--- title: "A Student-First, AI-Native Vision for the Future" slug: "brian-hemphill-old-dominion-university-a-student-first-ai-native-vision-asu-gsv-2026" author: "Brian Hemphill, Jeremy Singer, Pradeep Khosla, Sian Beilock, Tim Cleary, JP Novin" date: "2026-04-14 12:00:00" category: "Premium" topics: "ASU+GSV 2026, conference transcript, AI in Education" summary: "A senior leader from Western Governors University (WGU) presented a comprehensive vision for how AI can fundamentally transform higher education from a provider-centered model to a learner-centered one." banner: "" thumbnail: "" --- > **ASU+GSV 2026 Summit** | Tuesday, April 14, 2026, 2:10 pm-2:50 pm | Global Higher Education ## Speakers - **Brian Hemphill**, Old Dominion University - **Jeremy Singer**, College Board - **Pradeep Khosla**, UC San Diego - **Sian Beilock**, Dartmouth - **Tim Cleary**, Risepoint - **JP Novin**, Plexuss.com ## Key Takeaways - A senior leader from Western Governors University (WGU) presented a comprehensive vision for how AI can fundamentally transform higher education from a provider-centered model to a learner-centered one. - Using Amazon's "working backwards" methodology, he outlined WGU's approach of envisioning a desired future state and reverse-engineering the systems to achieve it. - He presented seven concrete examples of AI application at WGU: redesigning programs with personalized GPS-like roadmaps, using cognitive task analysis to build micro-courses from job requirements, personalizing and adapting learning journeys so two students in the same physics class may never traverse the same content, scaling Bloom's two-sigma one-on-one tutoring model through AI instructor bots, implementing continuous mastery assessment instead of episodic testing (yielding 10+ percentage point increases in first-attempt pass rates), creating interoperable digital transcripts (already serving 120,000 students), and redefining access to serve the entire workforce rather than just traditional enrollees. - He closed with five strategic considerations including using digital twin/synthetic student technology for low-risk innovation, taking a holistic integrated approach, maintaining a pro-human stance, ensuring credential portability, and contemplating whether we are entering a "post-institution" future. ## Notable Quotes > "Of all the supply chains that exist in our economy, the least digitized of it is the talent supply chain." > > — **WGU Speaker** > "If both of us are in the same physics class, we may never even traverse the same content because it's composite in its nature for us individually and specifically." > > — **WGU Speaker** > "More than 90% of individuals can develop mastery in any subject... that's the design approach we're taking." > > — **WGU Speaker** > "We have to be able to prove that the human approach to a thing is better than the AI native approach to something and vice versa." > > — **WGU Speaker** > "Transformation is actually realized through action." > > — **WGU Speaker** ## Full Transcript Thank you for the opportunity to be with you this afternoon. I look forward to just sharing with you a notion of what a future can be envisioned that makes it very learner-first and really creating an environment that in fact is what they deserve and actually how AI enables us to think about that or envision that future state without constraints. That's the objective of this. And I'll share with you a model or a method or an approach, if you will, as to how to think about transformation and how we do that, but I'll then provide seven examples of how we're utilizing that AI to help us redesign, reimagine, rethink about how we can do that in instruction and in learning, in design of programs, curriculum, et cetera, and then leave you with five considerations. So first and foremost, the pattern of transformation is like many of you have seen these charts before, but there is simply the reality is that over the course of human history, that there are these catalysts that exist that create a moment or an inflection point where transformation actually does occur and actually splits what actually is going on, that whatever model existed during the first period starts to actually begin its death. It's reached its apex, and you utilize that catalyst to start driving the transformation into the next phase of what the evolution is going to be. And then comes another catalyst and another period of transformation that then brings in a new era, and on it goes. If you actually apply this model of transformation in higher education, you might see this look like this somewhat, and this isn't a perfect example, but certainly you could go back to the Industrial Revolution, you realize that preceding that, but really starting to scale it, you saw the emergence of a modern baccalaureate degree, that that modern bachelor's degree moved beyond education and teaching and theology into actually providing degrees or pathways or programs that were relevant to the opportunities that individuals were going to be pursuing. But then came about this statement of principles, followed not too distantly after that with the GI Bill and the Higher Education Act that started to actually put in place certain design inputs that brought about the modern university. By that, what I mean is like you can think of examples like the Carnegie unit or the credit hour. You can think of academic freedom or tenure, but you can also think of the federal student aid complex. And during that period, that modern university started to advance. But then came the Internet, and institutions like WGU became these first kind of digitally native universities. That Internet allowed us to democratize access in a way that we didn't imagine before, that we could reach and teach individuals where they were. Well, what is now AI going to do to that? One of the simple things that we think about is how does that potentially democratize learning and the access and persistence and traversing of that in a way that is much more accessible, completable than it was before. If you also look at the time patterns here, it's like one thing you will have noted is that the time starts to compress between these catalysts and the periods of transformation while the impact scales and compounds differently than how the time compresses. You can even see this in how it existed in higher education. So let me share with you kind of a notion of how to manage through that period of transformation. Many of you may know that my background was not in higher education. I jokingly somewhat say it's like I don't know what the board was thinking when they decided to hire me 10 years ago. But I did learn from a wonderfully innovative institution, Amazon, that there is a model that says you have to actually envision a future state. It doesn't need to be perfect, but you have to have a sense as to what, in fact, are you desiring as the outcomes in that vision future state. From there, you start working backwards. The first thing you start to understand is what then are the strategic tenants or the design principles and the key metrics or key results that need to exist for you to realize that desired future state. And then from there, you actually go into then the purpose-built products, models, processes, solutions, everything else like that. So rather than focusing on where we are or the tradition or the practices that already exist and building from there, you actually start from the end in mind and work backwards. So you can actually envision your upward slope rather than staying on your current trajectory, which is a downward slope. Here's a sense of what we imagine this might look like in the future state, powered by AI, learner-centered. First and foremost, that is at the center of everything you do. Around that, you have to absolutely have in place trust. That trust exists not only with the individual, but also that individual in an ecosystem of talent or a talent economy, if you will, that exists between the individuals, all of us who are providing the skills into the workforce, and all of those consumers of our talent, employers, the workforce more broadly. There has to be a system by which, in fact, we have digitized that in a way that hasn't existed to this point. But the other thing I don't want to lose sight of is that we are the principles here. The individual is, in fact, the one who has the agency to decide what is it that they want to do for themselves to live a self-directed life. If you start with that core, you start to see certain design principles come about in our envisioned future state. These are some of the ones, not some of the ones, these are the ones that we actually focus on at WGU. You'll note some key phrases in there that are very unique about if you want to be learner-centered, you're actually thinking about how do you personalize, customize, individualize, whatever that verb is, to ensure that every individual has the capacity to access it, to traverse it, to attain it, and ultimately live a self-directed, economically mobile life. Well, to do that, you have to think about the relevance and the timeliness of that pathway to the opportunity. You have to think about that architecture of it from a skills architecture, a competency-based model. You also have to ensure that it's networked in a way that, in fact, gets us from where we are to where we want to be. This envisioned future state, for us, actually recognize this is at the center of the talent economy of tomorrow in our sense, that really empowering individuals to be economically mobile in their lives, to do that in a self-directed way. In turn, you're also powering a workforce of the future, that those employers are also discovering and sourcing and hiring all the talented individuals that they need into those jobs of the future. Let me share with you these seven examples of how we're doing this at WGU. This notion here simply to present is like it's from a university to dot, dot, dot. It's not an AI university. It's actually trying to solve for how do individuals acquire the knowledge, skill, and ability they need to progress on the pathway to the opportunity they want in the future. That may or may not be in the construct of a university. First example, you have to think about redesigning programs and pathways. If a program is a credential or something that, in fact, exemplifies the knowledge, skill, and ability that I have that's relevant to an opportunity, those programs can in fact be not just the first one but a second one. But the pathways have to recognize that all of, in this case, nine different starting points, that all the modes and methods and costs and time, et cetera, and duration and routes of how they get from where they are to where you want to be are different. There's no standardized model by which we're doing that. We're following a personalized roadmap to that. It's not one time. It's another time. And it's another time. And it's another time. So keep this kind of GPS orientation in your mind as you think about this second example. How does AI actually help us enable cognitive task analysis starting from the opportunity and the role and the job at the end and identifying actually what are the skills then required to actually be successful in that role? Designing then the kind of competencies required, the assessments, et cetera, and you start to realize you get towards more micro-courses rather than large-scale duplo-block-like courses. You start to think about varied and hybrid delivery, meaning all can occur in a classroom on a campus or, in our case, even in a one-on-one digitized interaction between you as a student and instructor. And you move from this kind of fixed path model and academically start from the provider side into a future, and you actually start from the market side and drive that into the customization of those paths, the programs, and the curriculum. Next step you can see is how I'm going deeper into this, is how we're utilizing AI. You heard some great examples from the panel before this, is that personalizing learning and adapting the journey means that in the specific subject matter that you're trying to develop proficiency and competency in right now, the instruction has to occur in a way that is right for you, not in the standardized delivery of content model that most of us have known. But that adaptive journey also means that it has to be sequential for me personally, the depth of it has to be relevant to me personally, that if both of us are in the same physics class, we may never even traverse the same content because it's composite in its nature for us individually and specifically. That also means you start to realize that you're going to get to this variable assessment model rather than singular summative assessments. Because if there is no standard DUPLO block course, then the assessments themselves are no longer standard final exams. Next example. Many of you in the room have probably known Benjamin Bloob's study over four decades ago at Northwestern Chicago who highlighted this simple point, that in a conventional instructional model where we have a single faculty member broadcasting all this content to a classroom of 30 to 300 individuals, that's a content delivery model. In that model, 20 to 40% of individuals develop mastery, and it's a wide distribution. As you have mastery with corrective instruction, you in fact improve the proficiency and the percentage of those who achieve it. But what he really realized is when you combine that corrective instruction with one-on-one tutoring, you in fact see that the median proficiency of that group of individuals is two standard deviations better than those in a conventional instructional model. But more importantly, more than 90% of individuals can develop mastery in any subject. That's the design approach we're taking to WGU. Now many of you already know this, is that we designed an instructional model that is already one-on-one. It's like 24-7 office hours between students and faculty. But what does AI do for us now? You can so dramatically scale up that tutoring one-to-one model in a way that we at even at WGU have never contemplated before, because why? So many of those instructional interactions are actually enabled by an AI instructor bot who's utilizing that custom composite content and curriculum that's relevant and timely to where you as an individual are in your journey to develop mastery in that specific thing. So for the first time, we actually have the opportunity to combine both Bloom and Kettinger in 2023 et al that highlighted that pretty much there's an astonishing regularity in the pace of learning, that we need seven attempts to in fact develop mastery in any given thing. Well, how do you do that then for the vast, diverse population of individuals that we need to enable? Going deeper. You just move from episodic testing to now to continuous mastery. How do you do that? Because now you're actually assessing proficiency as you're moving along this personalized content. This leverages that cognitive task analysis that I talked about earlier, but then it structures it into those domain models to ensure that the mastery is developing continuously rather than only measured at one high-stake endpoint. What we're finding at WGU is it dramatically increases the attempt rates of individuals for the exams in the first place. And the first attempt pass rates have gone up notably, like 10-plus percentage point increases in the pass rates. But what you also recognize is that it's actually continual. What does this help us do? You start utilizing this AI-enabled continual assessment model into work-based non-academic journeys too, which are in fact much more about demonstrating learning at tasks, activities, engagement, et cetera. The other cool thing about AI is AI is getting really good at observing even and inferring what are the proficient outcomes that you're actually expecting as you observe different models of learning. It even helps design what those assessments are going to be like. Lastly, let me share two final things on this in terms of some of the examples. Coming back to this core trust and agency problem, you have to actually have a means and mechanisms by which you have a trusted verification of an individual's mastery, proficiency, competency, whatever you want to call it, and the credentials that they've attained as a result of demonstrating that competency. That means that you have to have a digital transcript model that is highly interoperable, not across only institutions where you acquire the skills, but with the employers who are actually seeking those skills into those jobs. Of all the supply chains that exist in our economy, the least digitized of it is the talent supply chain. And AI is now enabling that is in fact a screen of the digital learning and employment record that we offer all of our students. We have nearly 120,000 individuals now on it. We are trying to get all million of those that we've served in our journey to be on that. It's actually surfacing that for them. What are all the roles and opportunities for me given the skill profile that I currently have? As well as for all the jobs that I aspire to, what are the skill gaps that I need to address? And then how do I actually recognize the custom and personal pathways that I can acquire those skills to be ready for that job? Lastly, AI helps us redefine access. Even heard it up here in terms of like all those individuals. We're talking about hundreds of millions of individuals in the U.S. alone. They're trying to figure out how to actually change my life for the better, and they know that learning and acquiring skills is the single biggest catalyst for that to occur. But when we get the opportunity to redefine access and convert that skill into opportunity, you know, one individual, our lens goes from this kind of view to say, hey, there's 20 million people enrolled to all of a sudden everyone in the workforce is now someone that needs to be served. Why? Because this is actually about what my next opportunity is, and the job that I'm in is going to be disrupted by AI, and now what skills and capabilities do I need to acquire for the next opportunity? It allows us to now think completely differently about access and enrollment. Because now you start to figure out ways, like I can mass customize at scale learning journeys that serve every individual from where they are to where they want to be. That to me is amazing, because we'll start talking about enrollment in education totally differently in the future. So for all of you in the room, for all of you leading institutions, for all of you on the workforce side of that who's trying to figure out how to upskill and reskill all of your individuals, what will you choose to transform? How are you going to utilize AI to remove the constraints and the parameters that you previously lived under to think in an unconstrained way that's focused on the individual? That you can create de novo, that allows you to abandon the past to kind of settle the future? What are you going to choose to do? So let me offer in closing five considerations in this regard. First and foremost, AI actually should accelerate the pace of innovation while lowering the risk of it. How do you do that? Because in fact now with digital twin technology enabled by AI, meaning it observes all of our students across their entire journey, that you can effectively make a synthetic student for every single individual that you've actually observed. That you can model and test, prototype and pilot things utilizing these digital twins. Synthetic students at a rapid pace that you did not deal with before, that you had no availability to do before. But more importantly, you're reducing the risk, because now you can actually prove in effectively a randomized controlled trial with synthetic students. That which you had to enroll thousands of individuals to start to measure whether or not what you designed was going to work. You get incredible predictive power. Second consideration, and this is certainly true from WGU's standpoint, please take a holistic integrated approach. Most of us who went to conventional degree programs at more conventional institutions, we traverse 40 different courses with 40 different instructors and no one is paying attention at all to how I experience across those different courses and we are hardly utilizing technology in those specific classrooms in those 40 different courses. Think holistically across the individual's journey from that GPS kind of mindset from where they are to where they need to be. How do you utilize a data-driven decision intelligence architecture to drive that? Third consideration, this is a pro-human approach to applying artificial intelligence. We are the principles here. However, you have to be willing to prove it. We even take that mentality with respect to the thousand plus mentors we have serving the nearly 200,000 concurrently enrolled students, that we have to be able to prove that the human approach to a thing is better than the AI native approach to something and vice versa. Fourth consideration, please keep in mind that again, we are the principles here. Think about credential portability. If your employers are on the workforce side and you are thinking about how to re-skill and up-skill things, do not assume that they are going to be with you for decades. So how do they take all that knowledge, skill, and ability they acquired through the learning and training they developed on the job or through other academic models and it travels with them? They are the sovereigns. Lastly, are we entering into a post-institution future? This idea that in fact institutions should provide a one and done education coming of age experience, it's actually freeing us up now to think about structuring learning that is proximal to the individual and the opportunity. It is at the intersection of where they are and the next opportunity. That means there are so many 25 plus year old individuals who now we can serve in a way that thinks differently about an institution of learning and instruction. Let me just close with the philosophy. Tale of two Zenos. The first Zeno is a Velia. Many of you know him, 5th century Greek philosopher, he basically put a lot of these paradoxes in front of us. The most notable one was Achilles and the tortoise was captured by my professor or my teacher Mr. Allison in calculus in high school. You can leave my class at any time, but you can only go half the distance with every move. You can never get out. It ends up unattainable. So you don't bother. You just give up. You just try to polish the rock that I already have. But Zeno of Citium, founder of Stoicism, 3rd century BC, his argument is like you start with small achievable steps and with each step you learn so much that you actually lengthen the stride in the next step such that action matters more than precision. Transformation is actually realized through action. I wish you all the best in your transformation. Thank you so much. Have a great afternoon. Thank you. --- *This transcript was put together by our friend [Philippos Savvides](https://scaleu.org) from Arizona State University. The original transcript and additional summit resources are available on [GitHub](https://github.com/savvides/asu-gsv-2026-summit-intelligence). Licensed under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/).*