Fellow Q&A: Logan Whitehair

This is part of our fellow Q&A series to give prospective applicants and other interested audiences a glimpse at what it’s like to be a Horizon Fellow.

What path brought you to the Horizon Fellowship, and what made you apply?

Before joining the fellowship, I was working primarily in AI policy. I was very interested in AI policy generally, but I wanted to get more into the national security space and explore the intersection of AI, emerging technology, national security, and defense.

I think the structure of the Horizon Fellowship gives you significant freedom to explore whatever intersection of niche topics you’re interested in. There aren’t many fellowships, at least to my knowledge, that can say the same. I would say that is what drew me to the fellowship in the first place.

Speaking of those niches, what did you concretely work on during your fellowship placement, and what did the role involve in practice?

A common conversation in D.C. during my fellowship was about understanding the value of basic research investments for national defense. If we make these investments in basic research, are they actually as valuable as making the same investment in applied research, or in research later in the research cycle that would more quickly create tangible benefits?

A confident and well-connected peer told me that, to their knowledge, no one had been able to show the empirical value of basic research investments. Frankly, those were my marching orders early in my fellowship.

I quickly realized that there was a considerable amount of open-source data on research grants and contracts, and then, separately, on patents. But these sources were fundamentally unconnected. They existed in multiple places, and people frequently analyzed one or the other, but not both at the same time.

So during my fellowship, I built an open-source data pipeline that normalizes data on research grants and connects it to patent data. In aggregate, you can use it to analyze the outcomes of basic research investments in the defense sector. Which universities do defense agencies frequently fund? Which companies frequently cite those universities’ papers in their patents?

I also published a companion paper with my colleague at the time, Dr. Evan Johnson, a professor at UNC-Chapel Hill. The paper was published at the Naval Postgraduate School’s Acquisition Research Symposium.

What were the most interesting findings that you drew from all of that data?

One interesting finding, which is probably obvious to a lot of people, is that these top defense research agencies frequently fund top universities. We probably know intuitively that a handful of universities do most of the basic research in the United States, and the data showed that empirically.

What was more interesting to me was what happened after that research was published. We found that the path from basic research to technology was not really linear. It was not a case where one university produces a piece of research, one company picks it up, and that eventually becomes a defense technology. Instead, it looked much more like a network. Individual universities were producing research that was later cited across patents from many different companies, and individual companies were drawing on research from many different universities. That reinforced the idea that basic research is less like the first stage of a pipeline and more like a foundation of knowledge that gets recombined across the broader innovation ecosystem.

We also saw meaningful differences between the research agencies in how widely their work seemed to diffuse into patented technology. DARPA stood out in particular. Its research received substantially more downstream patent citations, even after accounting for the amount of research it funded. There are a few possible explanations for that. DARPA has a different mission and funds work that can sit somewhat closer to application than some of the service research offices, but it was still striking to see that difference emerge quantitatively.

What are you working on now? How, if at all, did your Horizon Fellowship experience help shape the path you took after the fellowship?

I’m currently at a company called Air. We’re a defense software company, and we build advanced AI systems for military and defense customers to close the enterprise readiness gap. Enterprise readiness is fundamentally the capability and capacity to meet enterprise objectives during both steady-state and dynamic operations.

One of the surprising things about my current job is the level of complexity public servants are responsible for working with. If you were to feed this complexity directly into a foundation model, you wouldn’t get the outcomes you want. These models lack the context or skills needed to tackle these challenges.

During my fellowship, one reason I became interested in the implementation side of the defense sector was that I realized the data and intelligence available to decision-makers were sorely lacking. We have access to incredible amounts of data and, now, with foundation models, an incredible amount of intelligence. But they aren’t being leveraged in the best ways and often aren’t being used for the important problems and decisions that decision-makers are tasked with tackling. That’s what first made me interested in this work of creating more specialized models for national security applications.

Going back to your experience before the fellowship, what initially drew you to these overlapping fields of emerging technology, open models, and national security?

When I was younger, I was always into technology and, as silly as it may sound, certain simulation and grand-strategy video games. Those games sparked my interest in using computational methods to understand or model economics or complex social systems.

After graduating, I wasn’t sure where you could do that, because it’s so weird. The Horizon Fellowship was unique in providing the resources and opportunity to follow my interests wherever they took me.

You also had some previous research experience at a few think tanks and academic institutions. What was your focus at that point, and how have your research interests changed since then, or stayed the same?

I was at the Abundance Institute in the lead-up to the 2024 primary elections. Generative AI was also coming onto the scene, so there was concern about how deepfakes would affect the information ecosystem surrounding the election.

I developed a fairly rudimentary web-scraping and machine-learning tool that collected data on how the media was talking about deepfakes and on instances in which deepfakes were used in legitimate media relating to the election or the candidates. I then wrote a series of blog posts based on that data. The data is also open source and available online.

One thread I’m picking up here is that your work has involved a lot of comfort with data and analysis. How do you feel you’ve been able to use your particular skills, and is there any advice you would give people hoping to do similar, data-focused work?

I would just say that you don’t have to be a good programmer. I’m not a good programmer myself. If I were, I probably would have been a software engineer.

You don’t have to be a good programmer; you just need some familiarity and comfort with the tools. Especially now, the tools available to the public enable a lot. But you do have to have a good idea of what you’re researching or trying to analyze.

If you’re interested in using data, I would also say that many people would be surprised by how much even a little data can strengthen and justify your research and analysis.

When you first came to D.C. and started working in the emerging technology policy space, was there anything that particularly surprised you?

One surprising thing was coming here young and not really having any idea what I was getting myself into. It’s easy to assume that everyone here has been here longer, is more experienced than you, has it all figured out, and knows exactly what they’re doing. Frankly, that’s frequently not the case.

In one of the orientation classes before the fellowship began, I think Remco Zwetsloot showed an image that still resonates with me. It depicted a cycle of learning and knowledge: when you first start learning something, you’re pretty certain you have a good understanding of it. But then you learn more, and you start to realize just how much you don’t understand.

Even among very intelligent and knowledgeable people in D.C., there’s still a lot that they recognize they don’t know. There’s always space to contribute.

What advice would you give someone currently considering applying for the Horizon Fellowship?

I would tell everyone to lean into what makes you unique, whether that’s your interests or your skills, and truly try to have agency.

Horizon will enable you to pursue those niche interests. When you focus on something that few others focus on, it will be valuable to somebody.