Skip navigation
Podcast

The Five Most Promising Uses for AI within Academia

Episode 178

January 9, 2024 35 minutes

Summary

EAB’s Michael Fischer and Afia Tasneem discuss ways that institutions are moving beyond their initial resistance to artificial intelligence and are now incorporating AI technologies within academia. They highlight some of the most promising opportunities for leveraging AI to reshape curricula, improve student support, and gain efficiencies across hundreds of administrative functions. Finally, the two offer tips to higher ed leaders on how to focus their energies and governance processes to gain the greatest institutional benefit from AI.

Transcript

0:00:10.8 S1: Hello and welcome to Office Hours with EAB. In past episodes, we’ve explored some of the dangers and risks associated with students using generative AI to take shortcuts in their work. Today we wanna talk about how institutions are adapting and incorporating AI into academia for the benefit of students and the institution. We’ve captured an interesting fast-paced discussion, so give these folks a listen and enjoy.

0:00:43.5 Michael Fischer: Hello, and welcome to EAB’s Office Hours. My name is Michael Fischer and I’m a Senior Research Director with EAB. Almost exactly, gosh, one year ago today, I sat down with my colleague, Ron Yanosky, and we discussed over an episode of this podcast the budding technology known as ChatGPT, and the possibility that students would never write another essay or fill out another exam or maybe even stop going to college because of budding generative AI technology. Now, it’s been a year, there’s a lot we don’t know and a lot we’re still debating, but the conversation amongst Higher Education leaders has shifted to determining how we might use generative AI and other AI tools to help reshape curricula, improve student support and success and gain better efficiencies across vast portions of our functional responsibilities on college and university campuses. So today, we’re gonna dive back into the generative AI conversation and talk about some of the most promising opportunities that we at EAB are seeing when it comes to this new technology, and the insights that we’ve gleaned from having hundreds of conversations with university and industry leaders over the past year. And with me today to unpack this topic is one of EAB’s foremost AI experts, Afia Tasneem. Afia, welcome to Office Hours.

0:02:14.1 Afia Tasneem: Great to be here, Michael, thank you.

0:02:16.4 MF: It’s great to have you with us as well. Afia, tell us a little bit about yourself and how you ended up spending hours, days, I think maybe months at this point of your life researching generative AI technologies.

0:02:30.7 AT: Sure Michael. I’m a Senior Director and Head of our IT Strategy Advisory Services here at EAB, so it’s not all me. My team and I, we spent a ton of time on AI this year because that’s what our partners have been asking for. We generally serve the Higher Ed executives, CBOs, CIOs, provosts, presidents, and all of them have been asking many different questions when it comes to AI over the last few months, sometimes I would say it’s been daily or even hourly questions. One conversation especially comes to mind, a president told me that he’s getting tremendous pressure from his students, faculty, staff, boards and legislatures to get smart on AI. So we actually had our work cut out for us, we needed to understand what the opportunities are, what the risks are so we can help our partners figure out the implications of AI in the context that matters most to them, which is Higher Ed, and curve out next steps.

0:03:34.6 MF: I think one of the first things that I found when I tried to get smart on generative AI and AI technologies is the realization that AI is not actually that new of a technology, AI has been around for decades. If you’ve ever used Google Maps or searched for something on Google or Bing or engaged with any application on a smartphone or a computer, there’s probably AI functionality that’s at work behind the scenes, but generative AI has swept Higher Education and society writ large by storm over these last few months. What would you say is the big difference between this new iteration of the technology versus the past experiences we’ve had with AI tools?

0:04:16.7 AT: You’re absolutely right, Michael. AI is nothing new. In fact, it’s been around since the 1950s, and those of us in the technology field, we have seen AI go through many springs, many winters over the last seven years. But what we are noticing now is really an inflection point, and there are two main reasons why it’s been such a big deal. First, is the technology itself has evolved substantially, AI has moved from primarily being discriminative, which is focused on categorizing and analyzing data to being generative in that it can generate human-like content across domains.

0:04:57.4 AT: We talk a lot about texts, but it’s also images, software, a scientist I was listening to in another podcast was talking about genomes, so there’s been a lot of new possibilities that has been unlocked through the launch of generative AI tools and we are seeing this technology more as a general purpose technology that can bring about a lot of changes across multiple industries. The second point I would note, and I think this is even more important from a general perspective, is accessibility, while advanced functions like finance and IT have been using AI for a while now, ChatGPT really sparked the transition to AI for everyone.

0:05:47.2 AT: Anyone with a computer or a smartphone and internet access can access this technology. So with access came usage, so people started really using it for their benefits. And the other thing is you don’t have to be a tech whiz to use these products, no coding required, no user manuals, because tools like ChatGPT can process natural language, you can just talk to them and they talk back. These two factors, the revolution in the technology itself and the accessibility factors have made this quite a big deal.

0:06:28.8 MF: And I think it’s true that it’s near universal in its uptake across campuses, there’s been a couple of surveys that have found that 90% plus of students on an average campus have dabbled with generative AI, and one Provost told us that that only suggests that 10% or so of students aren’t willing to answer honestly on anonymous surveys because everybody is taking this up, but it’s interesting that that’s the case because historically, Higher Ed doesn’t have a reputation of being much of a technological innovator or one that readily embraces a lot of these new innovations to our curricula and to our processes, so how are you seeing Higher Education leaders respond and pay attention to this technology this time around?

0:07:17.9 AT: That’s a great point, Michael. An interesting stat there on the 90% of students using it, 10% may be lying because they fear stigma. So coming back to higher education being laggards when it comes to new technology, I agree, that’s true. But I also think people in general are often resistant to new technology, Higher Ed maybe more so. This mold of technology adoption, taking years and years to adopt new things, that got broken with ChatGPT. If you think about the technologies that are mainstream today, they took ages for users to adopt, folks are, as I mentioned, hesitant to take on new technology. For example, it took 75 years for the telephone to get to 100 million users, eight years for the internet.

0:08:12.3 AT: Facebook, that started this whole social media revolution, five years. For ChatGPT it was just two months, two months to get to 100 million users and become mainstream. And many of those users right from the start were university students, so I guess what’s different about this time is that the technology didn’t trickle down to our IT offices and then slowly to the rest of campus, students started using it first, so the rest of campus, faculty staff, cabinet leaders, all had to confront what this new tech meant for them a lot sooner than usual.

0:08:47.7 MF: Yeah, and I think that that lead by students really dominated the conversation early on. I remember a year ago when Ron and I were talking about ChatGPT and other generative AI tools, most of the discourse, most of the questions that I was getting when I talked to university leaders was around academic integrity, plagiarism, how would we have to change the way that we did exams, how would we have to look at essays, there’s a general risk aversion or a pessimistic mindset, is this something we would have to put guard rails on, something to avoid, but you’ve been hosting round tables, having research interviews with hundreds of leaders recently, do you feel like the discourse and the sentiments have changed across university leadership?

0:09:32.3 AT: Absolutely, I think there’s been a remarkable shift in mindset of university leaders and faculty and staff. And if I reflect back on what you just said, which is folks were really concerned about academic integrity, that makes sense because students were using it for their assignments and people didn’t know what to do with that, so they turned to AI detection software to prevent cheating. Over time, though, we realized that those plagiarism detection tools are actually very ineffective, and if you think about the future of work and how students need to know AI to be effective in their workplaces, those tools are outdated as well. And I’m so glad to see that university leaders have recognized that and they have since shifted their focus to how to accommodate or even encourage use of AI in classrooms so that students can learn to use AI in a responsible way.

0:10:36.3 AT: You’re also right about this risk aversion mindset that was there all across Higher Ed. Most people a few months ago had a very negative view of AI. They either believed that AI is all hype and it will go away or it’s too risky for Higher Ed to adopt. Now, I think there’s a growing recognition that AI is here to stay. We like it or not, it’s here to stay, Genie is out of the bottle, so how do we use AI to benefit our students, faculty and staff? The risks are there, don’t get me wrong, the most common ones being security, privacy, hallucination and bias, I could rattle them off because that’s how common they are. Everyone is talking about them, but leaders are now taking a more balanced approach, they’re thinking about the risk-benefit ratio, they’re thinking about how to manage these risks better while taking advantage of the enormous benefits this technology can provide.

0:11:35.4 MF: Well, let’s dive into it. I’m sure you stumbled across some really interesting ideas or promising opportunities that universities are pursuing with their generative AI experiments, what’s been some of your favorites that you’ve seen?

0:11:50.9 AT: Sure, I love talking about these, what are the biggest opportunities? Where are we going to see the most impact? But before I get into the opportunities themselves, if you don’t mind, I want to set the stage for that conversation, because it’s been helpful and I talk to university leaders to really think through what AI is doing for us. It’s not solving brand new challenges no one has ever heard of before, but it can accelerate the innovation curve and solve some of our long-standing problems in Higher Ed. In that way, it’s similar to the internet. Of course, sitting here right now, it seems hard to think of a time when we were ever skeptical of the internet, but we were and like the internet, AI is poised to have widespread impact in Higher Ed across all areas of campus. I’m only going to talk about a few of my favorites, five stand out, because I think those hold the most promise, but really wanted to make the note that these are not going to be the only ones.

0:12:52.1 MF: And with that disclaimer and legal caveat out of the way, Afia, what are the five that most stand out to you?

0:13:00.4 AT: Okay, the first is, drum roll, how do we prepare students for the future of work? As AI disrupts fields as diverse as film-making, legal services, graphic design, software engineering, how should we prepare students for changing workforce needs? The second one concerns student success, how do we leverage AI to provide real-time, personalized support to students across their life cycle from enrollment to graduation and beyond? Third is all about productivity. A partner recently told me AI makes a 1x person 10x and a 10x person 100x. While we may not be looking at those exact same multipliers, I concur that AI serves as a powerful productivity enhancer and can significantly shift or boost both faculty and staff efficiency and ultimately help institutions with cost optimization.

0:13:58.2 AT: Fourth is maximizing enrollment and advancement yield with AI, how do we use AI to generate hyper-personalized content at scale that can boost universities’ recruitment and fundraising efforts. And last but not the least is in the research space. The first four I talked about is about AI doing things that humans can also do. In the research space, we’re actually seeing ways to boost innovation with AI to expand the frontiers of knowledge, to think about… To expand the boundaries, really, to foster new discoveries, find new solutions in ways that we haven’t seen before. So those are the top five that I have seen a lot of excitement around on college campuses.

0:14:46.8 MF: Those are some meaty topics and listeners, you don’t need to check your podcast application of choice, I promise this is not a three-hour episode that you need to run at speed x100 to get through in your regular allotted time. We probably won’t be able to get into all five of these at the level that they probably deserve, but more information on eab.com if you’re curious about the ones that Afia and I don’t have a chance to unpack over the course of the next little bit of time. But let’s dive into this first one here, and it’s perhaps the most obvious one, work is going to change, the way that people do things in the economy but also the way that we prepare students to be successful when they exit our hallowed halls of colleges and universities. So where are people already making those experiments to try to adjust the types of skills that we develop in students and prepare them for the experiences that companies and governments may want for them when they graduate from our universities?

0:15:48.5 AT: Yeah, that’s a big one, because work is changing across industries, as I talked about. If you think about the examples I gave earlier, I picked them at random. So I hope it’s the same, but think about the TV and movie industry, writers and actors went on strike recently because of concerns that AI may replace their jobs. A year ago, we wouldn’t have thought of this kind of disruption because those jobs seemed secure, they seemed so human in our opinion, that they could not be disrupted by AI. We’re similarly seeing huge disruption in legal services, recent studies showing that nearly half of legal work could be automated. AI can sift through vast volumes of legal documents, assist in due diligence, even draft contracts. Something very different here, graphic design used to be a very labor-intensive process requiring deep understanding of design principles.

0:16:49.4 AT: With AI power tools now, even complex designs can be generated in an instant based on simple text descriptions. I can go on and on, software engineering is another one where a McKinsey study found coding can be done twice as fast using generative AI, so lots and lots of examples, but the key here is our students need to learn how to use AI tools to be more efficient in their work and also to innovate with AI no matter what field they’re planning to be in. The most forward-looking approach we have seen so far comes from University of Florida, they actually infused AI across the curriculum into every single discipline, so every student, regardless of their major, has the opportunity to develop foundational knowledge about AI.

0:17:37.2 AT: They have courses like AI in Agriculture, Business Applications of AI, AI in Healthcare, and their students and faculty are collaborating to come up with diverse research applications, things like how to detect strawberry bruises before they reach market, how do you improve equity in workplaces with AI, all the way to how do you use AI to predict and prevent surgical complications? This is the fundamental shift that needs to happen in Higher Ed. As workforce expectations change, AI may become as fundamental to learning as math and English, as essential in the classroom as textbooks and chalk today. How do we move towards that future?

0:18:17.7 MF: I think it’s important to double-click on the point that when people think about AI skills, they’re probably thinking of training AI models or developing AI tools, but that’s… Now with the vast majority of white-collar workers are gonna do with AI in the economy, they are going to be doing the same things that we might use like with Microsoft Word or Google Docs, using a tool for their own features in a relatively foundational or intermediate way. I think we saw one estimate, Afia, that maybe 90% of the professional population will need to use that level of generative AI tools while 10% will be the ones training the large language models or developing the AI software itself. So even if a student isn’t interested in going deep in that AI field, they’re going to need to have that exposure and experience and are gonna be expecting their universities to give that to them.

0:19:12.0 AT: Yes, absolutely. I agree. I think it’s going to be less than 1% that really trains large language models and then 90% will need those basic AI skills like using ChatGPT, using the other AI tools that are publicly available, there’s the 10% or 9% to 10% in there, they are actually going to be developing AI applications using the large language models that are already out there. So think of using OpenAI API or open source options like Meta’s Llama 2, and that’s an interesting space because you’re gonna see a lot of vendors develop applications, schools may want to develop applications as well, and that’s standing on the shoulder of giants to solve your problems with specific applications. That’s, I think, going to be a very interesting space to watch out for.

0:20:16.3 MF: Turning from students’ experience in the classroom to students’ experience around the classroom, the second opportunity you suggested was around student success and the student journey on campus. Universities have spent so much time, so much energy and resources trying to support students and intercede for them in order to ensure that they graduate, support their mental health and wellbeing. Where do you think AI provides the greatest opportunities for helping to close the equity gap and drive better outcomes for students on our campuses?

0:20:51.5 AT: This is interesting because the concept of providing seamless, personalized 24/7 experience is not new either. We have been talking about it for a while and that’s where we have said that students need personalized support, they need it round the clock and we struggled with it because how can you scale that effort? We tried chat bots from the previous generation. They were not useful or user friendly. They gave us scanned answers and ran out of steam within a few minutes. But now with generative AI, I think we finally may have the technology to power conversational AI bots that can interact with students in a human-like manner, helping them solve their problems in real time, starting from enrollment, academics, career services, all the way to alumni relations. That’s where I think we can move the needle a little bit, providing personalized support. One example that comes to mind is AI powered tutors or teaching assistants that can help students learn concepts, write papers. I know that’s controversial, but it’s…

0:22:04.4 MF: It’s happening, right?

0:22:04.5 AT: It’s kicking off the learning, it’s happening, that’s the first thing to acknowledge and it’s really about learning. Where do you need to learn foundational skills versus where do you need to get things done quicker, faster, more efficiently by collaborating with AI? And that’s where we need to find the balance and help students use AI-powered tutors to develop their own skillset.

0:22:33.3 MF: That balance is…

0:22:33.6 AT: And Khan Academy and… Sorry, go ahead.

0:22:36.5 MF: I was gonna say, and that balance is important on the backend as well because students want both and, it’s not that they want to replace their in-person interactions with counselors, with advisors, with coaches with AI entirely, it’s a complementary effort because they want to engage with the tutors in after hours, or they wanna be able to speak their native language to someone in an environment where there aren’t a lot of people who know linguistically how to do that, or they need a more intimate level of help on one thing that is easier for them to access through the AI than other aspects of their life that they may want to consider with their in-person or personal tutors.

0:23:19.0 AT: Absolutely. I think the key there is, can we today provide personalized support to students the way they need it? And the answer is no, we don’t have the human capital to do that. If not, what is the second best option there? And do we have the technological capability to deliver? And the answer slowly is now becoming yes, we do. So why wouldn’t we use it to get students to perform better?

0:23:46.0 MF: Is there anyone you feel has been doing a really good job of driving that student success vision of AI forward?

0:23:54.0 AT: It’s very early in the space, but I know many are experimenting. Khan Academy released their AI tutor recently, and it looks promising because it doesn’t give students answers to questions, it really is programmed to help students learn. Georgia Tech is another school that’s been experimenting with this for a while. They have AI powered teaching assistants for some of their courses, and they have been quite happy with it. So it’s really… The space is still nascent, but we are seeing developments and interesting experiments and feel quite optimistic that these would help move the dial on student success in the future.

0:24:38.4 MF: Let’s turn to that third transformational opportunity around productivity and efficiency. Perhaps the only thing that’s been getting more headlines in the Chronicle or Inside Higher Ed these days than AI stories is cost containment and concerns around budgetary pressures and the potential that that will lead for staffing decisions on campus. And some people have talked about AI tools as being the silver bullet, the solution to all of our inefficient woes on campus. Now, as you can tell, I’m a little skeptical of this conversation, but where might AI supplement or complement the kind of work that we do on campus? And is there a role for AI to replace some positions or responsibilities that traditionally have fallen to human workers?

0:25:23.5 AT: You’re absolutely right. This is the area that’s getting the most attention. And recent studies from Goldman Sachs and other places have shown that two thirds of work may be augmented with AI. Now, what does it mean for campuses? How should leaders think about this stat, two thirds of work may get augmented? I think for the most part, AI is still going to be a co-pilot rather than a rival to human employees. But there are going to be certain areas where you see AI doing a job better than humans or the same as humans. That’s where we are going to see universities focus on cost containment more. And I think this is a tricky space to… It’s a complicated area. And I think the thing that I would want folks to remember is AI is a tool, a very capable one, but still a tool. What we do with it is entirely up to us. Some institutions may use AI to help staff with their tasks and reduce long standing staff burnout and retention challenges.

0:26:44.3 AT: Other schools, especially ones who are struggling with financial sustainability, you talked about the cost pressures, they’re very, very real, they may use AI to replace staff because for them, AI can be a lifeline that helps them reduce administrative costs and survive. The choice is entirely up to university leaders and they will make decisions based on their circumstances. One thing regardless of where you are in terms of circumstances that I would advocate for is upskilling staff who are vulnerable. Economist Richard Baldwin recently said, “AI will not replace your job, but a person who know AI will.” And this is a very important point because we need to make sure folks who are historically disadvantaged, women, people of color, they get the appropriate reskilling and upskilling so they’re not left behind.

0:27:44.8 MF: I know we’re coming up on the end of our time, Afia, that we have, but what about the last two transformational opportunities around personalized messaging and research? Are there any gleaning thoughts or important things that people should keep an eye on?

0:28:01.3 AT: Yeah, I think, let me just take on the personalized outreach part for advancement and enrollment because I think that will have more large scale impact across campus because enrollment is a big conversation in most college campuses. And what I wanna note here is personalization is not new in marketing. For the longest time, we have been trying to speak directly to our prospective students or donors in ways that resonate with their unique backgrounds, interests and aspirations. But the extent of personalization we could do was always limited by what humans could manually do. There are only so many people you can tailor messages to when your staff is 10 people or 20 people.

0:28:50.5 AT: With the advent of generative AI, we see a future where it may be possible to take personalization to an entirely new level. Instead of segment or cohort level customization, imagine being able to produce individualized custom content from different centers in distinct styles or tones that can automatically be distributed across various channels, emails, texts, what have you. And that’s where I think some of the large scale impact is going to come from. And schools are innovating in this space or experimenting with them, at least, to reach these new levels of personalization at scale.

0:29:30.4 MF: So, of course, the flip side of that is if everybody’s doing it, our recipients of that messaging are gonna get smart on it in the same way that dedicated mailers and emails became quite blasé. But that balance between personal connection and providing people with what they want and people’s exposure to it, there’s probably some happy medium that we’ll reach to at the end.

0:29:52.3 AT: Yeah, I think what’s important to remember is prospective students only care that you understand them, that you’re speaking to their interests. They don’t really care if that’s coming from AI or humans for the most part.

0:30:07.9 MF: At least now.

0:30:08.3 AT: Because email itself is so impersonal, you’re not really talking to someone in email. So how do we make that communication more effective is the way to think about it.

0:30:21.1 MF: Afia, I know that normally when you’re presenting to our partners on this topic, you usually go for five hours straight, I think is what I’ve heard and there’s no way we can keep you going for that long here. So why don’t I ask one final question, which is, what should people do knowing that all this change is happening, that there’s a lot of disruption and transformation, both opportunities and pitfalls that are down the road with generative AI, what do you recommend leaders and end users start to do in order to prepare for our AI future?

0:30:53.3 AT: Great question, Michael. I think everyone, all end users should get familiar with multiple AI tools. Everyone uses ChatGPT, that’s not the only one out there. So try to figure out what works best for your purposes. Use ChatGPT, Perplexity, Bard, Stable Diffusion, there’s tons and tons of them out there, and many of them are very good. Claude 2 is another one I didn’t mention, and ask them to write bedtime stories, recipes, create images, really start getting comfortable with these tools. Recently, my son and I asked ChatGPT to write a story with Star Wars and Harry Potter characters. There were some…

0:31:33.7 MF: Oh, the copyright issues that are gonna be there, Afia, if Disney finds out what you’ve done. I promise I won’t tell.

[laughter]

0:31:42.8 AT: Well, don’t worry, my son also objected to the plot holes. [laughter] But it was a fun exercise and it helped me, us, understand how to use AI for creative purposes better. For university leaders specifically, those are the folks I work with, my recommendation is to start documenting what’s happening on your campus. Right now most institutions find themselves in a stage we like to call innovation theater, pockets of innovation budding all across campus but a lack of organization around documenting and scaling promising innovations to their full potential. Now, it is still early, but at the very least, we want leaders to start surfacing and learning what’s happening with these grassroots experiments so they can identify successes, failures, and ultimately figure out which wants to invest in and scale to solve big problems in their highest priority areas. And you can do this in various ways. Some schools have pulled together campus-wide task forces, others are doing symposiums and working groups. The key is regular pull-ups to document progress.

0:32:55.3 AT: Second, I would say is invest in AI skills. While experimentation with AI has become much easier, optimizing AI still requires a new and different set of skills that’s distinct from traditional IT skills. So leaders, especially CIOs, need to train and sometimes hire people on these new skills to optimize their AI investments. And finally, AI tools are only as good as the data supporting them. We like to say it’s 10% technology, the rest 90% is the processes and the data and the people supporting them. With AI, especially, data governance and management is key because that’s how you can ensure you can get the most out of your AI technology. This includes thinking through how to collect and govern new types of data, like unstructured data that generative AI can process. This is a new development and we need to ensure that we are thinking through our data management processes in a holistic way to get the most out of AI in the future.

0:34:07.4 MF: Well, I’m gonna encourage everyone listening to take Afia’s advice immediately and type into your favorite generative AI of choice how to give this podcast a five star review and follow Afia on LinkedIn so you can learn more about the AI opportunities that we were only able to scratch the surface with today. But, Afia, pleasure as always to have a conversation with you. I’m sure there’ll be much more to unpack and talk about in the months ahead.

0:34:32.4 AT: Thank you, Michael, for having me. This was such a fun conversation.

0:34:36.0 MF: And thanks again for joining us here on EAB’s Office Hours.

More Podcasts

Podcast

Is Your University Headed for a Merger (or Worse)?

Experts discuss options available to university leaders who are running out of time to reverse the fortunes of…
Podcast

How to Prepare for the AI Economy

Experts discuss the importance of building AI literacy among students, faculty, and staff.
Podcast

Can Embracing AI Help Colleges Survive?

Former SNHU President Paul LeBlanc and others discuss why institutions must embrace AI to reestablish the relevance of…