there is a labor shortage and we see increasingly technology being introduced in the form of drones for remote sensing of fields, mountains. water, fertilizer, et cetera.
We see it in the introduction of robotics to actually harvest crops and pack them.
And so we believe that the different colleges will create AI training and AI syllabus that particularly speaks to what they would like to achieve in that discipline.
And so, for example, our College of Business, the Warrington College of Business, they immediately grasped the potential and the promise of ai particularly in the area of business analytics and fintech And they actually changed their graduation requirements.
So now when you enter the College of Business as a freshman, one of the first courses you take is a course in business analytics and AI.
And that serves as the foundation for future courses in your major, whether it's finance or marketing or accounting, whatever it is. how those students absorb AI and integrate it into their major. is going to be quite different from how students in the College of Medicine or the College of Agriculture or the College of Engineering absorb AI understanding and fundamentals and integrate it with what they are doing in those majors.
Right, which makes perfect sense from the experiences I've had talking to folks on this podcast speak to exactly what you're saying, the There's a phrase we like to use that data is the new electricity.
And AI really exemplifies how this technology is touching and shaping every industry, every walk of life.
It makes perfect sense. And yet, as we introduced and developed this notion of AI across the curriculum, I don't see it in many other places, right?
In general, when I do see AI in other universities and other educational settings, It does seem to be largely centered in computer science departments or engineering departments.
And I would argue, of course it needs to be there, but it also needs to infuse all of these other disciplines.
Because otherwise they're not going to benefit from the technology very easily.
I couldn't agree more. And this is the burden of being ahead of the curve, so to speak, right?
You get to show the way for for others to follow.
Are there examples that come to mind of some particular, you mentioned the business college changing their graduation requirements and infusing AI from the beginning.
Are there other examples that come to mind of particularly exciting or interesting uh, developments, uh, at U of F since, um, Hyper Gator.
That's what it's called the supercomputer since Hyper Gator came online.
Yes, it's called HyperGator. So I do think that we are seeing students becoming very excited about this and incorporating it much more into their projects.
I mean, as a university, we have always encouraged undergraduates even to become involved with undergraduate research.
And now they have the opportunity to use one of the latest pieces of technology which is which is wonderful for them.
We just had a competition for entrepreneurship on campus and the winning The winning case was a group of students who had developed a system based on artificial intelligence to determine when lightning strikes in a field or a forest would actually start a fire.
And that's a big problem, obviously. Lightning strikes and forest fires.
And so they have developed an AI-based system to actually... calculate the probability that any given lightning strike, based on its characteristics, will start a fire that you have to deal with.
Wow. So I think that's an example of something I would never have thought of.
And I think the possibilities are just limitless.
As I mentioned a little bit earlier, agriculture is one of our colleges that really has moved into the forefront in terms of adopting this.
And there we see students doing all sorts of applications in robotics.
I've actually seen a machine that will harvest strawberries.
It lifts the leaves of strawberry plants off the ground. it decides which strawberries fix those packs them and lets and moves on to the next plant They're all just amazing technologies here and amazing possibilities.
I mentioned in the beginning Syngatortron and you referred to its predecessor.
So this just came online about, we're in November now as we record, so a little over six months ago.
Can you speak a little bit to that project? and what UF Health is doing with the technology?
Yes. It started with GatorTron, as I mentioned.
That was the predecessor to SyngatorTron.
So we have a health system at the University of Florida.
It's called UF Health. And as you might imagine, over the course of 10 years, we collected a lot of patient data.
Whenever you go into your doctor and your doctor examines you, the doctor usually sits at a microphone and dictates into the computer something about your medical condition, your characteristics, et cetera.
So doctors believe there is a tremendous amount of medical data and information that is trapped in all those notes.
Using natural language processing and AI
UF Health took 10 years, something like 10 billion words of information and created a medical model and is using that to extract information and data which heretofore was inaccessible.
Now, Syngatortron takes that one step further, and as you mentioned in your introduction, generates artificial data for doctors to use to further that study.
I'm not a doctor myself, so I can't give you the technical specifications, but it was an application that used the full capacity of HyperGator for considerable length of time.
My understanding is that Gatotron is the largest medical language model that's been produced so far in the United States.
Wow. Earlier in my career, I taught at the K-12 level for a little bit.
My wife is an educator. We have kids now.
I've been struck, and again, this is one of those things in retrospect to your point about PC revolution, and then I would think about it in terms of the internet. revolution changing, the implications being maybe less about computing and more about communication, as you said.
And when I think about it in an educational context, I think about this switch to learning research tools being so important. and beginning with just a simple web search, but the idea that the information is out there and at students' fingertips,
And then the skills become around learning what questions to ask, how to filter good resources from bad and that sort of thing.
In your position, and I think having the background as a mathematician makes it extra interesting, you've seen a lot in changes in the world of education.
Do you... get a sense of how skill the skill set that you know, you want your students to graduate University of Florida with, and whether they're undergrad or graduates, whatever fields they're going into,
Do you have anything of a sense of how this rise of AI might change what that core skill set is and what you want to see your students graduate with.
Yes, and I think we're beginning that conversation on the campus.
We know, for example, right now, that there are AI systems that will confidently write essays for you, right?
And so what does it mean to ask a student to write an essay and then grade it. in a period when the student can turn to his or her computer and say, please write me an essay on this.
So in the end, I think that the purpose of education is to help the students develop the critical thinking and problem-solving skills that they want. that they're going to need as they move out into the world.
And I think that's what majors are in college, right?
You may have majored in sociology or mathematics or physics or engineering or whatever.
And that's really to give you a little body of knowledge.
And it mostly in your last two years of college, on which you can practice the development of your problem solving skills and your critical skills.
So now, as you acquire this fantastically capable new tool, which is called AI, it's going to mechanize or automate some things which formerly you had to acquire the skill to do.
So I think we in education need to think about how are we going to test the development of those faculties like critical thinking and problem solving that are still going to be critically important to your functioning as a competent citizen in this economy,
But understanding that you're going to make full use of the tools that you have available, right?
Before there were calculators, we all had to know arithmetic, and oftentimes you had to do pretty sophisticated arithmetic.
If you look back at a and a high school or college examination from 1900, you can't do most of the problems because people were trained to deal with problems without all of the mechanical aids that we have now.
I think the same thing is going to happen as we move into the fully blossomed AI age.
People are going to have much more sophisticated tools available, and so they won't necessarily need to learn or retain those skills that they had to before the AI was there.
They can focus on other things that are very important and that the AI does not do for you or does not do very well for you.
Is there a feeling on campus, specifically among students who or whatever year they're at, but as they look to their life beyond university, is there a worry about AI replacing them in the workforce about the robots coming to take the jobs away, so to speak.
What's the sentiment around that? We haven't felt that on our campus.
In fact, as all campuses do, we have a career connection center, which helps connect graduating students with prospective employers.
And for the first time this fall, we had a career day specifically devoted to AI.
And we had hundreds of employers showing up looking for students with AI understanding and competences.
And we had between one and 2,000 students who went and interviewed with those employers.
So we think we see students understanding that this is going to be part of their future and embracing it.
And we think that integrating it into the curriculum and making it a normal part of education is going to prepare them to deal with the future.
We certainly know that Whereas in our generation, we might have anticipated working for one company for 30 years, That's not going to be true of current generations.
And so they need to be flexible and understand that they're going to change jobs and they need to be able to move their skill set from job to job as they change.
So we usually like to end these conversations on a bit of a forward looking note.
And in some ways, this conversation, many of them, this one in particular, has been very forward looking throughout.
So as you look five or 10 years out, do you get a sense of anything you're expecting or expecting? you know, especially curious to see how it pans out as relates to both how AI and sort of the jumpstart that University of Florida has gotten infusing AI into the curriculum, how that's going to change the campus, but then also if I can be so bold as to ask, how this is going to change education, whether higher education in particular or just more broadly?
Yes, we actually have thought about that.
We thought about it quite a bit. And we think we are in the process of building the nation's first AI university.
Now, that is a little bit of an undefined term, so let me talk for a moment about what I mean.
We mean that AI is not just an add-on to the curriculum, an additional skill that students should learn.
We believe that AI five to 10 years from now will infuse every aspect of the university's operations.
It'll be teaching, research, our outreach services to the community through Agricultural Extension and UF Health.
It will be inside the operations of the university itself.
For example, we are developing an AI-based system to try to predict student success or where students are having difficulty so that we can intervene.
So we think that AI is really going to become part of the fabric of the university.
Had anyone thought about this in the mid 80s, the first university to adopt PC would have branded themselves a PC university.
Now, we laugh at that concept right now.
But if you were really on the cutting edge then, you might have done that.
We see the future in which AI is so infused into everything the university does. that it really does become ubiquitous and change the nature of what we're doing.
And we think of that in terms of building an AI university.
And we think that 20 years from now, that'll be the norm.
Nobody will talk about being an AI university because everyone will be an AI university.
Right. Everyone has PCs now. Yep. You're absolutely right. joe this is uh we've only scratched the surface obviously but a fascinating look into really um the cutting edge of higher education.
Congratulations for all the work that you've done at U of F and I was going to say best of luck and best of luck, but I'm just excited to track the progress.
This is really fascinating on so many levels.
There are a few blog posts on the NVIDIA blog that I mentioned talking about the unveiling of the partnership and some of the other events.
But for listeners who want to learn more about what University of Florida is doing, are there some web resources that you can direct them to start with?
Yes, we actually maintain a website, ai.ufl.edu. which we talk about what we're doing in terms of research and undergraduate education and outreach.
We are also trying to take the fruits of AI out to the working professionals through our continuing continuing education department.
So there are a lot of different dimensions and angles to AI at UF, and I would encourage people to log onto that website.
Perfect. Well, Joe Glover, again, thank you so much for taking the time.
And I hope we can check in again a year, two years, however far it is down the road to see what new and exciting things you and your students have come up with to leverage You know, this technology that, as you put it, is really the PC, the Internet, whatever you want to say of today and becoming a ubiquitous part of everything we do.
Well thank you so much for having me. ¶¶
