Subvolts take Simulating human society requires a new AI approach beyond raw intelligence, says Simile's founder. Jun 16, 2026 - Developers
- Learn about alternative AI architectures focused on simulating human behavior and values.
- Business
- Discover how simulating human preferences can be used for concept testing and market analysis.
This episode features a founder discussing a novel AI approach that simulates human society, offering a different perspective from current frontier lab developments.
Simulating Humans at Scale: Simile's Joon Sung Park
Audio streams from Sequoia Capital. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
Sequoia Capital's Training Data podcast features Joon Sung Park, founder of Simile. Park discusses his work on simulating human society at scale, contrasting current AI models' rational capabilities with his approach to encoding human values and preferences. He highlights Simile's success in predicting behavior and its potential applications, from concept testing to modeling societal dynamics.
This episode explores building AI that acts like people, not just solves problems. It explains how simulating 1,000 people's behavior accurately shows a new way to understand human choices and societal trends.
This episode offers insight into a novel approach to AI development, moving beyond pure problem-solving to simulate human behavior. It's for leaders and researchers interested in the future of AI, particularly those exploring how to imbue models with human-like understanding of values and preferences. This contrasts with the current frontier labs' focus on raw intelligence, suggesting a complementary path for AI development that could impact areas like market research and societal risk analysis.
- HostSequoia Capital
- GuestJoon Sung Parkfounder and CEO, Simile
- Guest: Joon Sung Park, founder and CEO of Simile
- Title: Simulating Humans at Scale: Simile's Joon Sung Park
- Show: Training Data (an official AI show)
- Length: 39 minutes
- Simulation Accuracy: 85% as accurate as people reproduce their own answers
- Simulated Population: 1,000 Americans
The race to build superintelligence is producing models that keep getting better at objective problems, but not at behaving like actual people. Joon Sung Park, founder and CEO of Simile and creator of Stanford's "Smallville" generative agents study, argues that simulating human society requires a fundamentally different kind of model. He frames today's frontier models as the "CPU of intelligence"—rational, superhuman at problems with right answers—and Simile as creating the "GPU of intelligence," built to encode the diversity of people's values, preferences, and tastes. It simulated 1,000 Americans and predicted their behavior 85% as accurately as people reproduce their own answers. CVS uses it for concept testing; some customers simulate their own earnings calls. Joon's larger bet: a "CERN of human society" that could one day model bank runs, climate cooperation, or the early signals of a collapsing democracy.
Hosted by Sonya Huang, Sequoia Capital
- generative agents
- Joon Sung Park
- AI programs designed to act like people
- CPU of intelligence
- Joon Sung Park
- AI that is good at rational problems with clear answers
- GPU of intelligence
- Joon Sung Park
- AI designed to understand diverse human values and preferences
1 h 08David Silver and Lasse Espeholt of Ineffable Intelligence: A Superlearner for the Age of Experience
WithDavid Silver, Lasse Espeholt
1 h 041 h 0552 minMaking Cities Awesome: Peregrine’s Nick Noone & Ben Rudolph
WithNick Noone, Ben Rudolph
1 h 12Yossi Matias on the golden age of research
WithYossi Matias
42 minGemini co-leads on project origins and what's next
WithJeff Dean, Koray Kavukcuoglu, Noam Shazeer, Oriol Vinyals
32 minGoogle I/O 2026 Recap with Logan Kilpatrick, Josh Woodward and Tulsee Doshi
WithJosh Woodward, Tulsee Doshi
31 minQuestions people ask
- What is Simile's approach to AI?
- Simile focuses on creating AI models that encode the diversity of people's values, preferences, and tastes, aiming to simulate human society rather than just solve objective problems.
- How accurate is Simile's simulation?
- Simile simulated 1,000 Americans and predicted their behavior with 85% accuracy compared to how people reproduce their own answers.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Sequoia Capital.