Democracy can be seen as an equilibrium, sustained by specific structural conditions. This project investigates how advances in AI could shift the conditions that sustain democratic governance, and aims to produce concrete forecasts for these dynamics.

We start by studying the "democracy equilibrium": what structural factors cause democracies to emerge or fade? This involves drawing on political science frameworks like selectorate theory, as popularized by CGP Grey's "The Rules for Rulers" (https://www.youtube.com/watch?v=rStL7niR7gs), and on concrete historical case studies. We try to really understand why we are currently in a democracy and create a model where we can quantify the stability of a democracy based on measurable variables. Then analyze how AI might change those. For example:

(1) the productivity of the economy depends less on the wealth of the population,

(2) concentration of wealth,

(3) superpropaganda, where AI-powered social media platforms or their future analogs shape public discourse, or

(4) stronger military-coercive imbalance.

Optionally, participants may

1) also try to derive recommendations for how democratic accountability can be preserved, and/or

2) try to figure out how to communicate this risk here well.

The project outputs are

(1) a research report and

(2) either a recorded talk, a video, or an accessible blog post communicating the key findings to a broader audience.

A lot of the work will go into deriving the general model, but then we want to apply it in more detail to at least one of US, Europe or Germany. I will also encourage mentees to try to get in touch with academics whose research is about history of democracy or so, because I think this is often a fast way to learn.