BU Virtual Connects

BU Virtual Connects - AI, Computer Science, and the Skills that Will Matter Most

• Boston University Virtual

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 33:04

AI is changing how we work, build, and solve problems—but some of the most valuable skills remain surprisingly constant. In this conversation, we examine AI's impact on computer science and technical careers, the growing importance of collaboration and communication, and how professionals can continue learning and adapting in a rapidly evolving technological landscape.

About our Guest

John Byers is Professor of Computer Science at Boston University and Executive Director of the AI Development Accelerator. A computer scientist, educator, and industry leader, he has spent more than two decades teaching, conducting research, and applying data-driven approaches to real-world challenges. His research focuses on data analytics, computer networking, and the study of internet platforms, while his industry experience includes serving as Chief Scientist and a member of the Board of Directors at Cogo Labs, a Cambridge-based startup specializing in online advertising, algorithmic marketing, and data science analytics.

Welcome And Why AI Feels Different

SPEAKER_01

This is Wendy Colby, Vice President and Associate Provost at Boston University and the host of BU Virtual Connects. I'm pleased to welcome John Byers, Professor of Computer Science and Executive Director of the AI Development Accelerator at Boston University. John has inspired students in computer science for over 20 years, balancing classroom teaching while also conducting research and working in industry. His research interests center on data analytics, the science of computer networking, and empirical study of internet platforms such as Airbnb and Yelp. He has also served as chief scientist and a member of the board of directors of COGO Labs, a startup in Cambridge, Massachusetts, that leverages a proprietary platform for online advertising, algorithmic marketing, and data science analytics. For today's conversation, we will focus on how AI is reshaping computer science, among other fields, and what that means for professionals looking to advance in these fields. We'll also explore how work is changing, what skills are becoming most valuable, and how learners and organizations can prepare for a future defined by constant technological change. John, such a pleasure to welcome you here today.

SPEAKER_00

Thanks, Wendy. Great to be here.

SPEAKER_01

Great, well, let's go ahead and get started. We're obviously in a current moment right now, right? Where there's a lot of conversation on how AI is transforming work, but technical fields are experiencing these changes, I think, in especially meaningful ways. And I'm wondering from your perspective, what feels most significant or different about this moment in fields like computer science, business, software, engineering? Can you share any real-world examples of new ways of working for professionals in these fields? Any stories you can share?

SPEAKER_00

Sure.

From Internet Waves To AI Shock

SPEAKER_00

I think it's helpful to go back in history and think about technological change over longer time periods. So as a computer scientist, we've seen many huge advances: the microcomputer, the internet, email, all of these things came upon us and have been with us now for 30, 20 years, decades. But I think when these technologies came upon us, we as computer scientists realized the potential quite quickly. But it took a quite a long time for these technologies to permeate into society and culture. I remember being the first cohort of undergraduates who had email access. And it was just amongst us computer scientists, and we had it, and it was incredible, but nobody else saw it yet. And it took years for others to see it. This moment is different because AI has come upon us so fast and it has impacted so many people so immediately that it's very, very different and has a different feeling that's that's real and tangible.

SPEAKER_01

We're gonna get into a lot of that, I think, here today. You know, what that arc and evolution has looked like, you know, since, as you say, the internet, other big moments in time.

AI Changes How We Code

SPEAKER_01

So the piece I'd really like to dig into is how work is changing, right? And we're seeing how I think I've heard you talk about, you know, coding has almost become democratized now, and how AI is beginning to redefine how software is written and it's opening access to many more people. And so I wonder, as you think about that accessibility to a wider audience, um, what do you make of that? How does how is AI sort of changing day-to-day practices? You know, for computer scientists now coming up, right, who are just getting their degrees or they're working in industry today, how should they be thinking about preparing for careers in the age of AI?

SPEAKER_00

That's a great question. I think we should really decouple two different thrusts. One is the democratization of AI, making AI more accessible to like a really broad audience, which is real and actionable and very interesting in its own right. But maybe we can first focus on this question of what about for practitioners, people who are already coding and writing software and doing so. Yes. So I think with that, I think the main upshot is that all of a sudden AI is super powerful. It's a very, very, very good coder and it's extremely fast. However, it's not perfect in many ways. So you have to use slightly different skills. I find myself writing a lot of code with AI, both in my research and in my hobbies, which by the way is a good way to onboard AI. If you don't want to integrate it into your work, you can integrate it into your hobbies. Yeah, talk about that a little bit.

Birding App And AI As Assistant

SPEAKER_00

Well, I'm a birder. That's right. So I'm kind of obsessive about birds, as many of us are, but I'm tracking. I've got a year-long tracker of which birds I want to see, where I want to see them, how I want to see them. I'm using AI as like a recommendation engine and a log of all of my observations.

SPEAKER_01

And then maybe work. Like, where are you starting to incorporate this into your own world of work?

SPEAKER_00

In the world of work, I'm having it really do cutting-edge research. Me and a couple of other faculty members are co-advising an instance of Claude code. And that means uh we treat it like a PhD student. We ask it to write software, run experiments, analyze the experiments, give us write-ups, give us daily reports. And what's really interesting about that is it feels like the same old uh way we work with the graduate student, but we feel sometimes a little disconnected because it moves so quickly that overnight an experiment which you thought was going to take a while is on your desk and you don't have that think time to really reflect in the same way you might, right? You might want to revise what you said to a student in an email and say, uh, I think about overnight, but you don't have time to do that anymore. Yes. Yeah. So this this whole approach is is jarring.

SPEAKER_01

And is your research kind of can you talk at all about what kind of areas that you're probing in your research today?

SPEAKER_00

Is it related to computer science or this particular project happens to be understanding AI itself? So we're looking into uh the underlying models and trying to, you know, decompose them, deconstruction. Of course you are, of course you are, and drill into understand why they're doing what they're doing.

SPEAKER_01

That's great.

SPEAKER_00

So in the age of interpretable AI, it's really important to understand why AI is doing what it's doing. So we're really trying to get at what makes some of these early models that are analyzable.

SPEAKER_01

I like your example of almost like an assistant, a tutor, a PhD student. You know, I think we're seeing more of that. You've certainly talked about that in the higher ed community and with students, you know, having their study buddies or tutors or those kinds of things. That seems like a use case that is continuing to sort of accelerate.

Skills That Still Matter

SPEAKER_01

You know, what we talk about a lot too is the skills that matter most still, right? And you've talked about this. Like we're always gonna need computer scientists, right? We're always gonna need people to understand how to write code and how to how to, you know, have the decision-making authority and judgment around that validation, right? And so I'm just probing a little bit more on skills that professionals are going to need as we advance in these fields. What's going to become most important? Is it all of those things to kind of validate and learn how to work with AI? I mean, how do you see this evolving as somebody who's been in this field a long time?

SPEAKER_00

I mean, I think it is validating, but I think it goes deeper than that. I think it's really communicating with AI. I find that reading and writing skills, especially prompting skills, are really, really valuable. I'll give you an example in a second. But what I'm finding also is that uh younger professionals have the opportunity to really manage and work with a collaborator and guide that collaborator, an AI I'm talking about, in a way that's maybe very different than you had an opportunity to do before. So I'm using these soft skills to manage the workflow of a collaborator. And I'm finding that to be, you know, that's something I've cultivated as a PhD advisor and a mentor for many, many years. And I find those skills translate really well.

SPEAKER_01

Absolutely, absolutely.

Generative AI Beyond Chatting

SPEAKER_01

Um, tools, you know, everybody's talking about tools, whether it's ChatGPT or Gemini or the OpenAI suite, right? Um, are you excited about how these tools, how some of them are opening up new opportunities? Uh, you know, first just want to talk about tools and then I'll back up from that. Um, but anything that kind of excites you, do you have your go-to tools today in terms of how you're interacting with AI?

SPEAKER_00

Yeah, so I think one thing I've found over the past maybe 12 months is that we're using generative AI, even with text, but in ways that go way beyond chatting. So it's it's interesting. We have this moment this week where Google has enlarged its search bar, right? So you can type in a longer query. But that's so transactional and so limited. It seems like a very weak response to me. That's not, that's such an incremental approach to using AI.

SPEAKER_02

Yes.

SPEAKER_00

Even generative AI is so much more powerful than a larger search box.

SPEAKER_02

Yes.

SPEAKER_00

Uh, so to give some examples, I was talking to a journalist earlier in the week, and she was saying, Oh, generative AI, it's transforming the newsroom. We can, you know, download articles and summarize them in ways we never could before. And I'm thinking, yeah, that's that's great. I mean, that's that's that's fine. But let me show you some other things that AI can do for the newsroom. Let's talk about a story that you're working on. Okay, let's go back in time with AI to 2025. We'll ingest all the content from that story and then ask it to project forward, give its best guess what happened in the year since then. And of course, it can give you very, very thoughtful responses. This is something that predictive stuff that generative AI can do is very, very powerful. And not all humans are great at doing this. And certainly not thinking about a whole wide range of possibilities.

SPEAKER_01

And what did she make of this?

SPEAKER_00

She was like, oh my goodness, like lots of those things, most of those things happened. And and yes, but that's really, really interesting and a pretty great synopsis. I mean, some aren't right, so some things didn't happen. I can give the reasons why. But what I told her next was, we're not done. Now let's run up to present day and ask it to project truly into the future. And it was very, very thoughtful. I mean, it really could really assist a journalist in writing a position piece or tracking what's going forward in a way that could be very difficult to get this broader perspective immediately.

SPEAKER_01

Yeah, like amplifying kind of all the other research you might have had to do, right? To just be ready for to write that article or to position that article, right? It's amazing, right? Fascinating.

SPEAKER_00

You have this person expert. Sorry, I said person.

SPEAKER_01

You have you have this AI anthropomorphosizing right here, yeah.

SPEAKER_00

You have this expert who just got up to speed on your topic in a minute, and now is with you projecting forward things you might not have surfaced or thought of.

SPEAKER_01

Yes.

SPEAKER_00

And that is, you know, this type of empowerment is for me braggage. And to see that excitement.

SPEAKER_01

Right, to

Job Fear And Practical Onboarding

SPEAKER_01

see that excitement. The flip side of that, right? So maybe a follow-up is I think we are both seeing the fear also that AI is going to replace jobs, take away jobs. Uh, we're certainly seeing this in industry manifesting in, you know, changes in jobs. And what do you think here? Is this more of a job change, more of an amplification, or is it an elimination? Like, how should the workforce be thinking about this today? Is this a time where working professionals should be embracing these changes because the example you just shared with the journalists, because of the opportunities it can bring, as opposed to thinking my job is going away?

SPEAKER_00

I would hope so. I mean, I've tried to remain optimistic in the face of a fearful moment, right? There's a lot of fear out there. And there are a couple of responses I think one can have to this fearful moment. First one is getting educated and getting skilled with the tools. I mean, these are not hard tools for industry professionals to get on board with, right? I we I'd say any high school student can be an expert in ChatGPT in no time. Learn how to prompt, learn how to do whatever it is you want to do, be a great communicator, but you're up to speed. So there's no reason in the world why you shouldn't, as an industry professional, get up to speed and understand what the tools' capabilities are.

SPEAKER_02

Yeah.

SPEAKER_00

The other practical thing I suggest to people is try to think of the most onerous, tedious piece of your work, the thing that you would really like to get rid of. You know, as a former associate dean, I thought there are about seven different things that I would have loved to have gotten rid of in my previous job and delegate to AI. Try to create an AI agent that gets rid of one task and pick one. And maybe uh, you know, hobby, the hobbyist stuff can lead into that. You don't have to do that, doesn't have to be your first engagement with AI. Right. But if you want that to be your first engagement with AI at work, I would suggest that. Because what that shows you is that there are these tedious, alienating tasks in administration and work today that people really don't want to do. It's better for AI to do. So it saves a lot of time. Saves a lot of time. And that time you can then invest in doing something you didn't have time to do. And we all have stuff we wish we had time to do. Yes. And just imagine giving yourself a little bit of extra time and what you would do with that time.

SPEAKER_02

Yeah.

SPEAKER_00

That can give you the feeling of this, okay. Uh if you can do that, you can feel like, okay, I can work with this technology. I can do something here. And yes, maybe you can shave off those things uh from the bottom, but at the same time, replace things at the top of your stack, which are really useful and empowering and and growth.

SPEAKER_01

Right. They're you're growing, you're still learning, you're still advancing, you're still adding your own unique, authentic voice.

SPEAKER_00

Yes. And I think that's not going away. The authenticity, the secret sauce, we'll get into prompting later. But you know, being a great communicator, these things are not going away. People with these powerful soft skills, like we talked about before, authentic voice and specialized knowledge can really matter. You know, and I'll even talk about that with respect to computer science specialized knowledge.

Debugging And Better Prompts

SPEAKER_00

You know, as computer scientists, we learn all these things that AI can do well. So there's a question like, okay, well, what's the point? Well, like, why are we doing the learnings that we're doing? So one thing we teach computer scientists, and I went through this myself, is debugging code. Debugging code is difficult, right? You're a human, your code isn't working, you've got a bunch of code that you've written, it looks fine, but you've got to manually, tediously nitpick it apart, reproduce the problem, right? Sometimes the problem doesn't recur. It can be very, very time consuming. This is where you sync, you know, all nighters in the lab. It's debugging code that you just can't find the button. So you might think, well, that's a skill that's going away because, you know, hey, I can presumably debug code for you. But actually, no, it's still an extremely useful skill. And here's why I'll give you an example. I was writing one of working on one of my birding apps, and uh, I have a sync button to sync up uh some software. And one of the times I ran it, it didn't work. The sync failed, it didn't properly sync to the database that I had in the back end. And you could ask, you could prompt and say, hey, the sync button isn't working. Debug that.

SPEAKER_02

Yes.

SPEAKER_00

That would be a poor use of AI, right? That's delegating your thinking to AI. When you give a naive quote or naive prompt like that to AI, it thinks, okay, this is a naive user. What should I do? They don't know why the sync's not working. I'm gonna output the list of possible reasons why sync isn't working, walk through the sort of manual man page of like trying to debug this with them, and it's gonna be consuming a lot of tokens and spending a lot of time.

SPEAKER_02

Yep.

SPEAKER_00

But that's not the way I approached the sync problem. I said, I thought about debugging it myself without seeing the code, without thinking why this happened. Just as you would debug like a dishwasher. You know more about the problem than it's not working.

SPEAKER_01

Yeah.

SPEAKER_00

So I said, I think I know what happened. I kind of double-clicked the sync button, and that's an unusual use case because the second sync didn't have anything to synchronize. And that may have caused an edge condition or some kind of unusual failure that we didn't consider.

SPEAKER_02

Yes.

SPEAKER_00

And start there. And and and also for future reference, I also said, and no, and next time, if that is a problem, make sure sync is itempotent, which is a computer science term that it understands for like it works if you double-click it or triple-click it and does the same thing.

SPEAKER_01

Interesting. You know, John, many of our listeners to the BU Virtual Uh Connect series, they're working professionals, their early career, their mid-career, right? And they're at a point in time, perhaps, where they're considering, do I move into a field like uh computer science? Do I advance in the field of computer science? So much change has happened in this field. So, again, you've been in this field a long time. What would be your words of encouragement uh or advice to these working professionals today? Is computer science still a field that is, you know, sort of long-lasting and sustaining? And can you talk a little bit about why the world continues to need computer scientists in this era of AI?

SPEAKER_00

Sure. Yeah, I mean, I think these complex pieces of software are, you know, very difficult for an AI to wrap its mind around, still, frankly. And I think um, you know, computer science professionals have this, you know, strong background that they should feel comfortable retaining and continuing to hone, because letting it atrophy would be a real waste, I think, personally. But it may be very different from what you've done in the past. So I personally don't write that much code anymore, but I'm still working with AI to do code reviews or thinking through the code that it wrote. Now, AI today is writing in languages that I don't understand. So you might say, well, how do you do a code review in a language you don't understand? Okay, you're not doing line-by-line code review, but you're doing things that are more modular and more conceptual.

SPEAKER_01

Yes.

SPEAKER_00

Like, how do we organize this data in the database? Is that really the right way to do a key value store? Like, why is it so slow? Like, what's going on there? And being able to think about these issues that we train computer scientists with about correctness, efficiency, design trade-offs, you know, making sure everybody's on the same page so that module A can speak to module B and we've got the right API. All of these things require interaction, whiteboards, uh, you know, engagement in the lab, human interface, human interface, collaborating with these human mixed teams of humans and AI. I mean, I think uh, you know, if you can be a great manager, that is never gonna go away as a skill.

unknown

Yeah.

SPEAKER_00

It is concerning that Facebook is saying, well, we're gonna reduce the number of, or we're sorry, we're gonna increase the number of people that report to a single manager, but it also showcases how important those managers are going to be going forward.

SPEAKER_01

Yeah, trying to really manage across a much larger, larger group of individuals who also have now amplified their own work skills, right, with AI.

SPEAKER_00

That's right.

SPEAKER_01

It's fascinating.

SPEAKER_00

So one thing I maybe wouldn't advise is to go really, really deep and become a super expert in COBOL or something, some very specific siloed technology, because that may not be as valuable going forward. Having this broader, broader uh perspective and working with, you know, work work in a language you don't know. Have AI work, feel, feel what that's like. Yeah. Really interesting and helpful to know about.

SPEAKER_01

Interesting.

Higher Education Scrambles To Adapt

SPEAKER_01

Let me pivot now a little bit to academia and industry. You've spent considerable time working both across uh higher education and industry. You also now lead our AI development accelerator here at Boston University. That gives you a really a lens into the evolution that higher education leaders are now also going through, right? As they learn how to embrace AI. And so I wonder at first you can talk a little bit about how you see higher education responding to this moment.

SPEAKER_00

Yeah, I think it's it's really fascinating because usually it's in the course of history, it's been the academics who have been a little bit ahead. You know, their research has pioneered things, so they're seeing over the horizon and they can see things coming. So they knew, for example, that these cutting-edge technologies were surfacing. But with this one, it's different because it came upon us so fast that we really didn't have time to react. So we're learning right alongside our students. And we see, you know, when we train the students in AI, generative AI, and give them, you know, techniques on prompting, we realize, oh boy, the faculty need these skills just as much, if not more, than the students. And so do the staff.

SPEAKER_02

Yes.

SPEAKER_00

So it's really a new experience where we're upskilling our entire university at the same time. And we have no processes for AI. Like, where should generative AI be used? Can it be used to summarize documents? What about confidential data? Can we really upload that stuff into AI? Like everybody has questions, legal questions, copyright issues. There's just a whole panoply of things around AI that are complicated, all surfacing simultaneously.

SPEAKER_01

And how does it all get incorporated in teaching and learning?

SPEAKER_00

How does it get incorporated in teaching and learning, right? All the faculty members, all of a sudden their pedagogy has to change, right? A Gen AI can solve every computer science problem we pose in our undergraduate curriculum perfectly.

SPEAKER_02

Yep.

SPEAKER_00

So that's a problem like that faculty members have to grapple with pedagogically.

SPEAKER_01

But one of the interesting things you're doing in the AI development accelerator, right, is also bringing industry into our world, right, through advisories and other. And I'm just wondering how you're seeing that partnership perhaps evolving over time with companies, industry, advisory groups right now coming together with academia to create something really special going forward.

Industry Partnerships And The AI Integrator

SPEAKER_00

Yeah, I think there's a huge opportunity. And BU is certainly realizing it. I think other universities are realizing it as well. You know, being here right in Boston and seeing industry happening firsthand right at our doorstep is a real advantage that BU has. But we're definitely seeing that both students and industry partners want to engage with one another early in their career. Right now, we're finding that students, as I said, like they can be up to speed with Gen AI at with their high school diploma. So freshmen can do something with AI, integrate AI, or talk about an industry problem in their freshman year or their freshman after summer after freshman year. So there's both the possibility of deeper engagement, broad engagement, and engagement at scale across so many different disciplines. So one thing we're thinking about is a concept of an AI integrator, a person who's trained in AI and in a discipline who can take AI from the classroom or their experiences here at BU and help an industry partner with onboarding AI uh AI know how into their org and maybe transforming one of their AI programs. processes. But we're also finding that to do this well and convincingly, BU has to be a role model. BU itself has to onboard these processes and do what we teach.

SPEAKER_01

Yeah, absolutely.

SPEAKER_00

And make sure that as part of what we're doing, we're also becoming AI native and engaging in this institutional transformation around AI.

SPEAKER_02

Yeah.

SPEAKER_00

Which is another one of our president's signature initiatives, both institutional transformation and working with this AI Development Accelerator.

SPEAKER_01

Absolutely. You know, that made me think, John, of some of the work we're doing through one of the units I manage here, the Institute for Excellence in Teaching and Learning, that partners closely with the AI Development Accelerator. And, you know, the awards that we've given out recently through our Shipley Academic Innovation Fund. And that's been a lot of fun to watch, right? The faculty are really experimenting and really taking a leadership role and really championing some of the changes around this. And I wonder if anything stands out to you in terms of what you've seen in our faculty community in terms of that level of engagement and experimentation and perhaps also the importance of encouraging experimentation across the community in a time when we're still evolving in this new era.

Faculty Experiments With Learning Tools

SPEAKER_00

Right. That we're very innovation forward at AIDA because we think that encouraging experimentation is really the way to go. There's no, we don't have a silver bullet. We don't have the answers yet. We have to develop them. And we hope to develop them here at BU. I mean we saw I saw with you some of the lightning talks from the Shipley Awards and they were fascinating, right? One of our engineering professors is doing randomized controlled trials to see when we give some exposure, deep exposure to generative AI, how does it transform the learning of the students? Yes. And she interestingly the students weren't all I would think oh all the students must love this. And she said no actually some students are very guarded and hesitant about AI and really don't think this is a great value add. But I'm going to be we don't have the results for the RCTs yet. But we're when we see that maybe the students will actually be more convinced that oh yes actually deepening your knowledge of generative AI is actually a beneficial. So that was one. And the other one we saw which I've always been fascinated by is a professor in our data sciences unit who is doing something called paper buddy, which is thinking about right reading a technical paper. That used to be like a three or four hour endeavor. It was very difficult to do and we always ask our students to do that. It's and but it's a big lift heavy lift and students can take a shortcut and just have AI summarize it. But he is saying you know when AI summarizes it for you you can't actually contribute to an in-class discussion that goes deep on the paper because you don't know it. All you know is the AI summary. So that's not the right approach, right? No shortcuts. He has an assistive reader which facilitates reading of a paper where you can focus on the parts of the paper that you don't understand and need to know better. So you can shorten the time to onboard the paper using AI along the way, but using it not as a replacement but using it as a collaborator.

SPEAKER_02

Yes.

SPEAKER_00

And highlighting the text, getting the information you need from the paper.

SPEAKER_01

Yes.

SPEAKER_00

And I thought that's I mean I think this is a really great idea.

SPEAKER_01

What I love is you know, just as much as change is going on in industry in higher education, we're also rethinking the way in which we create instruction, the way in which we create assessment and that's a really great example of how now it's not just ask a student to go to chat GPT and ask a question. It's now here's a way in which we're kind of incorporating the instruction right inside of this assistive tool. Yes, exactly. Very interesting. So now let's do just a little visioning forward, right? Preparing the next generation. You talked about this in the journalism talk, right? Let's think about where the future is so as you know now think about a working professional who still may have 20 odd years or more in their career, you know, what makes you hopeful about what the future holds for a computer scientist and where do you think this all goes in the next few years?

SPEAKER_00

Yeah it's hard to prognosticate. So a crystal ball goes there. We should ask generative AI. Maybe we should type that in right now because it's going to have tons of ideas and we can winnow through them and pick the ones we like and explain to the AI why the ones that thinks are good are not going to happen. But I I find it to be an exciting time. I mean I think the folks that I work with who are by their nature entrepreneurial and innovators themselves, whether in research or in industry most of them are skew optimistic. I felt there was a little more pessimism last year when I think people felt like general artificial intelligence might be on the door and that's very, very scary for many. But now it feels like okay no maybe it's just very very advanced gen AI, regular old generative AI. And that we can get our minds around as an assistant and a helpful and a collaborator, an expert that's empowered to do things and on our level but not super intelligent.

Agentic AI And Your Comfort Line

SPEAKER_01

Yes.

SPEAKER_00

I think where we're well I think the one of the most interesting things is this concept of agency and figuring out where we're going to go, what we feel comfortable with letting AI do. So I personally, I mean I think this is something for everyone to think about like where does your line where do you draw the line? You know when you use something like OpenClaw very very powerful tool open claw wants you to delegate email access to it. It wants you to be able to it wants to write emails on your behalf. It wants to be able to delete emails it wants to send texts on your behalf. It wants to do an awful lot. I personally don't feel comfortable with AI being that agentic where it is where I've given it basically a power of attorney to do whatever. So but some people do. And I think this is really um a a place where people should think long term is where is the future of agentic AI going? I mean we've gone through so many cycles just in the last year from chatbots to Python sandboxes to agentic AI to what's next. I mean these the somebody I heard somebody say you know the half life of technologies now is about six months. So by the time you've learned something in six months it's halfway to obsolete or so I think this is like a very frighteningly fast moment. And why I'm hesitant to prognosticate, I think the best one can do is just to stay really, really current.

SPEAKER_01

Well that ties in so nicely John too when I I think about you know you talked about the innovation forward focus of the AI development accelerator and Boston University's really strategic approach to AI in general. You know, I think about as a university Boston University and certainly other universities the focus too is on upskilling and reskilling and how do we meet learners in this moment so that we can evolve. I mean this in effect for the slight plug for BU virtual is really the focus here on how we can help learners in their world of work gain these skills, continue to evolve because this world is going to be ever changing, right? Where this is a moment now and it's only going to evolve, right? So as we start to come to a close, I'm wondering if you have any final words of wisdom or any fun fact about you know how you've used AI and you've talked a little bit already about how you used it in your personal and professional life, but any kind of final words of wisdom?

Teaching After The Lecture Era

SPEAKER_00

I was actually going to riff on uh the way uh pedagogy and instruction is unfolding right we've seen the the demise of a frontal lecture to a giant hall of students has been predicted for a long time but I think the death knell is really really here in the age of you know podcasts and people getting content from YouTube and influencers there's a new way of transmitting information that's a younger fresher faster way that really makes it feel like you know going to a lecture and sitting in a lecture hall and being talked at is just not it was never for most people but it's really not going to work going forward. And I think the same is true for digital learning like if we just have you know corporate trainings where folks come in, sit in a room listen to somebody talk about generative AI, watch a video videos like that's not what we're talking about. So I think we're really talking about deepening the engagement somehow and I think we can do this through both synchronous and asynchronous learning. We have lots of ways to engage the students with peer collaborations, peer grading, working on collaborative assignments online, taking things from their work and doing work integrated learning where we really try to understand okay what's going on in your work like bring a problem from your work into the classroom and let's really bat that about these are the types of experiences that students really want to have. I was fortunate to teach a small elective class last semester of 20, 25 students and I said, oh we're going to be using AI in this class. You're going to write way more code with AI than you could have on your own. So your projects are going to have a lot of AI. But the way we're going to be assessing you is largely through oral presentations and teamwork and collaborative exercises we're going to do here in class in person. And students are like okay yeah like we that sounds good. We're game we're game but this is a very big challenge for universities. Universities have frankly gotten away with doing things at scale because it's easy to pack butts in seats and put hundreds of people in the classroom it's hard to have tenure track faculty, tenure stream faculty teaching 20 students in this bespoke way at a time but we kind of have to figure out how to do that. And digital helps because it removes the barrier of having to have everybody in person. Absolutely and the challenge is to can we keep the engagement super high as if everyone were really there anyway.

SPEAKER_01

And I'm optimistic I think we can come close yeah and that's I think our goal with these programs I agree right thank you for all of those reflections. You know I think we are at such an interesting point in time of transformation a time where we need leadership we need the kind of work you're doing at the AI Development Accelerator. We need the strategic focus right that universities like Boston University are taking now. And that is helping us evolve and modernize and advance and reach more students and drive student success all of those things that really become important to this broader ecosystem that includes both higher education industry working professionals and beyond. So I want to thank you uh for your time today. It's been really a fascinating conversation and look forward to all the continued work you're going to do, John, in the years to come. So thank you.

SPEAKER_00

You're very welcome and there's plenty of work on our plates don't worry there's so much to do ahead.

SPEAKER_01

Absolutely I don't think anyone's jobs are going away anytime soon. Not in this space no thank you again podcast.

Closing Thanks And How To Subscribe

SPEAKER_01

Special thanks to my colleagues at BU Virtual and to our media team who produces this podcast under the leadership of our studio director, George Faco. To keep up with our BU Virtual Connect series, be sure to subscribe wherever you listen to your favorite podcasts. You can also learn more about our portfolio of online programs at BU Virtual by visiting BU.edu forward slash virtual