A Carnegie Fellowship and a Year (ish) of College Azimuth

Happy Monday everyone. It's been a little bit and I promise it's been because I've been working on a few things I think people will find interesting.

The big thing to announce is that I've started a Senior Fellowship at the Carnegie Foundation for the Advancement of Teaching.¹

The scope is "Study how AI systems inform college decision-making and shape the propagation of the Student Access and Earnings Classification, and how Carnegie can better leverage data infrastructure to support opportunity colleges".

So what does that mean and how did this all happen?

The Carnegie Foundation for those maybe unfamiliar is the place (amongst a lot of other things) that classifies higher education institutions in the United States (these days together with the American Council on Education).² If you've heard the phrase "R1 Research institution" or something of the like then you're implicitly talking about Carnegie. It's run by Tim Knowles, its president, a man of many hats and someone I've known for about a decade now.

One of the big initiatives of the foundation under his tenure has been a reframing of sorts, of how it thinks about its classifications to focus on the work institutions do to enhance social economic mobility.³ The 2025 Student Access and Earnings Classification is the concrete version of that: it recognizes 478 Opportunity Colleges and Universities for enrolling students who reflect the communities they serve and for graduates who go on to earn more than their regional economies would predict.⁴ I've of course, talked about this in passing before and it's clearly been a big influence on a lot of the work I've done here at Azimuth and how I think about things.

It's no small endeavor to change how we think about the role, and the value of college institutions. Is it purely academic research oriented? Is it about educating a population, or is it a means to get people into good jobs to contribute to the economy? Or some mixture of the above.

I have my own views on that informed by my own experiences, and things I care about. Others of course will have theirs but it's notable for an institution with the history, prestige and platform of a place like Carnegie to openly change. The next question of course being does that change how anyone else thinks about these things?

Relatedly, we're more or less at a year since I started working on "College Azimuth" and it's worth regrouping a bit tot think about where we started, and potentially where this is going in the context of the above.

As I shared early on, this was mostly a way for me to get some of the ROI analysis I did for my last company out there in the public domain because hopefully someone could find it useful. That appears to be the case! As I've written my thoughts on a lot of things I've had a lot people from various domains reach out to share thoughts, notes or ideas on the work, and the data. It's been very fun in that regard.

One of the largest threads has been how this all intersects with the emergence of consumer AI applications that are increasingly stepping into the role of de-facto college advisor, well meaning mentor, or even financial planner.⁵ EAB found that the share of college-bound students using generative AI in their search went from about a quarter in spring 2025 to nearly half by the end of the year.⁶ So as you can probably surmise things have shifted a bit from simply "talk about ROI" to figure out how people are talking about ROI and helping them do it better.

This in turn led to some big realizations about AI, data, and the world in which we live in which I think inform how I have been thinking about things in the last few months, and a few of the things I've been working on.

In no particular order:

• AI is not some blob that just benevolently or comprehensively absorbs information, and how it is displayed. All of the conversations we have about "What's the role of AI", "What's the ideal return on investment" etc etc are being picked up by various Ais, in various ways with no real rhyme or reason to it (as far as I can tell).
  
• AI is fundamentally changing how people engage with information. This is both a risk, and an opportunity. The risk in my mind being that we just passively sit back and let AI be something that happens to us, and the opportunity being that we can take an active role in ensuring it has the data, information, frameworks to be net positive, while also having the evaluative framework to identify and remediate when it isn't.
  
• There's a necessary role to be filled outlining the acceptable bounds of AI use, outputs and how it interacts with people in various contexts. A lot of the attention in this space has been around things like mental health (or health care in general), privacy, et al but I'd posit that the same thing applies to when someone is talking to it about making some of the most consequential choices someone can ever make e.g. what college to go to, what career to pursue, how to pay for it.

So what do I think this means in the context of the work I am doing, and this fellowship with Carnegie? It's a few things:

• How do we evaluate what AI is, and is not saying about colleges, and career decisions.
• Does it know about the classifications, what does It know? Does it apply it consistently?
• To the extent the classifications are about helping highlight institutions doing the Yeomans work of being an economic engine for the United States, and helping people understand how they did (so they can do it to) how we think about being proactive about collecting that data, curating it, and making sure its available publicly in a way that reaches the right people in a world where the conduit is increasingly AI.
• How do we then hold it accountable when these gravitate to other polarities that may or may not be harmful or unproductive.

These are some of the questions I've been working on, and some of the things I've been doing over the past few months. I'll be sharing some more of my notes on this going forward, and more importantly some of the ideas and the work I am doing in response to it.

Carnegie changed how it thinks about what a college is for. I'm curious to see what everyone else changes, including the machines people now ask. I look forward to exploring this with you all.

1. Carnegie Foundation for the Advancement of Teaching, Our Team (senior fellows): https://www.carnegiefoundation.org/about-us/our-team/ ↩
2. Carnegie Classifications, 2025 methodology (administered with the American Council on Education): https://carnegieclassifications.acenet.edu/carnegie-classification/classification-methodology/ ↩
3. Katherine Mangan, "New Carnegie Classification Will Reflect Social and Economic Mobility," The Chronicle of Higher Education, Feb. 9, 2022: https://www.chronicle.com/article/new-carnegie-classification-will-reflect-social-and-economic-mobility (subscription). Tim Knowles and ACE's Ted Mitchell on the plan, Inside Higher Ed podcast, episode 71: https://insidehighered.podbean.com/e/ep-71-injecting-social-mobility-into-the-carnegie-classifications ↩
4. American Council on Education and the Carnegie Foundation, "Carnegie Classifications Redesign Better Reflects the Higher Education Landscape and Recognizes Student Success," April 24, 2025: https://carnegiefoundation.org/newsroom/news-releases/carnegie-classifications-redesign-better-reflects-the-higher-education-landscape-and-recognizes-student-success. Methodology for the Student Access and Earnings Classification: https://carnegieclassifications.iu.edu/carnegie-classification/classification-methodology/2025-student-access-and-earnings-classification ↩
5. Chalkbeat New York, "On-demand college counseling, courtesy of AI," March 27, 2026: https://www.chalkbeat.org/newyork/2026/03/27/high-schools-embrace-ai-tool-for-college-counseling-help/. Times Higher Education, "AI visibility key as students turn to ChatGPT for college searches," Feb. 2026: https://www.timeshighereducation.com/node/741387 ↩
6. EAB, survey of more than 5,000 high school students on AI in the college search, Feb. 2026: https://eab.com/about/newsroom/press/ai-in-college-search-survey/. The same survey found 18 percent of students had removed a college from consideration based on AI search results. ↩

Further reading

• Erika Tyagi, Kristin Blagg and Emily Gutierrez, Urban Institute, Evaluating How LLMs Handle Education and Workforce Pathway Data (May 2026) and Improving How LLMs Handle Education and Workforce Pathway Data (October 2026).
• The Chronicle of Higher Education, Can't Get ChatGPT to Cite Your College? Try These Tactics (subscription).
• Carnegie Foundation and ACE, Opportunity Colleges and Universities profile series, for example the University of Illinois Chicago profile.