SimLife platform enables long-term understanding of human behavior patterns
SIMLIFE: Pattern Understanding for Long-Horizon Human-Agent Partnership
Artificial Intelligence
Summary
Understanding how people behave over days or months is hard because routines change over time. The authors created SimLife, a tool that simulates household life with detailed observations and conversations to help study these patterns. They tested current AI models on SimLife and found that the models often guess based on common patterns without truly understanding the underlying rules, and they struggle when habits change. This shows there is still a big challenge to build AI that can learn and adapt to long-term human behaviors.
What this means in practice
- •For smart home developers: Improve smart home assistants by training them to recognize and adapt to long-term user routines and changes in household behavior.
- •For robotics engineers: Create robots that can understand and assist with daily household tasks by learning behavioral rules over extended periods.
Authors
Run Peng, Zinnia Nie, Jing Ding, Yinpei Dai, Yichi Zhang, Zengqing Wu, Yao Fu, Ziqiao Ma, Jiayuan Mao, Joyce Chai
Abstract
Understanding humans over long horizons requires agents to infer not only what people need in the moment, but also how routines form, why they repeat, and when they change. We introduce SimLife, a scalable platform for simulating long-term household life with rich visual observations, ground-truth action logs, and synthetic dialogues with audio. Built on SimLife, SimLife-BP evaluates long-context pattern understanding: the ability to infer latent behavioral rules from weeks or months of everyday observations. The benchmark contains 106 episodes averaging 15.49 hours and 38.57 in-game days, and 1,439 question-answer pairs. Each task probes direct, counterfactual, noisy, and inverse reasoning under different levels of rule hints. Evaluating frontier models and architectures, we find that current models often achieve surface-level prediction without comprehensive rule understanding, rely on frequency-based heuristics rather than if-then reasoning over evidence, and struggle to adapt when behavioral patterns change. These findings suggest that long-context pattern understanding remains a major bottleneck for future embodied agents, while SimLife opens a broader space for studying memory, personalization, adaptation, and long-horizon planning in everyday human-AI interaction.