8 BILION DIGITAL CLONES

2026-02-15 · Source: Wes Roth · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Emerging Technologies & Innovation · Depth: Intermediate, long

Summary

Simile, a new venture by Stanford researcher Jun Park, is expanding on the "interactive simulacra" concept to create large-scale digital societies for simulating human behavior. Building on the original experiment that used 25 LLM agents to simulate a village's social dynamics, Simile aims to model entire demographics based on real-world data like transcripts and transaction logs. This platform allows users to ask specific questions about societal reactions to events such as tax changes, marketing campaigns, or news dissemination. The project has secured a $100 million seed round with notable investors including Andrej Karpathy, Fei-Fei Li, and Adam D'Angelo, and is already working with enterprise clients like CVS Health and Telstra. Simile's predictive accuracy has been demonstrated in forecasting analyst questions during simulated earnings calls, achieving 85% accuracy, suggesting a potential shift from "big data" to "big simulation" for market research and social science applications.

Key takeaway

For AI Product Managers or strategists evaluating market interventions, Simile offers a powerful sandbox to test hypotheses without real-world costs. You can run thousands of simulations to identify optimal strategies and anticipate unintended consequences, significantly reducing the "innovation tax." This capability allows you to explore niche reactions and emergent behaviors that traditional statistical analysis might miss, providing a more robust understanding of potential outcomes before deployment.

Key insights

Simile creates large-scale digital societies using LLMs to simulate human behavior and predict real-world outcomes.

Principles

• Simulations can predict social dynamics.
• Memory streams enhance agent capabilities.
• Simulations can reveal idiosyncratic behaviors.

Method

Populate digital environments with LLM agents based on demographic data, assign backstories and schedules, then introduce variables (e.g., news, campaigns) to observe emergent social interactions and information flow.

In practice

• Test marketing campaigns on simulated populations.
• Predict analyst questions for earnings calls.
• Analyze public response to economic shocks.

Topics

• Digital Societies
• Large Language Models
• Predictive Simulation
• AI Agents
• Market Research

Best for: Investor, Executive, AI Scientist, AI Researcher, Data Scientist, AI Product Manager

Related on AIssential

• Farewell to Paper People: From Human Simulation to Human Simulator — LLM agents must evolve from shallow persona simulation to robust, runnable human simulators with persistent states and population alignment.
• The Doomers were always WRONG! — AI alignment is a co-evolutionary process driven by market selection, not solely an engineering problem.
• What Can You Learn from a Deterministically Chaotic Market Simulation? — Agent-based market simulations can reveal emergent financial phenomena like the co-existence of value and momentum strategies.
• SimGym: A Framework for A/B Test Simulation in E-Commerce with Traffic-Grounded VLM Agents — SimGym enables rapid, risk-free A/B testing of e-commerce UI changes using VLM agents grounded in real traffic data.
• Improve Your Agentic AI Trading With a Great Data Pipeline — High-quality, diverse data pipelines are crucial for successful agentic AI trading, outweighing model sophistication.
• The first ‘AI societies’ are taking shape: how human-like are they? — AI societies and digital twins offer novel ways to simulate and study complex human social dynamics.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Wes Roth.