A Deep Dive into Synthetic Panels: The Case Against Them

Synthetic respondent panels promise faster, cheaper sample at the scale researchers have always wanted. But what are you actually buying when you purchase AI-generated data, and is it appropriate for market research insights?

In this session, Josh Speyer offers a technical breakdown of how synthetic panels work, without the marketing hype. Drawing on a background in AI engineering and market research, he walks through the technical realities, math, and outcomes of using respondent models for consumer and market insights.

It covers the core problems with synthetic respondent data: the black box problem inherit in LLM-based panels, the illusion of sample precision, the false individuality of AI-generated respondents, and what widespread adoption could mean for the insights profession.