Curious Conversations, a Research Podcast

Travis (00:04)

What are the questions related to artificial intelligence that aren't only going unanswered, but are often going unasked? Obviously, there is no shortage of conversations and questions related to this topic, but in my experience, most of those boil down to some flavor of what can it do? Or is all this going to end up like Terminator 2? So, what are the questions that we're simply not talking about enough? What are the conversations that we really need to have? Well, obviously, if I knew that, we'd already be having them on the podcast.

But thankfully, Virginia Tech's Rishi Jaitly is an expert in this very subject.

Rishi Jaitly (00:37)

I think one layer of question missing often from the conversation about AI acceleration or AI doom is a question about human and human institutions and their evolution and elevation.

Travis (00:51)

Rishi is a professor of practice and distinguished humanities fellow at Virginia Tech. He founded the Institute for Leadership and Technology at the University and is also a co-chair of Virginia Tech's AI Futures Working Group. And until recently, he served as a senior advisor to OpenAI. So Rishi was the perfect person for me to ask all my questions and to answer them with even better questions. I'm Travis Williams, and this is Virginia Tech's Curious Conversations.

Travis (01:23)

I thought maybe maybe a place to start this conversation might be to simply of lay out the landscape of what we're talking about with artificial intelligence because I know we have large language models, and then I know there's generative AI. But with artificial intelligence, what are the the different, I guess kinds or flavors of it that we're talking about in this conversation.

Rishi Jaitly (01:41)

Well, thank you again, Travis, for having me on Curious Conversations. I'm a big fan. I think you're right, artificial intelligence is a broad, big picture, catch-all term to describe many things. in fact, if you take a step back, the origins of the term artificial intelligence date back to the nineteen fifties and a conference at Dartmouth College featuring a range of computer scientists wrestling with questions, early questions of machine learning and machine capabilities and data science. So the term has actually been in the zeitgeist and in the conversation in and around computing for the better part of three quarters of a century. And of course these days there are conversations in Washington about rebranding the term and and swapping out the term artificial to include the word super. But in broad terms, what we're talking about is computers, computers that have the ability to do something generative, which means predictive, actually, right? So these are machines that with large amounts of data, large amounts of energy, large amounts of chips, are able to teach themselves in using statistics and prediction engines, really, all all range of things. So I think the the specific consumer and enterprise manifestation of this technology, of course first occurred in the fall of twenty twenty two with the launch of Chat GPT, but the term itself refers, I think, to the to the underlying science, which is about data, which is about compute, which is about energy. And when you scale all three of those to unimaginable levels, what you end up having is systems capable of prediction, of learning, in ways that frankly many AI researchers are still trying to understand.

Travis (03:33)

Yeah, well I loved you mentioned that they this thing started it sounded like with a lot of questions. And it it seems like that it continues to generate a lot of questions.

Rishi Jaitly (03:43)

That's right.

Travis (03:43)

and so many of the conversations that I hear are about a lot of these questions, but I heard you recently in an interview say that you you didn't think that we were asking or most people were asking the right questions about this artificial intelligence. And so maybe I'm kind of flipping the script here of the normal podcast, but I'm just curious what

What do you think some of the questions are that we're not asking enough?

Rishi Jaitly (04:05)

Well, thanks, Travis. I think you're right that we live in a time that is AI ascendant, but you have a lot of folks jumping to answers, jumping to takes, jumping to opinions, right? So and it it's it's really interesting. In a time where we have a technology with enormous potential to challenge us, to teach us, to upgrade in in clear but also not so clear ways, human capabilities, I think the right thing to do is to ask.

The right questions. And I think one layer of question missing often from the conversation about AI acceleration or AI doom is a question about human and human institutions and their evolution and elevation. Again, the extent to which humans and human institutions necessarily might, should, could evolve and elevate in this new era. You know, a lot of the conversation centers on.

Correctly, I think, on the environment, on energy, on compute, on global supply chains, on use cases, on guardrails, on safety, any number of things. But I think ultimately what we miss often is that humans, humans aren't just in the loop, but human institutions and human communities need to dwell and reflect on how we evolve and how we elevate and keep up with technology and dynamism in the technology landscape. And so that's the question that I've tried to offer in the public conversation, including of course, in and around my work here at Virginia Tech.

Travis (05:42)

Yeah, so and tell me if I'm misunderstanding, but it sounds like you're maybe reframing the question, rather than how is this technology changing? Like how do we need to also change evolve to meet this technology where it's at, or maybe how can it possibly help us evolve to a better state of ourselves?

Rishi Jaitly (05:59)

That's right. I think the conversation often regards this technology as doing something to us when I think we have an opportunity to create the future we want to see, right? Having worked in a lot of in technology most of my career, and I was an a senior advisor at OpenAI a couple of years ago as well, part of what I bring to the conversation is I've been in rooms where high stakes and low-stakes technology decisions.

Get made. And one of the things I've learned is these are human communities too. There are humans making these decisions, imagining what the week looks like or the quarter looks like. And so I think the role of human agency in thinking through all manner of things, from questions of human education to questions of governance, I think are really important to not lose sight of in a conversation where often the discourse seems to be in and around questions of what is this technology going to do to us.

Travis (06:59)

Yeah.

When you're in rooms like that, with folks having these high level discussions, what are what are the types of skill sets that really set people apart?

Rishi Jaitly (07:06)

Well, first would be, this the theme of this podcast, which is curiosity and asking the right question. Often what I have found in my career in and around Global Silicon Valley is that folks in the room who ask the most cutting, sometimes most uncomfortable, big elephant-y question in the room without necessarily presenting the neat, tidy answer grounded in a kind of certitude, are folks that break through. In addition to that, I would say, perspiration matters just as much as inspiration. Both matter.

I think finally, and this gets at our opportunity as an institute of higher education and as a nation really, folks who are grounded both in the fuzzy and the techie. To to quote a friend of mine who wrote the book The Fuzzy and the Techie almost 10 years ago, grappling with questions of soft and hard skills and whether hard power is becoming soft power and soft power is becoming hard power in in the technology industry and in society more generally, and sort of

It's been my experience i at these companies that folks that lead with imagination, introspection, with curiosity and critical thinking, with perspiration and and inspiration tend to tend to break through.

Travis (08:17)

I'm glad you mentioned the fuzzy and the techie. I remember you speaking to a group of us communicators at Virginia Tech and mentioning that then. I don't even that that's been several years ago, or a few years ago.

Rishi Jaitly (08:24)

That's right. That's right, yeah. A long time in technology and AI years.

Travis (08:30)

Exactly. But, you know, it it resonated with me because as soon as you said it, it took me right back there.

 

I've read or I've heard you talk about the humanities as a as a superpower, particularly in this age of AI And I'm always curious when we talk about words, especially when we're not higher ed, well, how we're actually defining them, what they actually mean. So we were talking about the the humanities. What are we what are we talking about there?

Rishi Jaitly (09:08)

Well, it's a great question, because the humanities are a term that, have been been around for a long time across higher education and the academy, and they mean many things to many people. For me, I try to embrace the biggest tent definition as possible and and try to tell a story of a of a capacious accommodating humanities. To me, the humanities gesture at education and importantly, experiences that cultivate Durable, enduring, timeless skills, sensibilities, and superpowers. So, yes, they flow from liberal arts disciplines like philosophy, history, literature, creative writing, and the like. But I think they also flow from the natural sciences, physics, biology, chemistry, experiences that and education that bring intrinsic value, intrinsic without Urgent regard for their extrinsic application. Right. We live in a time where, understandably, and I've been an entrepreneur and in business a lot of my career, you you press X button, you want to know what Y outcome is. I think the humanities offer us not just a respite, but a reservoir of, again, education and experiences inside and outside the academy, in higher ed and outside of higher ed, that.

Do something to us that cultivate longer form skills, sensibilities, and superpowers. And I think in my experience in business, in my experience in technology, I have noticed that the differences between a senior manager of engineering and a vice president of engineering, a senior manager of sales and a vice president of sales amounts, I have felt, amount to their the extent to which an individual's been exposed to what we're calling, what we're calling here the humanities.

Travis (10:57)

Yeah, you spoke a little just now about some of the in the institutional experiences and stuff. But I am curious when we think about it being a superpower, what should that mean for our institutions and maybe how we we think about them?

Rishi Jaitly (11:10)

Well, I think that's a great question that I don't think we ask enough, which is it's one thing to speak about individual superpowers and institutions like ours understandably speak about the individual. What what are people gonna need in the 2030s? If you're in high school now and you're finishing college in the 2030s, what are you gonna need to bring to bear? Right. And I and I can wax poetic about that, how it's important to both bring timely and timeless skills and sensibilities and have got some theories there. But equally important is the is institutional evolution and institutional elevation. And I think there part of one way to think about, at least from where I sit, is it's important for institutions to be the change and not just preach the change. Right. I think we have an opportunity in how we be, what we signal, who we hire, who leads institutions. It's not just Virginia Tech. It's institutions writ large across the public and private sector in the United States and abroad, right? Institutions have an opportunity to signal, to signal that if we're gonna suggest to young people that today, capacious, eclectic, zigzaggy, hard to put in a box experiences, education are how you grow and how you lead a full life.

 

And how you break through. Well, I think it's important for institutions too to find ways to take leaps of faith to do the same thing and in how they're organized, in how they compartmentalize, in who the protagonists, you know, who are in the front row, so to speak. And what's been really fun about my experience at Virginia Tech, and this, you know, of course, relates to the institute I've been able to found here, I think it's no no accident that someone like me has ended up here and gotten to build the way I've gotten to build here. And I think in our culture at Virginia Tech, we're grounded both in tomorrow, but also yesterday. And I think that's I think that's a wonderful signal to send to young people and to all kinds of stakeholders for our institution.

Travis (13:19)

Yeah, it's really that's a really interesting way to to think about. I like it because it's a little bit getting away from the, do as I say but not as I do mentality, but kind

Rishi Jaitly (13:27)

That's right.

Travis (13:28)

that servant leadership. you lead by doing that, you know talk here by athletes doing that a lot of time. Like this person

Rishi Jaitly (13:33)

That's right.

Travis (13:34)

leads by showing you the way. that's that's a really cool way to think about it.

Well, you mentioned some of the future and what people may or may may not need. I I'm curious, maybe just from a purely selfish perspective.

When we're looking at this age of artificial intelligence, what are the the the skill sets that may set people aside five, ten, fifteen years into the future?

Rishi Jaitly (13:57)

Yeah, another great question. And lots of people have lots of different takes on this and ways of coming out of this. I'd be curious what yours is. I grapple with what mine is. I think, to start to go bottoms up, I gestured earlier with the importance of being grounded in what's timeless, right? So bringing to bear really human sensibilities in how one communicates, intuition about oneself, intuition about the world, being broad horizontal. But you're also gonna want to be fluent and literate in in.

Today's technology and today's dynamism in today's trend lines, right? And I the metaphor I use to talk about that is full stack human. And the question I ask is, well, what does the full stack human of the future look like? And of course, both tech, you know, technologists often think in terms of stacks, front end, back end, et cetera. And folks in the humanities, well, I guess used to live and work and study in the stacks of libraries, right? So that's one way to think about

 

it.

But if the question is what's true north, what's at the level of compass, where where are we aiming here? The frame I tend to use, at least in recent weeks, and this relates to my work as co-chair of the Virginia Tech AI Futures Working Group, is elevated AI judgment. Judgment is an interesting way to think about this, right? Because when you're telling a story of someone that's got really strong AI judgment.

What are you getting at? I think what you're getting at is someone who is well-versed and fluent and literate in today's technology trends. Yes. I think you're also pointing at someone who's thoughtful about today's technology debates, right? Who's able to be an advocate and a citizen in and around all kinds of questions about data centers, around geopolitics, around Doom narratives, whatever it may be. But you're also getting at someone who's bringing bringing to bear

A kind of st a spirit of stewardship to the technology, right? who has understood and asked questions about the ethical implications of a use case here or the morality of using technology in this way, or is asking questions themselves about how do I protect my learning, right? So I think for me, elevated AI judgment is a really interesting frame to for folks to keep in mind because it it's accommodating. It says it doesn't It doesn't presuppose someone's for or against. It doesn't presuppose you gotta be at 60 miles an hour. It's just you can be at 30 miles an hour in your usage. But it's incumbent on all of us to, I think, develop a point of view that that meets our moment. And I think the history of technology teaches us that these things do become pervasive. And so a spirit of judgment, both for individuals and for institutions, is really important.

Travis (16:47)

Yeah, that's that's really interesting. And it does sound like that that staying curious may be a a key component to that.

Rishi Jaitly (16:54)

Absolutely. You're on to something with this podcast for sure.

Travis (16:57)

mean, I'm really glad we didn't name it casual conversations. we really

Rishi Jaitly (17:00)

Ha ha ha.

Travis (17:00)

dodged a bullet there. so but yeah, that sounds I I really I love that. I love curiosity. I think about those types of things a lot because I have a seven year old. he's gonna turn eight soon. And so you know, you're always thinking about like what is it?

 

What are the keys that I can give this kid that can unlock their best future? so for me it's it's things like being curious, being empathetic, being genuine. those are

Rishi Jaitly (17:21)

That's right.

Travis (17:22)

some of the things that we talk about a lot because I hope and I do think that those will hopefully transfer across any type of technology advancements that we make.

Rishi Jaitly (17:31)

Well, you're well on your way. And I think, part going back to the earlier question on institutions, right? I think we we tend to live into the stories we hear, right? So I think institutions also have an opportunity to celebrate and communicate around the thing, the thing we want folks to live into, right? So we know as parents, right, when you when you reinforce a kind of behavior, you tend to see more of it. And so I think I think we're living through a really interesting time where individuals are trying to keep up.

But w what I'm delighted to see, particularly here at Virginia Tech, is is our interest in in keeping up as well.

Travis (18:08)

Yeah, absolutely.

Well, we've talked a little bit about the the the human side of things and I'm really curious as I I look at artificial intelligence and depending on who you talk to, they're it's gonna tell you that it's gonna be able to do this or that or or whatever. So I'm curious from your perspective as somebody that studies this. What's something that you just don't think that artificial intelligence will ever be able to do as well as us humans?

Rishi Jaitly (18:31)

Well, I think a few things in response to that question. One one word that comes to mind, which gets at, I think, something pretty deep about the human experience and certainly about learning, is the word friction. One of the things I've been I've been asking myself of late in and around questions of these systems, is how does learning happen? And what are these systems doing? And what is the thing we need to preserve in the human experience? in many ways, if you think about it, learning happens through friction, right? Friction, the moments of the the tug with a new concept, the wrestle with a problem set or a paper or a new idea, right? It's in moments of friction that growth and learning happen. And we're living through a time where these systems are actually able to remove a lot of friction.

And I think there's some interesting questions around well, what do we, well, how do we measure learning, right? In a time where, these systems can fabricate, can can actually help us help us be more productive. What's the metric, right? How do we actually measure friction when historically we've been measuring production or fabrication? And so I think what it gets at kind of in the human experience is I think we humans are uniquely able to enter the leave our comfort zone.

 

To experience healthy friction. one of the most interesting things I see right now in the conversation around AI is something called the Einstein benchmark, Travis, where a lot of researchers are asking this really interesting question, which is if you gave AI all of human and I guess non human knowledge up until 1902 or 1905 or 1910, could it produce

Special and general relativity, those two theories, the way Albert Einstein did. And my observation of the discourse is most folks tend to answer today at least, no. And so what it gets at is moments of eureka, moments of epiphany, moments of of you know, the big idea of inspiration actually flow from a kind of incoherent.

Thinking, right? The big idea you had while driving, right? The big idea you have while you're sitting on an airplane, right? Or the clarity that emerges when you hear a particular song. And so these systems are really powerful of what you might call coherent thinking. I think what remains to be seen, and there's lots of questions here around discovery and drug development, et cetera. And I gave you the Einstein example, this question of incoherent thinking. And I think that flows from humans who experience and intentionally so as well, a kind of friction and which pays dividends in sometimes the most unexpected moments. And so those are a few themes for for the audience to wrestle with and a a window into some of the conversation among AI researchers.

Travis (21:38)

That's really fascinating because I really

Rishi Jaitly (21:40)

Is it? Yeah.

Travis (21:41)

I relate so much to just the process of writing stories and storytelling and these ideas that come to you. I always tell my bosses, I'm like, it's really hard to charge time to writing because writing happens when you're mowing the grass or you're in the shower and and just these words that float up in your head and then suddenly you have a story. and how do you charge time to that? so it's really interesting. But another thing that you mentioned, this idea of friction is really fascinating to me. I heard a football coach used to tell his players, sometimes you win and sometimes you learn. And

Rishi Jaitly (22:11)

Mm.

Travis (22:12)

I always tell my son, you know, one of the think great things I preach is like, you learn so much more when you lose. You learn how to build resiliency. You learn how to do things better. and that comes from me having a lot of experience in athletics losing a lot of times.

Rishi Jaitly (22:25)

Ha

 

Travis (22:25)

but I really think that's true. And most most great movies start with loss.

Rishi Jaitly (22:30)

Well, and I think it, gets at this broader question that, is very exciting and exhilarating for people like me around the highest purpose of institutions like Virginia Tech these days. And, you know, I think the phrase that I've seen some other college presidents use is this phrase human formation, right? You're if we're in the human formation business, then all of a sudden it sharpens clarity that friction and opportunities for big incoherent but

Inspired thinking is sort of part of what we want to do for future scholars and future stewards and future ambassadors of what prosim.

Travis (23:08)

Yeah.

Well, I I'm curious, I guess maybe just on again a personal level, when people are going to start to use some of these, whether they're they're chat bots or just different forms of artificial intelligence, what are some maybe introspective questions they should be asking themselves as they enter into those spaces?

Rishi Jaitly (23:26)

I think fundamentally the thing to ensure is top of mind is self-awareness and self-actualization about not just the use case you're pursuing, whether you're a scholar trying to expedite your your research process, whether you're a student trying to trying to move more quickly through lower stakes work so you can get to the high stakes work, or whether you're, you know, an administrator, improving a

Customer facing workflow. I think the thing to do is to be high altitude self-aware about the thing you're doing, and also protect and preserve for your long-term interests these deeply human sensibilities. Have a c have a critical eye on protecting your own judgment. Have a critical eye on protecting your own ability to have superpowers across the human stack. So I I guess the thing that comes to mind is.

Deep self-awareness and self-actualization. one of the one of the metaphors I tend to use personally is that of a that of a barbell when I think about use cases personally around AI, which is to say, at one end of the spectrum, I tend to use AI for deep learning and deep teaching for myself. And at other ends, it's sort of, you know, mimics a little bit of patterns on search, you know, a little bit of a little bit of thesaurus action here and there. I tend to, I tend to not

Want to run towards it for longer form, high stakes you know, tasks at my end. Obviously, we're entering an agentic era now, and these systems are enormously capable of taking a lot of sort of life ops off one's plate. And so maybe this conversation in a year will be different. But I think a spirit of self awareness around one's use cases and the ability to articulate one's. theory of using AI, right? Before just using it, develop your own theory. Because if you can offer a pointed theory just really for yourself, I think it conveys a self-awareness that'll allow you to ensure you're you're you're thoughtful in other areas, and this this is the kind of thing we do at my institute for leadership and technology, where we work with technology leaders in and around the humanities and questions of AI, is ensure folks are able to use the humanities to To actually see themselves in new and interesting ways.

Travis (25:48)

Well, how how are you most often using artificial intelligence right now? Is it is it just to help with grocery lists or are you using it in a bigger way? I mean, that's where I'm using it most of the time. That's why I mentioned that.

Rishi Jaitly (25:59)

You know, to be honest, Travis, for me it's it's mostly as I indicated earlier, kind of in this area of learning, I I'm actually I'm finding the technology to be a great teacher on all kinds of topics that, you know, in in including about AI itself, right? So I you can talk to AI and tell it to spin up a, you know, five day course for you, for you on a particular topic in computer science or physics or this Einstein benchmark, for instance. And each day on the drive to and from work, you can kind of actually talk to AI and experience a mini module of sorts. And so for me, learning, big picture. And then as I indicated earlier, a little bit of I I love words. I'm a big fan of I take words and language really seriously. So when I'm when I'm struggling with the right word in a particular moment, I find it to be a really wonderful thesaurus to to use or or to draw on to get, get the right instinct. I I still I still take notes with pen and paper in terms of preparation for things like this, including today's podcast. And I try to I try to stay old fashioned there. I think with the rise of agentic systems, I'm really interested in, you know, maybe, maybe there's something to check in on in a year. I'm really interested in how these systems might might be able to, as I indicated earlier, be always on in the background from an operating standpoint you know, pers in personal life, and and for a lot of folks in professional life. But that's that's a little bit T B D at my end. And, you know, there are a lot of models now that are particularly tuned for what you might call personal and professional productivity use cases. So we'll see how I end up there.

Travis (27:42)

And with those the ancient gentic and gentic systems that you just mentioned, does that mostly mean that they just run all the time kind of doing these tasks versus being asked to do specific tasks?

Rishi Jaitly (27:52)

That's right. There are autonomous systems that can if you if if you set broad parameters can be helpful in things like taking care taking care of bills, anticipating travel plans, and making sure your reservations are squared away. you know, and it there's a there's a trade-off, of course, in terms of access to personal data and email and da-da-da-da. but but that's what agentic systems are. Now you have models that are pre-trained, which is the really the meat and potato torso, of of the science behind AI that are pre-trained for these kinds of use cases. As we're recording Meta, the company has just released a slate of new models called Muse that are particularly focused, for instance, on what you might call personal life, professional productivity for people. And so we'll see how this agentic era unfolds. Obviously a lot of companies From law firms to hospitals to sales organizations are embracing enterprise AI to speed up corporate workflows, including here at Virginia Tech. I think the rise of agentic AI for human for humans is going to be a really interesting thing to track. And I think the right thing to do, as I shared earlier, is to be enormously self-aware about what where one, where in the woods one lands with AI. Because I think it'll help you see the rest of the rest of the forest and ensure you're you're you're thinking big picture about your growth.

Travis (29:23)

Yeah, I don't know. It sounds like will those systems get me closer to the robot made from the Jetsons? 'Cause this is where I was hoping the artificial intelligence was going. I'm not even gonna lie. I like writing stories. I just need the house cleaned.

Rishi Jaitly (29:36)

Well, it's it's interesting, right? We all we all there's I think there's a famous quote that we we thought innovation, growing up would would show up in in the physical atomic world, but it showed up in bits and bytes. I was waiting for flying cars too. but

Travis (29:50)

Yeah.

Rishi Jaitly (29:50)

you know but these I think it's possible for two things to be true at the same time to bring a spirit of awe and wonder and wow and magic to the capabilities of these systems.

Again, these are machines that have essentially been trained on prediction. What's the next word in a sentence without being told how to write a great poem? That what we've been taught is the power of data and the power of scale in in uncovering all kinds of productivity and and magic. I think it's possible to have awe for those systems and also to have awe for the human experience and what makes us different and what makes us uniquely human and run towards these systems and their their productivity potential, perhaps their discovery potential and innovation potential, while also holding on to what makes us uniquely human, which we've talked about as well.