Julian Yocum is a graduate researcher at the Center for Human-Compatible AI at UC Berkeley, where he studies brains, human and artificial. His work has revealed novel structure in neural representations, giving insight into how brains see the world. His work also approaches human values and rationality as evolving through a creative, open-ended process aimed toward higher consistency or harmony. This contrasts conventional approaches that presume consistency while permitting uncertainty. In his view, human inconsistency implies a more radical form of unknowing.
Summit 2026
Friday ·
1:30pm - 2:30pm
Salon
The Couch
Consistency has been upheld as a central principle in modern formulations of rationality. However, human inconsistency presents a more radical form of ignorance than mere uncertainty. Whereas information is the familiar antidote to uncertainty, the human inconsistency reveals a new rationality in which creativity takes center stage.
Saturday ·
7:00pm – 7:45pm
Fireside
The Couch
Researchers working on representation, multi-agent systems and evaluation on the experiments that would move the field, and the ones nobody is running.
Summit 2025
Friday ·
1:30pm
What if the real challenge isn’t teaching AI right from wrong, but designing it to carry meaning? What if AI was built not just for efficiency, but for belonging — for creativity, trust, community, and personalization we actually want? Following their provocations, we all gather in topic circles to ask why this matters, how else we might approach it, what we might do about it, and how we can communicate it to the world.