PIBBSS fellowship final presentation. Preserving human power is increasingly proposed as an AI objective, but how “power” is formalised determines what maximising it actually does. These slides relate two existing measures, show their failure modes, and argue that Heitzig and Potham’s framework can measure collective power when the model encodes organisation. I also discuss two diagnostics for who holds power and how much of it others can withdraw.
You may navigate the slides by clicking on it or using arrows once active, or open the slides fullscreen. You can watch me present these slides on YouTube.
Presented at the PIBBSS 2026 summer symposium. See also: the rest of my PIBBSS outputs.