The field of human behavior modeling is experiencing a significant shift with the emergence of large language models (LLMs) and foundation models. These models have demonstrated remarkable abilities in understanding and generating complex data, and are being applied to various aspects of human behavior, including travel behavior, user behavior, and decision-making. Researchers are leveraging these models to infer sociodemographic attributes, predict behaviors, and generate insights about contexts. The use of LLMs is also addressing the limitations of traditional statistical and machine learning approaches, which often require large sample sizes and strict statistical assumptions. Overall, the field is moving towards more behaviorally grounded and data-driven approaches, with a focus on interpretability and scalability. Noteworthy papers include:

• BehaveGPT, which introduces a foundational model for large-scale user behavior prediction, achieving over 10% improvement in macro and weighted recall.
• Be.FM, which presents one of the first open foundation models for human behavior modeling, demonstrating its capabilities in predicting behaviors and inferring characteristics of individuals and populations.