Episode cover
16 Jun 2026
38m

Simulating Humans at Scale: Simile's Joon Sung Park

Podcast cover

Training Data

Simulating human behavior through generative agents offers a transformative approach to decision-making in social sciences and corporate strategy. Joon Sung Park, founder and CEO of Simile, details how his team evolved from the "Smallville" experiment—where autonomous agents exhibited emergent social behaviors like planning parties—to a platform that models diverse human populations. By grounding large language models in real-world behavioral data, such as randomized control trials and life-story interviews, the platform addresses the "say-do" gap that limits traditional polling. These simulations allow organizations to test product concepts, marketing strategies, and complex downstream societal impacts at scale. Beyond commercial applications, this technology aims to provide a "telescope" for humanity, potentially solving fundamental mysteries in economics and collective action by modeling the irrational, subjective, and longitudinal nature of human society.

Outlines

Sign in to continue reading, translating and more.

Open full episode in Podwise