‘Simulate’ to ‘stimulate’: generative AI for causal and anticipatory service research

Service research faces a temporal paradox: the futures most needing investigation often emerge faster than the methods available to study them. A central constraint is methodological lag, reflected in the limited capacity to create controlled variation in rapidly evolving service contexts. This paper treats Generative AI (GenAI) as a governed infrastructure for stimulus design in causal service research. It introduces CAUSAL (Conceptualize, Align, Utilize, Scrutinize, Assure, Legitimize), a six-phase protocol specifying how GenAI can responsibly generate, verify, and disclose research stimuli under controlled variation. CAUSAL distinguishes two pathways aligned with causal identification logic. In experimental designs, GenAI supports identification by generating parameterized, multimodal stimuli that vary only in focal elements while minimizing and empirically assessing non-focal variation. In quasi-experimental and observational designs, GenAI supports measurement and stimulus standardization, reducing noise without creating identification, which remains a function of research design and random assignment. The paper integrates GenAI into stimulus design while preserving the distinction between stimulus control and causal identification. The protocol provides actionable guidance, reporting standards, and decision rules that lower barriers to multimodal stimulus construction. Finally, the paper clarifies the inferential boundaries of anticipatory service research, identifying when stimulus sophistication enhances behavioral realism and when it amplifies speculative responding.