Two new ways to see your university
The first AI instruments arrive
Two fun websites launched this year to compare how seriously American colleges and universities are responding to AI. Scott Latham’s AI Campus Index (headlined “Is Your University AI-Ready?”) ranks colleges and universities along multiple domains from the classroom to launching graduates. Kyle Saunders’s Mapping the Structural Divide (headlined “Where does your institution stand?”) maps 1,556 four-year institutions on two axes: resilience and market position. Check them both out and then come back here.
I’d seen a few scattered rankings of AI degree programs and universities doing research on AI previously, but as everyone knows, a tiny part of AI on campus is about students and faculty studying AI. AI is everywhere, from student affairs chatbots to enrollment management data scrapers, to faculty and students using LLMs in general education courses. Both Scott’s and Kyle’s sites are headlined with questions. Everyone is wondering whether their institution is ahead or behind.
Thinking about how to make a whole campus AI-literate (let alone AI-competent) is a challenge, which is why a few months ago I wrote a template job description for a university AI czar. The job involves both building capacity and contracting offices that may need one-third the staff in the AI era. So much is going to change.
I still see a lot of flailing among higher ed leaders who are treating AI as something to be “adopted” and “implemented” rather than understood. If the top priority for a university is knowledge—its creation, transmission, and conservation—then a new technology that transforms how humans can engage with knowledge must be central. AI should make institutions more productive, which is different from making them more efficient.
Scott Latham, a professor of strategy at the Manning School of Business at UMass Lowell, developed AI Campus Index with Michael Braun, recently of the University of Montana, and Gage Light of Stakforge. I first started following Scott after reading his breathtaking April 2025 Chronicle of Higher Education piece “Are You Ready for the AI University?” (subtitled “Everything is about to change”). “The big mistake faculty members make is underestimating the existential threat AI represents to their livelihoods. Professors need to dispense with the delusional belief that AI can’t do their job,” Scott wrote. “Early and mid-career professors who hope to survive will need to adapt and learn how to work with AI. They will need to immerse themselves in research on AI and pedagogy and understand its effect on the classroom. How does AI alter how we teach engineering? Or pursue basic science?” Nobody has written a better piece yet.
Scott also predicted that accreditors and U.S. News & World Report would begin factoring AI capability into their assessments, foreseeing “the emergence of the AI haves and have-nots.” AI Campus Index focuses on six domains: the classroom (from AI major to AI gen ed requirements), AI in campus life (from accessibility to advising), AI in operations (from admissions to financial aid), AI in governance (ethics, board of trustees, privacy), AI in research (new faculty, commercialization), and workforce readiness (career outcomes, industry partnerships). You can plug your institution in and see how it’s doing.
Right now, Carnegie Mellon, MIT, and Stanford are topping the list, with Berkeley and UIUC close behind.
Kyle Saunders, a professor of political science at Colorado State University, announced his map in March 2026 and it’s terrific: eight indicators drawn from public sources like IPEDS, the College Scorecard, O*NET, WICHE projections, and Anthropic’s Economic Index. Kyle then maps 1,556 four-year institutions. You plug a school in and it shows up as a green dot (high capacity), an orange dot (structurally exposed), a red dot (high stress), or a blue dot (market misaligned). Kyle’s working paper “Mapping the Structural Divide: Institutional Resilience, Post-College Market Position, and Artificial Intelligence Exposure Across U.S. Higher Education” is also available here.
Latham’s site seems like it will be more useful over time, as universities begin to provide more student access, do more faculty training, bring AI into operations, and show that governance is addressing things like privacy and ethics. The danger is that Latham’s site is more hackable by university PR offices: it is constantly scanning and will pick up press releases trumpeting AI initiatives that sound really great but aren’t backed up by real investment. I have confidence that the site’s mechanisms will see through the fluff over time, but I suspect with the institutional interest in AI Campus Index, Scott and his team will get a lot of lobbying. I’m glad it’s being run by people with integrity.
Saunders’s site measures current strengths and position with public data. One axis charts endowment per student, revenue diversification, enrollment trajectory, and selectivity; the other charts completion rates, earnings relative to debt, regional demographics, and graduates’ AI task exposure. I tested the map with St. John’s College, which came out “high capacity,” which seems right but in a different way than R1 universities (82 percent of which are “high capacity”). St. John’s may not ever use this capacity, but if it wanted to become an AI Great Books campus, it could. The point of Mapping the Structural Divide is to look at resilience: who must pivot and who can afford to wait and watch (or opt out).
The biggest challenge I see on campuses today is lack of leadership and lack of messaging about why knowledge about AI is critical. What careers are most at risk and which offer the most opportunities?1 An AI czar needs to develop capacity in AI literacy, understood as the ability to ask what a model was trained on, whether it is open or closed source, how it should be used. This is why I wrote my recent piece on inference chips. Faculty need to understand clearly the gap between using a tool that produces fluent output and understanding what that output contains. Every AI czar should have both tabs open.2 Latham’s is for provosts wondering if they’re doing enough; Saunders’s is for people who may want to jump ship and wonder if where they’re interviewing is going to survive.
Neither site is meant for families and future applicants. I expect U.S. News will include something soon, and they may draw on both Latham’s and Saunders’s sites. It’s unclear to me what students are going to want in an AI metric. Some may look at academic integrity metrics. Some may wonder if there will be courses on how to use AI, though I suspect most students will be better users than many faculty. They may wonder if courses adapted for frontier-model agent use are available. Prospective math majors may want to know how the curriculum has changed to incorporate mathematical proof discoveries by AI, and methodologies to solve more difficult problems. Prospective humanities majors may be interested in courses on knowledge and ethics. Prospective arts majors may be interested in both analog courses and high-tech ones.
Inside most universities there’s still chaos and confusion. The state of affairs reported in the recent Chronicle piece, “The Rise – and Fall? – of the AI Czar,” has not changed much. In this climate, it is good to have places to compare institutions.
Future instruments may address questions like whether a university has completely revamped a curriculum, including requirements, to ensure students are ready for the AI world. Every existing curriculum, assessment, and workflow was designed before the AI era. Everything should be rethought. I am impressed Latham’s site includes campus operations, since AI offers new opportunities to think operationally across divisions. In the meantime, universities may start with press releases, and that may be okay.
I’ve seen two new papers that measure AI exposure. Jacob Light, an economist at the Hoover Institution, sees exposure in statistics and data science, English, and computer science, but minimal exposure in the skilled trades and the clinical fields. Evidence Matters’s Jennifer Steele and Isabella Cruz average the exposure projections of seven economic models, seeing strength in occupations pairing above-median salaries with below-median exposure: healthcare practice, public safety, construction and extraction, maintenance and repair. Their graphs are interactive in a similar way to Latham's and Saunders's.
There are a handful of other new sites but all are in beta and none do anything close to Scott’s or Kyle’s sites. https://www.airedex.ai, https://www.irex.org/universityaireadiness · https://www.digitaleducationcouncil.com/resource-library-items/ai-in-higher-education-global-survey-2026 · https://thinklytics.com/insights/2026-higher-ed-ai-readiness-map
"I’d seen a few scattered rankings of AI degree programs and universities doing research on AI previously, but as everyone knows, a tiny part of AI on campus is about students and faculty studying AI. AI is everywhere, from student affairs chatbots to enrollment management data scrapers, to faculty and students using LLMs in general education courses."
YES. And beyond.
We've been living with a wide variety of mechanical, electrical, and electronic devices for decades to centuries, depending on the device. All of us use and interact with some of these devices, but most of us know relatively little to nothing about how they work. However the culture has developed common sense ways of thinking about and interacting with these devices which are sufficient for most of us most of the time. These concepts allow us to use these devices effectively and to interact with experts when we need their help.
We need to develop common sense means of dealing with AI. The AI concepts and practices needed by students in general education courses need to spread to the general population – obviously, when those gen ed students graduate they'll be part of that general population. And then there's AI concepts and practices for primary and secondary schools, which need to be developed in and taught in colleges and universities. This amounts to a massive population-wide effort to reconstruct our common sense view of the world.
Within the college and university system, someone has to read and think deeply about a little book that John von Neumann wrote while on his deathbed, "The Computer and the Brain," which he didn't complete. It was published posthumously in 1958. You might think that we've learned so much about the brain since then and computers have changed so much that this book from seven decades ago is obsolete.
Wrong.
It's about how computation can be implemented in physical stuff. That's where the problem begins. Now, yes, the book needs updating in various ways. But you can't begin to see that until you've understood what von Neumann was saying seven decades ago.
https://www.linkedin.com/posts/jim-hutchins-26a97413_thanks-to-hollis-robbins-substack-anecdotal-share-7491242059752833024-pxye/?utm_source=share&utm_medium=member_desktop&rcm=ACoAAALP73IBY2IoVY9NflqHYPSiZdDBDQqMji8