Data and Models for Behavioral AI

This unit examines data strategies and modeling approaches for behavioral AI, covering experiment design, dataset sizing, synthetic and survey data, and a spectrum of AI techniques from classical to generative. Students will learn to select and apply appropriate models, build and augment datasets, and translate outputs into actionable segmentation, intent prediction, and customer journey maps while assessing bias and ethics.

Learning Objectives

• Analyze experimental designs and dataset requirements for behavioral AI tasks, including trade-offs in data volume, how to define a Minimum Viable Dataset (MVD), and criteria for choosing real, synthetic, or survey data sources
• Apply classical AI, machine learning, deep learning, and generative model approaches to behavioral prediction and segmentation problems, justifying model selection based on data characteristics, complexity, and resource constraints
• Demonstrate methods to create, preprocess, and augment datasets (including synthetic data generation and survey-derived data), and document procedures for quality assessment, bias mitigation, and privacy protection
• Evaluate segmentation strategies beyond demographics, predict user intent, and construct customer journey maps from model outputs and experimental results, articulating uncertainty, limitations, and ethical considerations to stakeholders