This week’s convo features Ismail Kurun, a fellow at VALIANT, Vanderbilt University’s interdisciplinary AI lab. Kurun works at the nexus of AI and PPE (philosophy, politics, and economics) and is a Paul Lewis Fellow at the Mercatus Center. You can follow his work on X and PhilPeople.
How did you get into emerging tech? What are you working on now?
I have followed emerging tech for a long time. I was blogging about cryptocurrencies back in 2014 because I was interested in blockchain’s decentralized architecture. Then I got interested in AI through my work in political philosophy and mainline political economy. My research investigates how societies sustain or develop formal and informal liberal institutions, particularly epistemically. The transition to the AI economy is putting pressure on these institutions. In 2024, I joined VALIANT to work on this.
I am currently most concerned with markets and the practices of public justification that support democratic decision-making. My recent paper on AI and markets contributes to the economic calculation debate. I ask whether AI systems can reproduce the epistemic functions of markets. And I argue that entrepreneurial discovery in particular resists computational replication. It depends on embodied, historically situated judgment rather than computation alone.
What work of art has most shaped your views on emerging tech?
Jules Verne’s books. I read them when I was a kid. They tied technology and adventure together in my mind at a young age and made me curious about the future in a way I never really outgrew.
What’s your most contrarian take on AI or emerging tech policy?
My most contrarian take on AI is that the loudest disagreement in AI is less fundamental than it looks.
Accelerationists and existential risk theorists disagree about whether the arrival of AI systems matching or exceeding human capability will go well, and about how fast we should move toward it. But they share the assumption that competence is separable from the kind of being that has it.
But researchers working on embodied AI and world models argue that grounding in sensorimotor experience is necessary for general competence, that you cannot achieve general competence through disembodied prediction alone, however much text you train on.
They treat the body as an engineering problem. If you build the sensors and run the robot, then it will be grounded. But that solves only half the problem. You can build sensors, but how do you build judgment?
Entrepreneurial judgment is a case in point. It is tacit, something you must live through. Judgment and even knowing what’s worth attending to both come about when people with skin in the game have decision-making power and when dissent is tolerated. Judgment withers when these conditions are missing.
The rules that matter here are not only the formal ones but the informal —— commercial norms, trust, and the shared expectations that make formal rules work at all. These are learned through participation, which is why transplanted institutions often fail to take root.
So, questions about AI capability become questions of institutional survival. And if AI drives down the cost of execution while leaving judgment where it is, orders that are good at producing judgment (i.e., open societies) should see their advantage over others widen. But the same technology strains those orders, so the outcome depends on whether we can renew them faster than AI erodes them.
What worries me in the meantime is a version of what Hayek called the pretence of knowledge: the assumption that because AI is impressive, we already know which human functions it can replace.
To be clear, I’m optimistic and excited about AI’s potential. But my worry is that, because of that assumption, we hollow out the institutions that have produced so much human progress, including AI, over the past few centuries.
What are you reading right now?
First, there is the technical literature on world models, Yann LeCun’s joint embedding predictive architecture (JEPA) line in particular, and on the vision-language-action (VLA) policies coming out of robotics. This is one direction the economic calculation debate may go next.
Then there are the interpretive and cultural foundations of market processes. I want to understand how growing up in different commercial cultures might change how a person sees business opportunities. So, I am reading Virgil Storr’s Understanding the Culture of Markets and Don Lavoie’s essays on hermeneutics.
Both strands of reading bring me back to the same questions. What kinds of knowledge can be automated? What kinds are generated through participation in markets and other liberal institutions? And what happens to those institutions when AI is introduced to them?
Go-to emerging tech music track?
Well, I like ethnic, religious, or otherwise culturally significant music, much like Tyler Cowen. I have a large collection of world music —— from Balkan Roma and the muwashshahat of Arab Andalusia to Greek Orthodox hymns and Vedic chants.
There is a great deal to be gained from “cracking cultural codes,” from immersing yourself in unfamiliar forms of life to understand and appreciate the aspirations of different cultures.
I have talked about this in more detail in an interview for the American Philosophical Association’s blog. Liberalism has drawn on a humanistic tradition concerned with understanding cultures from within. This is a line running from Alexander von Humboldt through Isaiah Berlin. I share that spirit.
Edited for length and clarity, with hyperlinks added.