Beyond the Deceptive Simplicity of AI Recommendations: An Observational Framework for AI Recommendation Behavior

AI-generated recommendations often appear to be straightforward answers to consumer questions. This study examines how recommendation responses adapt as consumer information changes, using responses generated by four major AI platforms across home services recommendation scenarios. Across three independent investigations, geographic refinement, trust-oriented requests, and changing consumer decision contexts produced different patterns of behavioral adaptation within recommendation responses. These observations led to the Behavioral Influence Framework, a descriptive observational model that organizes recommendation behavior into three analytically distinct layers: Recommendation Selection, Authority Construction, and Trust Construction. The study establishes a reproducible observational methodology for examining AI recommendation behavior and provides a common analytical language for future empirical investigation.