Today, we’re embarking on a fascinating journey through time, text, and technology with our guest, Brent Seales, is the Gill Professor and Chair of the Computer Science Department at the University of Kentucky, as well as Principal Investigator for the Digital Restoration Initiative, funded in part by a major grant from the Andrew W. Mellon Foundation.

Brent, a visionary in the intersecting worlds of computer science and digital imaging, will take us through the transformative power of digitizing historical texts. Together, we’ll delve into the seismic shifts in teaching and career paths brought on by the pervasive reach of computer science into every conceivable field.

Yet, as we embrace the advancements of the digital age, Brent doesn’t shy away from addressing its darker side—particularly the tangible consequences that rapid tech adoption can have on society’s younger generation via social media’s expansive influence.

We explore Brent’s unique trajectory from a music student to a luminary in computer science—a journey shaped by the orchestral strands of a music teacher’s advice and the reverberating chords of curiosity and faith. We’ll uncover how his work at the University of Kentucky intersects with unlocking the secrets of ancient artifacts and contributes to our collective understanding of cultural heritage.

From discussions on AI’s ethical quandaries to the evocative narratives of science fiction that mirror our technological quests, Brent’s insights promise to both challenge and inspire.

So tune in as we unravel the threads of Brent Seales’ story—a tapestry woven from the codes of the past and the pixels of the future—right here on the UpWords podcast.


Brent Seales [00:00:00]:

You. In a way, it was sort of the prehistory of computer science, because if you look at the calendar, you know, 1995 was when everything really changed, and it. It moved computer science from just a sort of another field of study into a part of every field of study.

Dan Hummel [00:00:16]:

Right, right.

Brent Seales [00:00:17]:

Think about the. The redemptive act of pulling out a text that is on its way out of this world. And then we cross this chasm, because digitization creates a kind of eternal life that what you digitize is no longer subject to material degradation. It's digital. And if we preserve it. Right, it will be digital and digitally perfect from now on. So the two things that I have thought about, actually are this idea of a redemptive act which resonates with the christian mandate, and that is that because God is capable of redemption, he has made us capable of that, too. And our presence in this world can be an act of redemption.

Brent Seales [00:01:00]:

Right. And then the other thing is this idea of eternal life, of life beyond the physical frame. And so it's sort of a metaphor for that idea, which is also a strong christian idea, is that this world is not our home. Right.

Dan Hummel [00:01:13]:

Welcome to this episode of the Upwards podcast. I'm one of your hosts, Dan Hummel. I'm excited today to introduce this conversation with Brent Seals, who's the Gill professor and chair of computer science at the University of Kentucky. He's also a principal investigator for the digital restoration Initiative. And these titles should go to show that Brent is a world class scholar in computer science and digital imaging. This conversation ranged widely from Brent's training at University of Wisconsin Madison in computer science, where he got a PhD, to his work with cultural heritage artifacts and imaging them and extracting data from them, to our thoughts on science fiction, to even talking about AI. And Brent's thoughts on AI. He's been working with AI going all the way back to the 1980s.

Dan Hummel [00:02:07]:

So without further ado, here's my conversation with Brent Seals. Glad to be here on the podcast with Brent Seals. Welcome, Brent.

Brent Seales [00:02:17]:

Well, thank you, Dan. It's a real pleasure to be here.

Dan Hummel [00:02:19]:

So, Brent, you've been staying here in Madison for the week as sort of a guest of upper house. How's the week been going?

Brent Seales [00:02:26]:

It's been fantastic to be here and to enjoy the environment that you have and interact with the people. The staff's amazing. Great week.

Dan Hummel [00:02:33]:

Awesome. And I know you have some connections, particularly the UW. We'll get to those, but we're happy to have you here. We're actually talking on the day of a big public lecture you'll be giving this evening that will include some discussion of AI in it and we'll get to that topic at the end of this conversation as well. I thought we'd start Brent we really want to get to know who you are, what you do. And then just as a foreshadowing for the audience Brent's a leading scholar, a leading practitioner in computer science in a particular part of the world of computer science with digital imaging. So we're going to get to that and try to explore some of what it's been like to be in the field of computer science from basically the, which is a really important period of computer science history and then some of the particular work you've been doing. But let's start with your background Brent.

Dan Hummel [00:03:31]:

Just give us a sense of who you are where you come from and then what you do today and then we'll fill in the gaps in between those.

Brent Seales [00:03:38]:

Well my academic background started here really when I went to graduate school and stayed five years in Madison from 1986 to 91 and got my computer science phd here. Before that I had been an undergraduate in the South Louisiana and I was raised in western New York. So outside of Niagara Falls and Buffalo.

Dan Hummel [00:03:59]:

What drew you to Louisiana for your undergrad? That's far away from New York.

Brent Seales [00:04:04]:

Yeah that was a big step. I was eager to get out of my home. I think that's true for most 1718 year olds and I had experience with the deep south because my father was from eastern Texas which borders Louisiana and opened up that part of the country as an avenue for me maybe academically.

Dan Hummel [00:04:25]:

So tell us a bit about growing up in New York and particularly I think audience is interested in your relationship to Christianity. Yeah whatever you want to share about that part of your life. Sure.

Brent Seales [00:04:39]:

I grew up in a christian home and my grandparents were the founding members of a christian and missionary alliance church in Springville, New York and that was my maternal grandmother and grandfather and so my mom was the pianist for the church and we grew up doing the music and also worshipping there and I stayed in that church through my teen years and then until I left.

Dan Hummel [00:05:01]:

When did you start becoming interested in computers or in computer science as a field?

Brent Seales [00:05:07]:

Well my childhood spanned the moon landings. I was five when Neil Armstrong walked on the moon and the rapid acceleration of technology during the 70s was inspiring but what really did it for me, what captured my imagination was what was going on in Silicon Valley with the Steve Jobs story Apple computers that came online about when I was in high school and we were converting curriculum in high school, the teachers were trying to figure out how to incorporate computation into what they were offering, and I was on the forefront of that. It was a very exciting moment for me.

Dan Hummel [00:05:47]:

Can you tell us just a bit, what was so exciting about it for you?

Brent Seales [00:05:50]:

Well, my first experience with computing in high school was through punch cards that got sent off overnight to a collecting area that sort of ran them. And then we'd find out what happened the next day. And we lived through the transformation from that to a Steve Jobs Apple computer that showed up in a room in our campus. And the difference between those two things was night and day. I could feel the acceleration and the excitement around it.

Dan Hummel [00:06:19]:

Did you have computers in the home? Was that part of sort of how you grew to live?

Brent Seales [00:06:23]:

No, we did not have any computers at home. So the school was a real attractor because that's where the machine was. We had small things that you could do, like calculators. Believe it or not, hand calculators weren't a thing until the early seventy s, and some of the programmable ones were just coming out near the end of that period. And that was something I did. I got a small handheld, and you'd call it a graphing calculator now and figured out how to code up something on that, which was mean.

Dan Hummel [00:06:49]:

I think a lot of people, me included, when I was young, I was really interested in computers. I even took a Cisco networking course at my high school, but I did not make it a career choice. I ultimately was happy to play computer games and do other things on computers. Did you know, even in high school, that you wanted to make that your career? Sort of specialize in that, really learn it deeply and actually try to make money from it?

Brent Seales [00:07:13]:

I wasn't really sure. It was exciting to me and I had aptitude, but I was also a musician and I played classical violin. And so I hedged my bets and I went as a double major playing violin, thinking maybe that could evolve into something. I didn't know if I was going to be a good player or bad player, and then also a computer science major, because I did want to explore what was happening at the university level, not just in my high. Mean, we had one Apple computer, and that was, you know, that gave me a taste of what was, you know, when I got to university, the whole world opened up.

Dan Hummel [00:07:45]:

Yeah. What happened to the violin playing?

Brent Seales [00:07:48]:

I went to the deep south to University of Southwestern Louisiana and Lafayette. I went there on an orchestral scholarship. I played the orchestra the first two years of my career there. And I also picked up an academic scholarship. So I carried both of those paths forward for a couple of years until it became clear I couldn't really do both and do them both well. My music teacher sat me down and gave me. My music teacher sat me down and gave me a nice lesson in helping me understand where my talents were. And he told me that I was a good player, but probably not a great player, and that he could tell that I probably could be a great computer scientist.

Brent Seales [00:08:23]:

So that helped me be able to understand that maybe music was going to be my avocation, but not my vocation.

Dan Hummel [00:08:29]:

Makes me think of sometimes some of the best conversations, not always, but sometimes are blessings when people actually tell you what you're not good at or what you need to treat as a hobby more than the central thing. And that can be really clarifying at times.

Brent Seales [00:08:43]:

No, honestly, I've always been grateful to Bill Hayden for telling me that. He was a great violinist and he was my teacher. He'd come from juilliard, and I was lucky to have him in the deep south. And I've taken that lesson on to some of my know as I'm now a professor. Sometimes it's really important for them to know what they're good at and what they're not.

Dan Hummel [00:09:01]:

Well, so you have a computer science degree. I don't know a ton about that world, but I know that not everyone goes on to graduate work after that. In fact, many people jump right into the private industry or other areas. What made you want to go to grad school and sort of really dive into the scholarly, academic part of it?

Brent Seales [00:09:25]:

What was happening in Louisiana at the time? And the university actually called itself Silicon Bayou. There's actually a lot of great computer science going on in Lafayette, Louisiana, at the time, but it was driving the oil industry because know were doing optimization. It was a lot of development in that region around energy. And so most of my colleagues went with their undergraduate degree into the industry, and they followed that route. What I felt, though, was that I wasn't prepared yet. I needed to know more about what was happening in the field. And one of the things that helped me know that was that I took the GRE exam, and there was actually a subject exam on the GRE in computer science, and I didn't do very well on it because there were parts of our curriculum that didn't cover what was asked on the GRE subject exam. And so what it sort of told me is that I needed to bake just a little bit longer to make sure that I really knew what the field was about and where my place in it was going to be.

Brent Seales [00:10:26]:

And graduate school is my path.

Dan Hummel [00:10:28]:

And uw, how did that emerge on the radar as the place you'd want to go?

Brent Seales [00:10:34]:

UW was ranked in the top ten public universities in computer science at the time. And I had this vision of Wisconsin that, of course wasn't accurate, but it was very positive. And I came up to visit. What they offered me when I applied was a guaranteed four years of funding. And I figured if I couldn't get a PhD in four or five years, I should move on anyway. So it just seemed like economically a very responsible move. It also was just a beautiful trip. I came here and I saw the lakes and I became enamored with the department, which had the kind of energy I'd never seen before.

Brent Seales [00:11:08]:

And it really did feel like a top ten department.

Dan Hummel [00:11:11]:

Yeah. And I think many of us who I also went to grad school at UW, we do that trip to figure out if UW is the place to come. And it's just a beautiful city, and for most people it becomes a good decision. And it actually indicative of certain parts of the year, not all parts of the year, when you come in the spring for the next fall. Well, I think it'd be interesting to just hear what it was like to be a grad student in computer science in the late 80s, early ninety s. And I know you teach grad students today, so I don't know how similar is it today versus 30 years ago? Is the same type of work or. I know it's obviously different equipment and instruments you're using, but what was it like to be a grad student? What's the type of work you did?

Brent Seales [00:11:52]:

That's an interesting question. I haven't thought about that in a while. But it was pre Internet. Okay. So the field was completely different in terms of the way we ran the equipment and the way that you had to gather around the equipment. So we had places you had to actually come and be on site. I would say it was a lot more social in the sense that we were gathered together on the machines in the same rooms, and that was required because you didn't have connections at home where you could ubiquitously connect to the machine. And I think also the field had not yet gone to the place where the personal computer was in every home.

Brent Seales [00:12:29]:

And certainly mobile computing hadn't happened. In a way, it was sort of the prehistory of computer science, because if you look at the calendar, 1995 was when everything really changed and it moved computer science from just sort of another field of study into a part of every field of study.

Dan Hummel [00:12:47]:

Right, right. So I've read a few books on the history of Silicon Valley and early computing, and one of the images that I remember from that is this reality that there were places on Stanford or wherever where there were computers, and if you wanted to use it, you had to reserve a time on it. The famous stories are people who have big breakthroughs, but they're reserving a computer from, like, two to 03:00 a.m. On Thursday mornings or something. Does that resonate with how it was at UW at the time? Like, you were sort of very social, as you mentioned, but jockeying for limited time with the few computers that were available.

Brent Seales [00:13:27]:

Yeah, that's right. And more of that happened during my undergraduate period from 82 to 86. We were moving toward deployed personal computers, but the computers were all networked and they still had to talk to servers. And so there was reserving time. There were machine rooms where you had to go in, and that's where all the machines were. And these things have all gone the way of the dinosaur, but that's the way it worked at the time.

Dan Hummel [00:13:51]:

Yeah, it sounds very frustrating, someone in the 21st century, that that's how. But I guess it's how anything like a library or a lab would have a similar dynamic, where there's certain hours you have to go there to actually access the things that are in there. But I think with computers, we all carry them around all the time now. Well, one other track, as you're in grad school, is that you're obviously thinking about what's next after grad school. And we talked a bit about this last night when we were meeting with some students. But just the question of sort of discerning where to go after grad school with a phd in computer science. Yeah, just talk through what were you thinking as you were ending grad school? And ultimately, where did you land after that?

Brent Seales [00:14:41]:

Well, it was difficult to know. We spent five years here, my wife and I, and we together graduated, my wife with her master's degree and me with my phd, and went off from there. And it seemed like the university life was going to be the right thing for me. I really enjoyed the teaching that I'd done, and the research environment was one that inspired me. But at the time, academic positions were few and far between. It was a difficult period economically, so I did explore the commercial side, and what I just found is that I wasn't getting excited by the directions the products were going, and there were possible spots in the defense community, but that didn't excite me and I really didn't know what to do. So I think I went to my first academic job with the idea that I would see how it went.

Dan Hummel [00:15:26]:

Okay, so describe with us, where did you land and what was that job like?

Brent Seales [00:15:32]:

I ended up at the University of Kentucky in Lexington, Kentucky. And it was a small department and there was a faculty member there who had left Wisconsin's department, actually, and so knew very well my advisor and probably really helped make a way for me there. The department gave me an opportunity to build my own thing, which was exciting because other departments that he could have gone to were either less developed than that, maybe not even a graduate program, or we're super developed, where I might not have a chance for a while to build my own program or think outside the box on the direction my research might go. So at the time, I don't think I saw it as the perfect fit. But then we grew into something that actually fit really well.

Dan Hummel [00:16:17]:

Yeah. And that's an interesting dynamic. I remember we were talking about this yesterday that maybe your top sort of, on a piece of paper, top ranked places might not have actually been the healthiest places or with the most potential for an incoming young scholar because there's bigger names in the department. There's so much attention and structure already, but in some of the less centralized places, maybe a place like Kentucky, there was actually a lot more open area to dream up new things and I guess develop your own team as well.

Brent Seales [00:16:54]:

Well, that's right. And I also started to realize that one of the things that was a driver for me was the interaction with students and being able to see them develop. The student groups that we had in the day in the 90s when I was first starting my career were, they were amazing people. I mean, they were coming from Appalachia or from southern Kentucky, northern Tennessee, those areas. And the opportunity they had at Kentucky was the biggest thing they could imagine. And then I got to see them develop. Yeah, I'm not sure it would have been the same vibe for me anyway at a place like MIT or UC Berkeley. Those are the places that people before me from Wisconsin were going to get faculty positions.

Brent Seales [00:17:37]:

And, yeah, I did kind of miss the idea that I would be associated with a name that had that kind of cachet because the University of Kentucky wasn't going to be on anyone's top ten list for branding. Right, right. But then I quickly resolved myself to the idea that maybe the brand is really not as important as what I'm enabled to do, and if it aligns, then that's where the important part is.

Dan Hummel [00:18:03]:

Okay, so you're at Kentucky. Most people, if they've heard of you, probably know you for your work around digital imaging and cultural heritage pieces, ancient texts. Was that always part of your research profile, or is that something that developed? And how did that become the thing that you really spent a lot of time focusing on?

Brent Seales [00:18:21]:

The things that I'm known for now, especially around heritage science, evolved fairly late, actually, in my career, they were only a part of the portfolio because I'm an imaging specialist, and I did computer vision here at Wisconsin. That thread found its way into a lot of different applications. So I had done work in surgery, for example, in theaters, where a camera is crucial for being able to do a surgical procedure, in places around visualization, where you want to set up an environment so that someone can remote control an object. We also had work that we were doing in navigation. So I'd worked a little bit in Europe on self navigation for space exploration, for example. So those things were all there. But then when the Internet happened, and especially the move toward digital libraries occurred, and imaging was so prominent a part of that, it sort of pushed me toward antiquities and digitizing objects.

Dan Hummel [00:19:21]:

And what were the first types of objects? You mentioned antiquities there. What were the types of things that you were imaging? And I assume people were inviting you to do this or wanting this. What was the early part of that, sort of that new type of work?

Brent Seales [00:19:36]:

Like early on, it was really just about facsimile. Digitization was new, and the idea was that we could create a model, a proxy that could be broadcast on a network. That was what the World wide Web was all about. Right? The protocol and then the remote network that allowed you access information. So we wanted the information to be visual and to be immersive. And so the idea was to build a very high fidelity model of something that maybe is in a museum or a library, let's say it's a painting, and then allow someone remotely to experience that painting. And the step required that I was interested in to do that was all about making that digital model.

Dan Hummel [00:20:18]:

Right? So it makes sense to me how you would scan something flat, like a piece of paper or a painting, but what you're really well known for, or at least how I know you, is for scanning things that are folded or crumpled up or three dimensional. How much harder is it to go from on that type of stuff?

Brent Seales [00:20:44]:

Well, in fact, we were doing three d at the very beginning, but the idea was just to make a facsimile. It wasn't to extract any new information or to do anything other than make an accurate representation and then transmit it over the network. So we were doing the acquisition and then the manipulation of all of that modeling, and then the end display. And those were sort of the three phases, and my work fit into that pipeline perfectly. What became apparent, though, was that we were going to move not from just making facsimile, but from collecting data. And that shift over that decade was something that I would like to believe we pioneered, or at least we're on the frontier of, because it was a shift in thinking about the digital library. It's like we're not just going to make copies of things that you can see remotely. We're going to actually collect data, maybe even over collect data that can tell us much more about the objects than just how they look, how are they aging? Is there anything hidden that we can reveal? Can we run a simulation using what we collected to be able to tell something more about how this object behaves? Right.

Brent Seales [00:21:50]:

All of those became new questions we could pose and we could answer if we started to think about the process as more than just facsimile.

Dan Hummel [00:21:57]:

Is there any moment as you're developing this? And by the way, you said this decade, what decade are you thinking of?

Brent Seales [00:22:02]:

Well, that was the ods.

Dan Hummel [00:22:03]:

The ods, okay.

Brent Seales [00:22:04]:

Yeah. Because that was five years after the Internet. Everybody was worried about just accessing the data. But soon we started to change our mindset from just digitization. Right. And acquisition to this broader idea of data collection around new questions that we could pose and answer.

Dan Hummel [00:22:21]:

Right. Okay. So the shift from mirror imaging to actually extracting data, were there any moments as you were making that shift, where it really clicked for you that this has potential in sort of a way that could be revolutionary for the way we understand, particularly ancient texts?

Brent Seales [00:22:38]:

Oh, absolutely. Early on, one of my projects was an imaging project with. It started out with the Beowulf manuscript as being central, but a bigger collection at the british library that included many more manuscripts than just that iconic one. We were imaging, and this was in partnership with a professor in the english department who was a medievalist at the University of Kentucky. We were imaging using different spectral components. So you hit the manuscript with part of the spectrum that maybe isn't even visible to your naked eye, like infrared, and then you'd image that to see if it would create the ability to see more on the page than you could with the naked eye. And that was when I started to put it all together. It's like, wait a minute, we're now extracting more than just a facsimile of what you would see if you sat in the room and looked at it.

Brent Seales [00:23:28]:

We're actually giving you a view of this thing that you would never be able to see if you just sat there and looked at it. Right? So that was a huge kind of leap forward, an aha moment. And we started to think, okay, what are the other ways that we could do this?

Dan Hummel [00:23:41]:

And if an aha moment for you. How was your work received within the museum archaeology curating community? How did they respond to this idea? Pretty revolutionary idea, that many of the things they were holding, that maybe there was significance, but there was much more significance if you could actually read what they were. And you're coming in as a computer scientist, I'm sure you have a team. Were they eager to share those things with you, or was that a case you had to make?

Brent Seales [00:24:10]:

It was a case I had to make. I think most of the diffusion of these new technologies into fields meet with resistance initially because people are skeptical as to whether a true transformation is going to take place or if anything new is really going to come of it. But computer science, even going back to the community, I felt as an undergrad in the way that we had to do computer science. It's always been collaborative. For me, that collaboration is compelling. And I don't mean just within computer science, but across fields. I mean, even back in my Wisconsin days, the center for the study of the atmosphere and climate Science was almost connected to the computer science building. They were right across the street, and those guys participated in seminars, and it was an application where computing clearly was going to play a role, because understanding climate fundamentally tied to these big models that we run in simulation.

Brent Seales [00:25:05]:

So, yeah, I've always known the collaborative part was there. And making the case in heritage science, it had its own challenges, but ultimately, of course, people could see the power of what was happening.

Dan Hummel [00:25:18]:

Right, okay, so you have this experience at the British Museum. Clearly there's a lot of potential here, but you've been doing this work since then, so it's been another 15 years or so. Just give us a sense. Is there a highlight or sort of maybe the highlights in the future and you're excited about it, but what's the sort of defining mark of this new way of really analyzing these ancient texts that we couldn't read otherwise?

Brent Seales [00:25:45]:

Let's see, what's the metaphor for people who ski in Wisconsin? I think I got out over my skis a little bit with some of the things that we wanted to do, but the technology wouldn't really support it at the time. So there were some things, like using computed tomography to see inside an object. We were really advanced on that thinking, but the technology wasn't as advanced as what we wanted to do. So some of that came in underneath us, but realizing the potential of that. In the old days, you would take an object like a closed book that can't be opened, and you would make a model just of what you could see, and you would concede that we can't open it. So the model that we would deliver would be what it looks like. But tomography, x ray, and other kinds of imaging allow us to open up the interior of that, even if it's non invasive. And then we have a completely new world of data, not facsimile, that we can then work on algorithmically.

Brent Seales [00:26:45]:

And we knew that was going to be a thing in the odds, and that's been almost 20 years ago now, but it's taken a while for the technology to catch up to the place where that's robust, where we can just do that every day, and we're there now.

Dan Hummel [00:27:00]:

Is that sort of the defining work you're doing now at Kentucky? Is this type of heritage work with all different types of objects?

Brent Seales [00:27:09]:

It is. We narrowed down our focus to a few iconic objects that no one had ever tackled before, because there were no known methods that could get at what might be inside. And so over time, we made contributions in a lot of these different fields. Flattening out wrinkled manuscript papers, for example, or using ultraviolet light to see things that otherwise would be erased to the naked eye. But ultimately, our place, I think, in the world has come from the interior of objects that cannot be opened at all. And doing that completely non invasively, using x ray.

Dan Hummel [00:27:46]:

What are some of the things you've decided or you did in the recent past that are these sticky problems?

Brent Seales [00:27:54]:

We've had two or three really amazing results that have come from our work in advancing the software and our thinking around this problem. One is with a manuscript in the Morgan library, which was established by Morgan and exists in Manhattan, Lower Manhattan. And I don't know if it's lower Manhattan, but the Morgan library is in Manhattan, and they have this manuscript there which J. P. Morgan acquired in the. It's one of the earliest copies of the acts of the apostles. So it's a biblical manuscript. It's damaged fairly badly, and so the pages can't be turned.

Brent Seales [00:28:35]:

And so we were able to work with the Morgan library to be able to image it using tomography. And we can pull out every one of the pages non invasively using virtual unwrapping. And so it's been an amazing experience to do that work. And also the binding of this particular manuscript is a transitional binding. It's a binding that shows how book binding evolves over time. So it's kind of a dual manuscript where you also have this binding part that makes it really, really valuable. And of course, we didn't have to damage it or even do any restoration. We put it into the machine, collected the data, and from there it was all virtual wrapping.

Dan Hummel [00:29:13]:

That's fascinating. That's one example, the book of acts. I know you've done other biblical related material as well. Is there anything as a Christian? Is there an extra dimension of either stress or excitement when you're working with documents that might actually affect potentially or at least fill in what we know about the history of the Bible, the history of the church? I imagine that could be a little more invested interest if it's coming from your religious tradition, but maybe not.

Brent Seales [00:29:48]:

Well, no. I mean, I am a Christian, and this material is valuable to me for more reasons than just its historicity or its rarity. Right. I think that I don't spend a lot of time, though, worrying about what something might reveal. And I guess that's because going back to my religious experience here in Madison, all truth is God's truth. Right. Sort of a reformed idea that we learned in church here. Right, at Geneva Chapel.

Brent Seales [00:30:21]:

Right. We talked about that and the idea that when we reveal through our work things that we believe are true, facts, stories, we don't fear those things because it's all part of what God's done, what God's created. And so I've thought more about how do I stay true to what I'm supposed to be doing and who should I be working with? Who can fill in the parts that I'm not very good at because I am not a biblical scholar and I don't really read the ancient languages. So I need to have teams of people who are willing to work with me and maybe they worry about it. I'm not sure, but I've been more captured just by the wonder of it. Because if you see how some of these things look, I've learned to put on the glasses of wonder and awe to realize that this thing, 2000 years old that looks terrible on the outside, is maybe one of the most valuable things in the collection once we realize what's inside.

Dan Hummel [00:31:20]:

Right. Yeah. And I can imagine. I hear you worry is one potential way. I think there's also just, like, an excitement. Maybe that's what you were just describing of some of the importance of some of this stuff. Could be. I wonder if you've ever reflect, as I just look at what you do.

Dan Hummel [00:31:36]:

And particularly there's an interesting sort of mediated way that I understand what you do, which is it's usually depicted through, like, a documentary or a film or something, where there's obviously a lot of visual representation of what the process actually is. And it makes it look almost like magic. Right. You see this really grainy thing or like a piece of charcoal that's actually a scroll, and then sort of there's special effects on the screen that are making it, and then suddenly you're reading an ancient language. I wonder. Obviously, that's a way to just give us laypeople an understanding of the process. But it does look to me like there's an interesting theological or christian perspective on this, which is something about pulling things out of a natural process. Or sort of, if you were to leave these things without this technology, they would be inaccessible to humans.

Dan Hummel [00:32:32]:

But through the technology, it does in some way do a miraculous thing. It allows you to do something people before the 20th century certainly could never imagine actually doing. Do you ever reflect on that, like, just how sort of interesting and almost and potentially resonant with a christian worldview this type of work is?

Brent Seales [00:32:54]:

Well, I have thought about that, and in prior interviews, I sometimes have used the word redemption, which is primarily now a religious word. But think about the redemptive act of pulling out a text that is on its way out of this world, and then we cross this chasm, because digitization creates a kind of eternal life that what you digitize is no longer subject to material degradation. It's digital, and if we preserve it, right, it will be digital and digitally perfect from now on. So the two things that I have thought about, actually are this idea of a redemptive act which resonates with the christian mandate, and that is that because God is capable of redemption, he has made us capable of that, too. And our presence in this world can be an act of redemption. Right. And then the other thing is this idea of eternal life, of life beyond the physical frame. And so it's sort of a metaphor for that idea, which is also a strong christian idea, is that this world is not our home.

Brent Seales [00:34:01]:


Dan Hummel [00:34:02]:

Very good reflections. That's deeper than I thought about it, but that's very interesting. And, yeah, I think there's a lot of types of work that have these redemptive elements to them. But it's a very interesting one when you tie it with also this interesting digital eternal state. Okay, two more big topics I want to talk about. The first is just getting some of your perspective as someone who's been on sort of right in the midst of the rise of computer science from the, as you mentioned, it's now become part of essentially every endeavor, certainly in the academy and in industry as well. Computers are involved, computer scientists are needed. I wonder, any general thoughts on just sort of.

Dan Hummel [00:34:47]:

You mentioned a little, but just the career of computer science as a field and how it's sort of, I know here at UW, it's the largest major. That was probably not the case in the 1980s or 19. In fact, I think it was just a few years ago that became the largest one. But as a professor for all those years, how have you observed computer science develop within the broader higher education world?

Brent Seales [00:35:11]:

Well, it has diffused into every field. And I mean, that's gratifying because the field has created tools that are usable enough and concepts that are usable enough that other fields can simply integrate them and then build. And we're not going to move away from the need for people to continue to innovate in the pure computer space, but we've created this whole new set of interplays right between the tools and the concepts that are now diffused into new areas. So you have data science people who are working in hospitals, you have data scientists who are working for sports teams, you have developers who are working in financial institutions, trying to keep people from stealing your credit card and so forth and so on. Right. I mean, how can I be anything but really excited about seeing a field that I sort of just stumbled into because of the enthusiasm back in the day of people like Steve Jobs. Right. Grow to the point where it's a global thing and it's diffused into literally every area of our lives.

Dan Hummel [00:36:20]:

Yeah. Does that diffusion have any, I'll just access directly any downsides. And I think particularly about teaching. Like if you're teaching in a computer science class, I assume at some point, maybe back in the maybe early ninety s or something, you could assume that most of the people in the class were going in a certain career trajectory, and so it'd be very clear what to teach them to get them there. And as you just mentioned, that that's not the case anymore. How have you seen teaching change, and has it been a challenge that sort of computer scientists, people are coming into your classroom for one of hundreds of different reasons of what they want to actually use it for.

Brent Seales [00:36:54]:

I would say for me, it happened about the late aughts, 2008, 2009. The iPhone was out. Social media was really on the uptake. And what I saw was the acceleration of an adoption of computing had exceeded the capacity of the institution of higher ed, right. To sort of keep up in terms of what they offered and the curriculum. And so there is a downside. I mean, people are eager to adopt new things, and sometimes those new things aren't incredibly healthy. I think we're seeing now studies around social media.

Brent Seales [00:37:26]:

Let's just take that and some of the downsides of teenage immersion in social media that we never really anticipated. And parents have been so eager to adopt the devices and the network and the content that maybe they've not been cautious enough, right. About allowing people to take a deep dive into what can be sort of a cesspool, right, of concepts and interactions that are just not appropriate for a nine year old or a twelve year old. Right. So that's just one example of where there's a huge downside, right. And I think it's the role of the university and the computer science department to try to have a voice in this. And I don't know that we've been that good at doing that, actually, in terms of research.

Dan Hummel [00:38:10]:

So you spent a lot of time doing research at Kentucky. Has research changed over the decades? We mentioned the Internet coming in the 90s as a big change, but fundamentally, are you asking similar types of questions, or has the research environment pretty radically changed as well since the 90s?

Brent Seales [00:38:31]:

Yeah, research has changed tremendously, and there's been a shift of research power, if you will, or focus from the universities to corporate entities. So you'll see Google, Facebook, Microsoft making advances because they have the infrastructure to do so that many of the universities can't make. And while they offer some of that infrastructure for hire, and while our national organizations like the National Science foundation and others try to build global and national infrastructures that can be used by academic researchers, I think right now the focus has shifted and it's hard to get it back. Economic power is with corporations, usually not with the government. And so you have researchers competing for enough funding to be able to make advances, but they're competing against deep pocket corporations that can make those advances directly. And I don't know what it's going to do for the research enterprise of the academic side of things. I think there's going to be some tension there.

Dan Hummel [00:39:38]:

When I think of, I'm a historian, I've really benefited from Google books and particularly the stuff that's not copyrighted. So right now it's 1922 or something like that. Anything before that you get total access to, and they're very clean, tend to be very clean scans. I noticed that many of those scans are coming from universities like they have University of Michigan or University of Wisconsin in the inside of the COVID As someone who's sort of in that field, what do you make of something like Google books? It's right at that intersection of, it seems like corporate, and yet they're also somehow relying on university labor or insight, I'm not sure. But what do you make of something like Google books?

Brent Seales [00:40:20]:

Well, I worked at Google as a visiting scientist back in 2012 13 academic year and was tremendously impressed with what at the time. They were doing follow on work on Google books at the time, which included Google Arts and humanities, what they were calling at the time the cultural institute, creating large archives of material that they could distribute for free, that you could search and research. I think those are all really good things. And as far as corporate social responsibility goes, I think those are tremendous assets if they can remain open and sort of revenue neutral. And the Google books thing, have you played with the Ngram viewer?

Dan Hummel [00:41:01]:

Yes, it's really fun. I don't know how much. Yes, I played with it a lot. I've never sort of footnoted that or something. I don't know if I need to know more about it to know if it's footnote worthy. But it's really interesting to track things, and it actually gives you interesting clues on intellectual cultural trends as you see words come and go over the years.

Brent Seales [00:41:21]:

Yeah, it's tremendous. And the paper that was initially written about that work, Stephen Pinker and others, I think it appeared in nature, maybe was kind of a seminal expose of things like distant reading and other things that you could do with a large corpus like that. And what's interesting is that that feeds in directly to where we are with artificial intelligence right now, because it's those corpora. Right. That the artificial intelligence algorithms are using to be able to do this next step, like large language models, for example.

Dan Hummel [00:41:51]:

Well, let's get to that. I know your talk here at upper house has AI in the title, and that's going to be something you talk about. And I've heard a few of your comments as you've been with us this week, but maybe to start it, what has been your relationship to AI? And I know there's like debates about what is AI? And that depending on how you define it. It changes how you tell the story. But as you understand it, what's been your relationship to AI going all the way back to your time in grad school?

Brent Seales [00:42:17]:

Yeah, I'm actually going to talk about this a little bit tonight. I entered grad school in 1986, which actually was the beginning of this large slide in AI popularity and ended in the AI winter. The second AI winter of the came with the idea that I might actually study AI, and I was really intrigued and excited by what I might learn. But what I found when I got to graduate school was that the field was struggling, people weren't making progress, the computing was behind. You could say that maybe the AI people were out over their skis because they had visions and dreams of what was possible, but computationally they didn't have the underpinnings for that.

Dan Hummel [00:42:54]:

Is that the winter? Is that what the AI winter is?

Brent Seales [00:42:55]:

Winter? Yeah. So there were big promises made in the late seventy s to early to mid 80s about what was coming, and then it didn't materialize. So by the time I got to grad school, people were a little bit depressed about pursuing those ideas. And I don't know that I could have even gotten a job had I at the time pursued a pure AI PhD. I chose computer vision because perception is a part of AI, and it turned out to be a masterstroke because vision and imaging has become a part of almost everything. And even now, if you look at the AI breakthroughs, they're almost all image based. They're built around the idea of generating an image from a textual query, for example, or being able to navigate using only images from a camera, which the Tesla can do now. And Elon Musk's been very vocal.

Brent Seales [00:43:47]:

I don't know if you've read the biography by Walter Isaacson, but in the book he talks about how he really didn't want to use any sensors besides just cameras. That's a pure computer vision problem. Really interesting.

Dan Hummel [00:43:59]:

Okay, so that's a big part of AI, and you've been relating to that then since for your whole career. I think most of us know AI based on Sci-Fi and sort of certain images of rogue AI, Terminator and other things. And then obviously in the last year or so Chat GPT and I guess it would be text based, but you can tell it to give you a picture. I know in the academic world, particularly in the humanities, there's a ton of concern around plagiarism, the idea you can just sort of tell it to write you an essay on this topic and then it passes for a B minus, and tons of students will do that. Where do you hope the conversation goes with AI? I doubt it's to the maybe, maybe. But what are we missing in the AI conversation, based on your perspective in, yeah, what should we be paying attention to?

Brent Seales [00:44:52]:

Well, I think we should be paying attention to the human know. Marvin Minsky defined AI to know, getting a computer or a mechanical system to do something that you think really only a human can do, that's sort of the definition of AI. And that definition includes what it means to be human. Right. And I think that's the part of the conversation we're not talking about enough, because the things that are uniquely human telling stories, does that mean that AI is going to be the big storyteller and humans aren't? What about human dignity? And what about the erosion of, or the support of dignity as humans with the things that we invent? So I think that it actually gives us a huge opportunity. The development of AI. And AI is a field of computer science, is what it is, and it's made up of various techniques that approach this ideal that Marvin Minsky sort of defined in one sentence. The techniques then come in as things like machine learning or other kinds of names that we have for what ends up becoming this field of AI.

Brent Seales [00:45:57]:

But I think it gives us an opportunity to talk about what it means to be human and to understand maybe better, or to refine or to not forget that that's really what's at stake here and that's what's central. Not so much what we're doing with the machine, but how it's going to affect what it means to be human.

Dan Hummel [00:46:14]:

For what I just said about Sci-Fi I like Sci-Fi I read a lot. I'm a huge Star Trek fan, all that kind of stuff, old Star Trek fan. And I found it interesting to just compare where we're at historically, as far as I understand it, with some of the visions of fictional visions, but visions that were the best of them, really trying to grapple with foundational moral or ethical questions and then using the imagination to set up interesting stories around them. Is there any science fiction, any sort of classic science fiction or anything around AI that you think got part of it right, or at least a moral question right, or something like that? I'm hoping you consume some of that.

Brent Seales [00:46:56]:

Yeah. I was a science fiction fantasy reader in the course. Tolkien played a huge role, as well as C. S. Lewis. As well as C. S. Lewis.

Brent Seales [00:47:06]:

And I read Isaac Asimov, and I think that would be the first answer that most people would say in the foundation series and the laws of robotics and that kind of know. I think that was a huge contribution at the time because no one was going that far. We were looking at things mechanistically, like sort of the know, but nobody was thinking about the deeper questions. And then you get some of the movies like the Blade Runner. Blade Runner series. Yeah. And again, deep questions that were going far beyond where we were technically at the time, but posing really important questions. Should we get there with the technology? Right.

Brent Seales [00:47:43]:


Dan Hummel [00:47:43]:

And Blade Runner. And we'll get off the nerd horse here in a minute. But Blade Runner was a very interesting movie, but based on a short story by Philip K. Dick, who was one of these very voluminous Sci-Fi writers of the. He's someone I found really, along with Asimov, someone who almost all of his work, he wrote dozens of novels. And everything else was about this question of what does it mean to be human? And using robots, androids, basically robots that look human as the interesting way to question that. Right. So a lot of his stories are about ais that look like they're human and questioning, are they really human? Or what would it mean for them to.

Dan Hummel [00:48:21]:

They're striving to be human, and what does that actually mean? So anyway, I find a lot of that really interesting and actually relevant to today, even though it was written many, many decades ago.

Brent Seales [00:48:30]:

Well, I honestly should go back and read some things that you're reminding me now of Robert Heinlein as well. Some of the things there were very motivating and even going back to Aldous Huxley and were, there were many strands of thinking in those early writers. I've always said the artists, the writers, the humanities scholars, they're really the Canaries in the mind. Right? I mean, they see things coming before maybe the rest of humanity does. These are really important questions. Yeah.

Dan Hummel [00:48:59]:

And I think it was C. S. Lewis, it might have been Tolkien, who talked about science fiction as the modern era's. You know, ancients had myths and gods in the clouds that did magical things and you'd sort of tell your morality tales through those. And today the magic is through lasers and warp engines and things that can actually happen but are the way we work through as a culture a lot of the moral issues that we have. And I think that's probably right. And certainly more, there's hard science fiction, which is much more engaged in the science and trying to actually be much more tied what we know and sort of maybe be a little more realistic. But then there's the more fantastical stuff that really opens up the moral world as well.