Building an AI-Powered University

Welcome to our virtual event, "Building an AI-Powered University," hosted by the OpenAI Forum. The session, moderated by Natalie Cone, OpenAI Forum Community Architect, began with a reminder of OpenAI's mission to ensure that artificial general intelligence (AGI) benefits all of humanity.

The event featured a keynote speech by Leah Belsky, VP of Education at OpenAI. Leah discussed the transformative potential of AI in education, emphasizing the goals of enabling global access to AI for students and teachers, powering AI-driven educational products, and fostering collaboration to enhance AI's educational impact.

The panel included Jason Horn, a technology investment banker and the evening's facilitator, who has a rich background in the education technology sector; Alex Persaud, Assistant Vice President of Enterprise Engineering and Operations at the University of Maryland, who manages critical infrastructure and co-chaired the university’s AI commission; Dr. Jaci Lindberg, who combines roles at the University of Nebraska to push innovation in digital learning and IT strategy; and David Brubaker, who leads AI initiatives at the Wharton School, focusing on aligning AI with educational priorities.

During the discussion, panelists shared their diverse experiences with AI implementation at their institutions. They highlighted the impact of AI tools like ChatGPT in enhancing both the student and faculty experience and improving operational efficiencies across university functions. The session transitioned into a virtual networking event, allowing participants to engage directly with thought leaders and peers interested in AI's future role in education. The event concluded with Natalie Cone encouraging participants to continue the dialogue on AI in education through upcoming OpenAI Forum events and platforms like LinkedIn and Substack, fostering ongoing collaboration and innovation in the field.

Welcome to the OpenAI Forum and our event, Building an AI-Powered University. I'm Natalie Cone, your OpenAI Forum Community Architect.

I like to begin all of our forum events with a reminder of OpenAI's mission, which is to ensure that AGI, or Artificial General Intelligence, benefits all of humanity.

Tonight, we're welcoming OpenAI's VP of Education, Leah Belsky, and a panel of thought leaders in the education space.

And after the roundtable discussion, we'll move into a virtual networking event.

Let me introduce our special guest for the evening.

Jason Horn, our discussion facilitator tonight, is a partner and co-president at GSB.

He spent the majority of his career as a technology investment banker, specializing in the education and human capital management technology sectors.

Throughout his career, Jason has personally executed or advised on a wide range of transactions within growth subsectors across the pre-K through 12, higher ed, and enterprise learning and talent continuum.

Alex Persaud is the Assistant Vice President of Enterprise Engineering and Operations within the Division of IT at the University of Maryland, a position he's held since 2017.

He oversees critical areas such as the campus network infrastructure, on-prem and cloud infrastructure, identity management, data warehouse analytics, and software development, including financial, HR, and student information systems.

In addition to his role, Axel co-chaired the University of Maryland's Presidential Commission on AI and was honored with the President's Distinguished Service Award in 2022 for his outstanding contributions to the university.

Dr. Jaci Lindberg has worked in higher education for nearly two decades, serving in a number of key leadership roles that have served to accelerate innovation in post-secondary education.

She currently serves jointly as Associate Vice Chancellor of Innovative and Learning Centric Initiatives at the University of Nebraska at Omaha, as well as Associate Vice President for Digital Learning and IT Strategy at the University of Nebraska System.

David Brubaker is the Senior IT Director for AI Initiatives and Strategic Partnerships at the Wharton School of the University of Pennsylvania. He leads efforts to align Wharton Computing's AI initiatives with the school's priorities, focusing on enhancing learning experiences, streamlining faculty support, and fostering innovation.

With 19 years of experience at Penn, David is recognized for his collaborative approach to creating positive organizational change.

Leah Belsky is a Tech Executive and Board Director with over 20 years of experience leading high-growth business and initiatives in the education, tech, and health sectors in both public and private companies.

She is currently the VP and GM of Education at OpenAI. Leah has been recognized by Fortune Magazine as one of the 40 Under 40 Leaders in Tech, by Inc. Magazine as one of the 15 Top Emerging Executives to Follow, and by GSV Ventures as a woman leading the AI revolution in EDU.

She also served as a member of Obama's Technology Policy Committee.

She is a graduate of Yale Law School and Brown University and was a fellow at the Yale Information Society Project and Counsel on Foreign Relations.

Please help me welcome to the stage, Leah Belsky.

Hi, Natalie. Hi, Leah.

So good to have you. Thank you for hosting us.

We're excited to be here. It's absolutely my pleasure.

I am going to leave the stage and hand you the mic, but I'll be back at the end of the panel discussion to support everyone with the audience Q&A.

All right. Awesome. See you soon.

Alright. Well, so good evening, everyone. First, I'd love to extend a warm welcome and thank you to our university partners, to GSV, and to all of you for joining us today.

We had a tremendous response to this webinar, and it's super inspiring to see so many education leaders here who are shaping the future of learning and up for co-creating together, because that's really what we're doing here.

As we've been building the education vertical at OpenAI, we've focused on three key goals.

The first is to enable every student and teacher globally to access AI.

The second is to power products and platforms that advance AI outcomes in education.

And the third is to engage the entire ecosystem and all of you to share best practices and drive progress in AI for education and for research.

Our conversation today on transforming universities hits on each one of these goals, and so we're grateful that you're here to dive in.

We imagine a future where AI is as integral to the university's experience as accessing the library or jumping onto the internet.

When a student enrolls in the same way that college students get access to their email account or their LMS, we imagine they would get access to the latest and greatest intelligence.

That intelligence could operate as a personal tutor, a mentor, or a career advisor, all seamlessly integrated into their student journey.

For professors and researchers, AI would supercharge their work and their jobs.

Students would be able to engage more deeply with personalized curriculum and research opportunities.

Professors would have tools to personalize learning, hold dynamic office hours in the middle of the night, and unlock new frontiers of research.

And on the operational side, AI has potential to streamline university business models, optimizing enrollment, financial aid processes, improving student services, and bolstering retention.

As a community, we are still in the early stages of shaping this vision and sharing our learnings together.

So today's panel is going to highlight three institutions leading the charge in becoming AI-native universities. They are each at very different stages of adoption and took different approaches to their deployment, some starting with launching pilots, others driving projects across teams, and others unleashing access to their students.

You should know that the universities sharing their stories today are all customers of ChatGPT EDU, our education-focused AI product.

And as you listen, I'd invite you to reflect on where your own institution is in the AI journey.

And I welcome you to continue the dialogue here in the forum, on the OpenAI Education LinkedIn page, and in our Substack.

I will be joining the Q&A session with Natalie and the group later today, alongside Jason and our incredible panelists.

And for now, let's jump in.

Please join me in welcoming our panel. Thank you, Leah. Great to see you.

And excited to have you back on for the Q&A session in a couple of minutes here.

And thanks to Natalie for kicking us off.

I'm Jason Horne, partner and co-president at GSB.

I'm really excited to be here with you this evening.

And as Leah mentioned, we've got a great turnout, which is really exciting to see.

I see this level of interest and enthusiasm on this topic.

And to me, it looks like there's pretty broad understanding that the AI-powered university isn't some future possibility, but rather our present reality.

And yet we are still clearly at the very early innings of this process and this wave.

And we have a lot to learn around the most appropriate deployment of AI, most efficacious deployment of AI.

So we are lucky to be here this evening with David, Axel, and Jackie. Three of them and their institutions have been out in front, early adopters of this technology, have been working deeply with the Chatspouty platform and OpenAI team and AI broadly for the last six months plus.

It was a lot to learn from their experience.

So I hope to dig into all those things and things Leah highlighted. There's a lot to cover and not a lot of time.

So I'll dive in, Jackie, David, Axel. Just speaking to you and other leaders in this space, it's clear that there is no one way to deploy AI.

Leah just mentioned even three of your institutions have varied approaches, varied strategies, priorities around teaching and learning, operational efficiencies.

I think it's helpful for the audience to hear this diversity and approach. So I thought we could start and really set the stage by having each of you briefly describe what your deployment of Chatspouty looks like so far, maybe who has access, how are you getting it into their hands, those sorts of things.

Jackie, maybe I'll hand it off to you first and then David and Axel.

Yeah, that sounds great. Thank you, Jason, and excited to be here tonight. So at the University of Nebraska, we're a four-campus system. That means that we do a lot of our work together, but not everything. And so even our four campuses that are at a little bit different place with our rollout of Gen AI. I spend a lot of my time on the Omaha campus, which is about 15,000 students. So I'm going to describe the approach we've taken there as we've moved ahead fairly quickly.

So AI is part of our strategy and the licenses with open AI are a part of a larger strategy. So in our college of IS&T, they've launched full degree programs in AI, which is really exciting, great opportunity for students to go deep in building their understanding. But then we've launched what we call an AI learning lab through the unit that I lead, which is a little bit of an R&D shop who supports other initiatives like online learning and teaching with technology and micro credentials, things like that. And that AI learning lab is really focused on breadth of use and piloting.

And so when we think about how we can get AI into a lot of folks' hands, we've started doing really cool things like working with faculty on course revision in all sorts of courses across our entire six colleges, which is really exciting.

The open AI licenses that we use, we have a thousand licenses, which is not enough for everyone. It is cost prohibitive at this time for us to have enough licenses for everyone. And we weren't sure how quickly the interest would come candidly. And so we run an RFP process where folks express interest once a semester, and they bring us examples of projects that they think they might want to do. We have a review team that takes a look at those, and then we award access each semester at a time.

We offer plenty of support to folks doing that. We offer compliance training to folks doing that. And then we do a lot of storytelling around those efforts, which has been a lot of fun. And so it's really exciting for me to see the different use cases, which to a lot of the introductory comments already. They really do span from research to teaching and learning to operational efficiency. And I think that's really aligned with Chancellor Joe Lee's vision for doing the pilot the way that we have, is just to see what broad impact we can have and have some fun as we innovate. So that's how we're getting started at the University of Nebraska.

Thanks, Jackie. And I'm excited to dive into some of those use cases and those adoption patterns you mentioned in a couple of minutes, hopefully.

Thanks, Jason. And it's great to be here. I wish I could see all of you, but in this case, that is not possible. So at Wharton, we're fortunate to have a smaller student population than what Axel or Jackie have. And we also have, I think, generally some more resources. And so we've been able to deploy ChatGBTEDU to all of our MBA students, staff, and faculty. And what that means, we're still figuring out.

Because a lot of, I think, and I'm going to say this in general, everyone thinks they're behind, but no one is that far ahead is the reality is. Because even though we have this and we've deployed it to all these people, we still don't know exactly the use cases and the places it should be used.

And so what we're doing is a lot of pilots, a lot of understanding what works, what doesn't, getting into the hands of people who are the power users, the people who are excited about it, willing to try new things. We're doing pilots at different chatbots. We're figuring out how research can be augmented with it. There's so many areas of possibility, but we don't know where the most effective ones are.

And even at a fairly resource rich place like Wharton, we can't just throw endless money at things. So we have to prioritize. So I think that there is, while there's differences in our ways we've decided to do things, we've been able to do things. There is, I think, a consistent theme across what we're all doing, which is figure out what works and put our energies towards that.

At the University of Maryland, our size is probably similar to Nebraska, where we have about 40,000 students and another 10, 15,000 faculty and staff. So we also have a thousand licenses of Chat2BT Enterprise, and we've allocated those to the power users on campus. So some of the people who are paying for Plus already, we converted them over to Enterprise and the ones that really wanted some rich features.

We've also worked with a contractor to help develop a front end to Gen AI, but it's all powered by Chat2BT 4.0 on the back end, 4.0 and 4.0 mini. So what we've been able to do with that is provide it to the whole campus. So we launched this late in December and then made it available to the students, actually on Monday, the first day of class.

And it allows us to do, to take a consumption-based approach. So with that, we're able to put quotas on everyone. So the students get a smaller quota, but they use 4.0 mini, so it's still pretty decent. The faculty and staff, we've allocated 4.0 to them and given them a little bit more quota. We can kind of cap it, so we know what our spend is going to be.

And as Jackie said, as David has alluded to, it's kind of an experiment. We're trying to see who's going to use this and just what the adoption will be like. But then we've also developed custom solutions for classes, such as these virtual study assistants for faculty to deploy in their class. They're able to upload their content. It'll connect into a Panopto, pull transcriptions from video recordings, and then students can ask questions to that. And they pretty much have a virtual TA available to them at two in the morning when they're studying.

So we've done that and also deployed other specific chatbots for administrative uses on campus. So we're kind of hitting everything. And the things that work well, we're going in deeper. The things that don't work well, we're just evaluating to figure out, is it too soon? Was our approach bad? Things like that. That's great, Axel. I imagine you're going to get a lot of questions in the Q&A or people following up and understand more your approach, an efficient approach to get an incredible power of the technology into the hands of more students, but a different set of features, perhaps, than David's MBA students. So I think that's fascinating to hear and hopefully unpack more a little bit later. Maybe staying with you, Axel, a moment ago Leo was talking about the AI-powered university. It sounds like that exists where a chat GPT and these AI tools sit in the hands of students, faculty, and staff. And I think at some level, each of you are doing that at some level, which is fantastic. I'd love to double-click on maybe the staff piece and the operational efficiency opportunity.

It seems like a huge opportunity from my perspective, not living and breathing it every day like the three of you in so many ways. And of course, you have an urgent one for some. We hear about demographic cliffs, enrollment declines, rising operational costs, and all sorts of things of that nature. I'd love to hear from a few of you, Axel, maybe starting with you, what you're thinking about that opportunity, what you've learned, what you've seen.

One massive opportunity we were able to take advantage of is in the middle of November, we actually transitioned to Workday for our HR system and financial system for the whole campus. Prior to that, we were on a custom-grown solution, and we got our money's worth out of that. So in transitioning to Workday, it was going to be a violent shift for the entire campus. So literally a few days before the actual day, my boss came to me and asked me, "Hey, can we spin up a chatbot for people who have questions about the transition and how to do things, just because the whole procedure is different?"

So my team was able to do that. Once again, all powered by ChatGPT on the back end. We worked with the Workday team, uploaded all the documents, and it was a massive success. So we were averaging for the first two weeks when it launched about 700 questions per day, within the, yeah, 700 questions per day and about 10,000 questions for those two weeks.

After that, it trailed off a little bit, but it was 300 questions a day for December and January, and 20,000 questions in total for that time period. That was a massive saving, because all of that would have hit the services. The other thing that was interesting was some of the heavy users were not necessarily the campus community that didn't know what was going on. It was the actual people that had to support the platform, because it was new to them. So they were using the bot to figure out answers, and every answer also provided a link.

link back to source material. So whether it was a video, it took them to the exact point in the video, showing them how to do that task or a PDF or something like that. That was a massive success with regards to an administrative deployment.

Well, that's amazing. Those numbers are incredible. Jackie, I think in the lab, it sounds like you're looking at some simple opportunities.

Yeah, and some of the staff and administrative uses are some that I'm the most excited about and some that we're hearing some of the most positive time savings was when we asked folks their experience. So yeah, we've got folks in campus security who are, because we use our RFP approach, we're asking folks what they want to do, right? So campus security has said, hey, we wanna try to use this to do some crime analysis and predictive mapping to anticipate future needs. We have folks in HR who are saying, we have all these staff, but we don't necessarily have them on learning plans of how university staff can continue to upskill and reskill. So they're working on that project. We've got advisors who are using this to try to do more immediate real seat availability in more real time. And then a lot of units who are looking at chatbots, right? Dean's offices, HR admissions, which I think is great.

I do wanna say though, so of these staff projects, we've asked them, we allocated our first set of licenses this fall and we said, how's it going, right? And 90% of the folks who are in there said, it's going well and it's making me more productive. So we asked a little more. We said, what does that mean in terms of numbers? The vast majority said it saves me a couple hours a week. But one of the things I found really interesting is about 15% of the people in our pilot, again, thousand licenses, said it saves them five or more hours a week or even 10 or more hours a week. And so I think that's really exciting as we think about all of the staff and administration across I think all of our campuses who are stretched really thin, who are always being asked to do more with less. I think this is a really great opportunity to have your friend chat GPT, do some of the really kind of monotonous work for you and reallocate your own time to doing really, really critical thinking tasks.

That's amazing. And 10 hours, I haven't heard that level of efficiency again, really anywhere. That's incredible. And I heard some of the examples, like advisory space is one, so advisors can advise and not worry about focus on things like scheduling and things that shouldn't be. Crime analysis is a new one. So I always love asking this question because there's some new things. Financial services, financial aid, it's another big one you hear often. A school we're close with at GSB, I think a third of their financial aid calls have been automated, which is a huge, huge lift. So, so many opportunities there.

Jumping to maybe the teaching and learning side of the house and David, I'll go to you. Your colleague, Ethan, has talked about the homework apocalypse. What's your reaction to that? What sorts of things are you seeing as students and faculty have adopted this, or your MBA students, I guess, specifically?

Yeah, it's interesting because there's like anywhere, there's a huge range of response. So you have the, on the one end, you have the ephemolics of the world who are diving in at the forefront, trying everything, expecting their students to use it for every single project, creating homeworks that are based on it. And then you have the other extreme where people are scared to death of it and don't really understand. And so a lot of our work is helping faculty understand where it's helpful to them and not forcing them to do it in a particular way. We find, obviously, you don't generally tell faculty what to do, that is rarely productive. But we have found it useful to provide some examples of what can be done.

So one of the pilots we've created is, we call it lecture recall. It's actually similar to what Axel was describing where we record all of our classes using Panopto. And so we have the transcripts from every lecture and we can ingest those into a database that then can be accessed from the chat GPT front end. So we use a custom GPT with the students, the enrolled students in that custom GPT and then using actions, which if you've never, if you don't know what actions are within a custom GPT, it's worth either you looking into it or ask your IT staff to look into it because there's a lot of power there where you can be reaching out to, calling out to an external data source and bringing in data. And so we're keeping the students in the ecosystem we want them to be in within chat GPT, EDU, protected environment. They're able to access and query this data, like this valuable, valuable data that they would have to go through and like listen to hours of lecture to find something. They can find things quickly. And then we link directly to that point in the video. So they could either get a video summary, I mean, a text summary of what happened in a lecture they missed, or maybe something they forgot about, or they could ask for a particular point and then link directly to that in the video. So that's been, I think, an interesting path to provide faculty a way to use it that is not way out of their comfort zone.

Yeah, it's interesting. In some ways it's a response to the homework apocalypse, which I think what you meant by that is any most cognitive tasks can be tackled pretty efficiently here and what your students have shown by using a tool like the one you've built is that they're not looking for a simple answer. They're digging into the material, they're watching lectures, and that of course can be pretty linear. You've made it more efficient, but they're still going through that process. So that's pretty fascinating.

I guess staying on the topic of that, and we've gone through a couple of fascinating use cases, but I'd love to double-click on adoption since you've just given some great examples of how it's being adopted. David, have you seen the level of adoption you anticipated? The number was, what, 1,000? Or no, your entire MBA student body.

Yeah, sorry, Jason, go ahead.

No, no, please.

To the extent you're willing to share, what does that look like?

Yeah, I'm happy to share. Please. It's, we didn't know what to expect. I think their expectations varied depending who you talk to. We have about half of our MBA students actively using it. Now, there's a lot of subtlety there because the reality is most of them are using Chad GPT in some form. It's just not necessarily on our EDU instance. So one of the things we're working on is figuring out how to encourage them to use it because as we understand why they're not using it, there's a couple of reasons. One is they're worried that they're being monitored, right? So I think no one would be surprised by that, but we have very clear privacy policies and expectations for privacy. They just don't necessarily know about them or they don't feel confident about them. And so reiterating what those are and being very clear upfront what they can expect from a privacy perspective, and then also creating reasons for them to be using EDU as opposed to a free account or whatever else they have outside. And so as we create more infrastructure, custom GPTs, other value within the EDU ecosystem, I think it will naturally go higher than that. We're just still in the phase of figuring out what those things are.

Yeah, that's fascinating. And yeah, I think it makes sense, as you said, pretty intuitive that that would be a concern and it sounds like one you're tackling.

Jackie, you mentioned in your intro that you also were unsure of what to expect around adoption. Have you seen similar resistance or concerns maybe from my students or faculty?

Yeah, sure. I mean, a lot of questions, great questions, especially right out the gate. I think faculty asked a lot of questions about, well, when are we gonna have a conversation about the philosophy of doing this before diving in, right? But I shared in my intro, I worked in online learning for a couple of decades and I personally compare the speed of interest in this just wildly different, right? So online learning, I think it took us 20 years and a pandemic to get the level of interest.

excitement that we've seen in a year with Gen AI. That's my personal experience. And so I think the pushback, the questions, the concern were all very valid. But at the same time, the adoption has been way higher than I personally was expecting.

So of the 1,000 license and the RFP process we use, we received 150 proposals. Now that's either a unit or a faculty member teaching a class saying, raising their hand they want to use it. That's high for me. It's high, especially when you think about other things that we've rolled out slowly in higher ed, maybe like online learning. I never had 150 people the very first semester say, I'm into it. I want to do it. It was more like two or three.

So I think that's really exciting. And I think then, as people go into this second semester, I mentioned that we do a semester by semester license allocation. Many of them wanted to roll forward again. That was another kind of thing we didn't know what to expect. If someone tries this out for three, four months, are they saying, oh, I'm good, not for me? Or are they saying, I want to do even more? I want to try some different work. And we did see a lot of people want to carry forward, which is a great sign.

Do you have a theory on why this has been so quick? COVID is a prerequisite for the pace of adoption we're seeing and the way we're all forced to move a couple years ago? I think that's part of it. But to me, it's also that it impacts all parts of your life. And you can see the benefit fairly quickly in a really basic way. It helps me plan a vacation very differently than I would. It helps me prepare to tutor my fourth grader differently than I would.

And so I think in some ways, that really changes the understanding, is when you're able to use it in very safe, approachable ways. And then you go into the workplace, and it's sort of like, well, why wouldn't we just ask ChatGPT for ideas of how we're going to run a breakout room at our next conference? I think it just has made it a lot more approachable, because it's not just specific to one part of our lives.

Yeah, I think that's such a great point. And it's something our team at GSB talks about a ton, how approachable this technology is versus other waves and things that we've seen. Remember when we all needed to be a coder to have a productive future? And it was such a steep climb to do that. Well, this is such an approachable and flexible and powerful technology. I can sit with my six-year-old. I can hand the phone to my mother. Pretty much everybody can pick this up. And there's certainly levels of aptitude and skill as you get comfortable. But I think that's such a great, great point around how flexible it is.

What I'm just fascinated by is the pace of change in this space. There's the expression, there are decades where nothing happens, weeks where a decade happens. It feels like for the last two-plus years, we've only been living in the weeks where decades are happening in this space. This week's no different. I mean, this company, OpenAI, a huge infrastructure, investments announced, amazing agentic product. We've seen huge international competition announcements that have rocked the AI world.

So it's really been continuing to be pretty remarkable how fast this is changing, the investment that's being deployed. I'd love to just, maybe David or Axel, if one of you wants to kick us off, I'd love to hear how the pace of change influences your strategy, some of the things you've talked about doing. How do you do that in light of how quick the space is shifting under our feet?

David, I'll have you kick off. All right, sounds good. Well, it's a great question. And our main focus is not to build anything massively custom. We just don't do big projects that would take significant amounts of time, significant amounts of resources. It should be all about connecting pieces you already have, using smaller pilots, trying things, figuring out what works, and then once you are there, chances are something has already come out that has either replaced what you were doing.

We're looking to our third-party vendors and partners to see what they're doing to understand what their roadmap looks like and to make sure that what we're doing is not trying to replicate work that is going to exist in existing platforms in the coming months. And just always assuming that there's massive change coming the next day and be ready for that, be nimble. Don't build massive custom things. That's our main mantra right now.

Axel, do you have a similar view? You've built the Panopto integration as one example. I don't know how much of a lift that was, but you must also be thinking about what's ahead, so as to not build something that's going to be off the shelf.

I'd say two things. So one, we are building custom solutions, but we're making sure that they're repeatable. So one thing that we're doing that's come up recently within the administrative space is document review. So this can be contract review or software licensing review or something, because our legal team and our procurement team, it has to have certain terms and services.

And for them to go through all this, it's kind of laborious. So for us to have a bot that has the prompts with the stuff that's relevant to the state of Maryland, the University of Maryland, makes their lives a lot easier. And it can really bring their attention to that. But that's one thing. But then once we have that platform, it's just a matter of cloning that and changing the prompts, because we have, I think it's two or three people that work with the state of Maryland's General Assembly.

And they told me that when it's in session, it's about 3,000 new bills that come out. About 1,000 of them are repeats, but 2,000 are new. And this man has to go through and read all of these things. So we were able to work with him, built an application that'll go and download the new bills each day, but then go through that document review process with the prompts that are appropriate for his office.

So to me, the custom stuff are things where a slight tweak and it can be reused for other things. There have been times when people have come to me with really interesting and exciting opportunities, but I'm like, ah, I'm not doing it, because it just benefits you and it'll make your life better but I'm focused on providing service for the entire campus. So that's kind of our approach to it.

And just kind of like what David said, if a vendor is doing something and it's general purpose for everyone, like what we just rolled out for our entire campus, then we'll partner with the vendors. Our custom solutions are very focused on our unique data and something that no vendor could potentially replicate. That's great. That's a great framework, kind of avoiding the incremental and looking at the things that really move the needle.

I don't know if you've quantified it, but the contract review tool must save more than 10 hours a week. We have a new record. That's a great one. Jackie, I think it was you. You mentioned an AI degree, I think in the front of the conversation. AI literacy is another kind of hot topic and something lots of folks are talking about.

There was a survey recently published by Cengage, a higher ed publisher that showed nearly two and three employers expect employees and candidates to have these foundational skills around generative AI, and yet more than half of students feel unprepared to use AI in the workforce. So many are, I'm afraid, feeling kind of left behind without these skills as they enter the workforce.

What is your take on AI literacy? Do we all need an AI minor? Should this be embedded in curriculum writ large? I'd love your thoughts.

Yeah, I think it's phenomenal that we've had an academic college want to move as quickly as they have to stand up, not one, but I think they're working on a second degree right now at the graduate level, and so I really cheer them on. But I think for me, as someone who does not have a technical background myself, I just think about the broad need for skill development and the alignment to so many fields.

To me, it's really critical that a graphic designer and a philosopher and a teacher and a historian all are really thinking about and understand how AI could be used in their field, what particular tools could be leveraged, how it will impact the way that they go into the workplace.

So I think that literacy is critical, both breadth and depth. At my institution, we're experimenting a lot with micro-credentials, and I think many are across the country, as we just look at the continual need to up-skill and re-skill. And so if we've had college grads recently that don't feel like they've yet had that learning experience, I think there's a lot of opportunities that don't look like going back to school for a four-year degree.

I think there's a lot of opportunities just to continually lean in and make sure that you have the skills that you need to do the work. And that's one of the things I really applaud higher ed for, is being pretty nimble, I think, especially in the last four or five years, to think about how do we reach learners that have already been with us, but need a little bit different skills.

And I would put AI in that category for sure, even for the people who are using it. Not necessarily like students going into the workforce, but if I take us more back to our own faculty and staff, they're hungry for professional development on AI. And so as we've rolled out things that might be skipped usually, like, hey, you're due for your annual compliance training. I think those are generally activities that people just do as quickly as possible, and move through at the required stages of their year. We've had a lot of people choose to do our AI training who are not currently required to. It's our most popular workshops through our digital learning department right now is around AI.

And when we bring folks, especially alumni and employers from our community to campus to talk about this issue, that's when we're packing the room. So I really agree with you. I just think people are really hungry to learn as much as they can, and also to learn from people doing things in a role or in a field similar to theirs to be inspired, and to get their kind of brain churning about, wow, okay, if someone does that in that field, even tonight hearing David and Axel, I've taken some notes, right? Like, oh, that's how Marilyn's doing that. I wanna go back and check that out. So I just, I hear you. I mean, I think that hungers there across the board, whether it's current students, recent grads, faculty, staff, or folks in outside fields. Yeah, thank you.

I'm glad my question really spoke to student AI readiness, but I should have asked about faculty readiness as well. So it's great to hear the adoption of some of the courses and workshops you've developed. That's amazing to hear.

Axel, I'd love your view on AI literacy question as well.

Yeah, so we've done a couple of things. So our Teaching and Learning Center, because I think similar to Jackie, educating people is important, but getting the university to actually develop a course and then put it into the curriculum, like that takes some time. So our Teaching and Learning Center has actually developed some online courses and made it available to the faculty. So they can just take it and insert it into Canvas or online LMS, and it's available for the students. So the faculty don't even have to teach them. They just have the resource to provide to the students.

That same group has also done several sessions with faculty, and it's great because they're awesome at teaching the faculty, but then they're able to provide me, and I take care of more of the technical side, some of the feedback. So jointly, I think that's where we're starting to shine a little bit more, because if one of us do it, we have a very narrow perspective on things, but working together with other units where we're all leaning into our area of expertise and not trying to do things that we're not good at, we'll really make it successful.

That's great. Thanks, Axel. And unfortunately, it looks like we only have a couple of minutes, so I'm gonna try to sneak in one more question. I'll do a quick, maybe it's a quick fire one, and maybe we could go around quickly. Love to hear what, we have a big audience here from a number of institutions. What one or two actionable steps should every campus, every institution be taking to prepare for what's coming here in the AI space? David?

I guess, in the end, don't try to do it all. There's so many things, and it feels endless at times, and pick a few things that you can do. Try them and just get started. Get access for your power users. It doesn't have to be everyone. Pick a few people, like what Jackie was talking about. Get the people who are excited about it to start using it. You'll start finding use cases, which then drives further interest, and you can potentially get more resources than if it actually shows significant value. So it doesn't have to be all or nothing. It doesn't have to be a magical, everyone gets an account, everyone gets like full access to every possible platform out there. It can be much smaller than that, and it can be just very, very effective. Because in the end, the power users, I think that are in many ways the most important, at least that's what I've seen, is the ones who get really excited, start like evangelizing it out to everyone around them, and also start building things on their own, creating processes. That's where it really counts, and that's what's gonna drive adoption longer term.

Yeah, that's fantastic. Thank you. Something you said really resonated with me, that no one really has fully figured this out. Everyone's figuring things out. I think that's a good message for folks to hear that this is a journey, and jump in and come on this journey, this process. And it's great to have communities like this one and others where we can learn from each other, communities to practice.

Jackie?

Yeah, totally agree with what David said, by the way. I think I would just add how critical communication is. And if you're gonna take a little extra time to do something, it's think about how you're communicating this strategy at the end of the day, at the end of the week, at the end of the month. Because to me, if it is a positively phrased, cheering people on, congratulating them on even trying something, making it exciting, that goes a long way right now.

And the reason I think it goes a long way right now is we've been through a lot with COVID, of course, but also there's just so many system changes. There's so many like technical changes, right? Like we're migrating our storage environment. We're changing LMS. And that, no matter how good of a job we do, there's a feeling of pain. There's a feeling of, ah, ooh, you know, like there's gonna be some loss there. This is not that, right? Like this doesn't have to be that. And so take the extra time to say, how am I going to build in storytelling? And how am I going to create energy and excitement to David's point with however many people are trying this to inspire others?

That's great. That's really helpful. It even resonates with me and how, we've said we're an AI first team and we celebrate in a dedicated Slack channel how people are using this and kind of create a playbook.

Axel?

I'd say I'd echo everything Jackie and David said. For us, it was also finding some of the low hanging fruit because we put out a wide net to people.

And one of the things I found most interesting is a lot of people want it, but initially there's only a subset of people who are actually willing to do the work needed to move their efforts forward. And at one point I had to pause and I was thinking, okay, am I trying to teach people to fish or do I just need to feed them right now? In some cases I work with people to identify their problems and then we were able to use Gen AI to help solve them. Then that sparked more curiosity and then it was easier to teach them. But trying to really teach a bunch of people upfront, as Jackie said, there's a lot of change and a lot going on and people have very minimal bandwidth in some cases to learn something new.

They just need their process or whatever it is they're working on made more efficient. And my focus has been shifting to that because, and like I said before, if they see it and they touch it and they experience, then more ideas, like Jackie said, would pop off in their head and then they will start hopefully to adopt it on their own.

Amazing, that's fantastic. Last thing I'll say is also reach out to community in terms of there are partners out there that are willing to come to your campus and do promptathons, do training sessions, do things like that. And I've taken advantage of that from a variety of different companies because there's certain things that the IT group is good at, certain things we're not good at and facilitating some of those sessions have been phenomenal when outside partners come in. So don't neglect that and they'll do it for free. What's better than that?

That was fantastic, brilliant conversation.

and so many amazing nuggets of wisdom there. That was absolutely amazing. Thank you all so much for your time this evening. The event isn't over, but we are gonna be moving away from the panel discussion and into the one-on-one matching for a networking event. So I just want to express our deepest gratitude.

Leah, thank you for joining us. It's always exciting to host you at OpenAI. Axel, Jackie, David, Jason, this was really fun, and we're really grateful for your time this evening. Thanks for having us. Thanks.

Before we go away, I'm gonna remind everybody of some upcoming events, which you can find in your events tab of the OpenAI Forum.

Next Thursday, February 6th at 5 p.m. PT, we'll be hosting a virtual talk from Microscope to Monitor Stanford Pathology's AI Journey.

February 12th at 4 p.m. PT, we'll be hosting AI Native MBA Slam Dunk Strategy with the San Antonio Spurs. The San Antonio Spurs, the first NBA team to adopt an enterprise contract of ChatGPT. And similar to Jackie's experience, they took a really human-centered approach. So can't wait to hear from Charlie more on that approach to adoption.

February 19th at 5 p.m. PT, we're hosting Sora Alpha Artists for an event called Preserving the Past and Shaping the Future. And I just wanna share a little sneak peek with you all in that this spring, those same Sora Alpha Artists will be teaching a Becoming More Familiar or Getting Started with Sora type educational event in Spanish for the Latin American Global Chapter and also in French for the Paris Global Chapter.

February 26th, we're hosting our first in-person event of the year. It's Math is Music, an interdisciplinary exploration of sound science and creativity. It'll also be live streamed. It's a hybrid event so you can watch it and tune in from anywhere in the world.

And then last but not least, March 19th, we'll be hosting How a Paris AI Startup Illuminates History and Culture with Ask Mona.

I hope you all had a wonderful evening. I hope this was helpful. I hope you continue to connect with each other and I hope you have a wonderful evening. We'll see you very soon in the OpenAI Forum. Good night, everybody.