In May, Professor Muthukrishna gave a talk titled “Cultural Evolution and Alignment” at the Post-AGI Workshop: Economics, Culture and Governance hosted at Lighthaven in Berkeley, California.
In his presentation, he drew on the Cultural Brain Hypothesis and human evolutionary history to argue that we have already lived through a cognitive-outsourcing event: human brains shrank in volume over the Holocene as cognition moved into externalized stores, institutions, and distributed networks. He argued that AGI represents a phase transition because tools do not deliberate, but AGIs act as self-replicating agents subject to multi-agent evolutionary selection regimes.
Rather than focusing solely on aligning individual AI models, he emphasized that civilizational alignment is a property of the selection landscape itself. Preserving human flourishing requires either maintaining human utility within competing AGI lineages or engineering a new cultural-overlay equivalent for AI populations.
Sincere thanks to the organizers, David Duvenaud, Raymond Douglas, and Jan Kulveit, for putting together such an interdisciplinary and forward-looking workshop!












