As large language models (LLMs) become faster and cheaper, many marketing and insights teams are testing them as stand-ins for real consumers, promising quicker insights at lower cost. But do these “synthetic consumers” actually behave like humans? 

In this webinar, researchers evaluate leading LLMs using a well-established decision task and find a critical disconnect between human-like language and human-like decision making. Even with advanced techniques like prompt engineering and fine-tuning, LLMs consistently fail to reproduce key patterns of human behavior, and often in unpredictable ways driven by small changes in prompts or framing. 

Join this session to learn when LLM-based research can add value, where it introduces hidden risk, and how to use AI responsibly without replacing the human insight your decisions depend on.