Ideation with Generative AI—in Consumer Research and Beyond

Looking for a guide on how to use large language models (LLMs) for ideation? Check out our new article in the Journal of Consumer Research, “Ideation with Generative AI—in Consumer Research and Beyond.” Whether you’re a marketing scholar, a behavioral scientist, or just curious about how generative AI tools can help spark original thinking, this paper offers a roadmap for doing exactly that.

LLMs are already transforming many stages of the research process—from analyzing data to
writing papers. But their role in the earliest and most creatively demanding stage—ideation—hasn’t been fully explored. This article steps into that gap. The central question
is simple: Can LLMs help researchers generate better, more original ideas? And if so, how?

Drawing on decades of psychological research on human creativity, the authors argue that LLMs mirror the two main cognitive strategies we rely on to think creatively: persistence (deeply exploring a focused space of ideas) and flexibility (connecting distant concepts in unexpected ways). LLMs can mimic both. They’re capable of producing a high volume of coherent ideas (persistence), and they can draw on a massive breadth of knowledge to suggest novel combinations (flexibility).

But LLMs tend to generate what the authors call “small ideas”—useful but incremental. They’re less effective at producing “big ideas,” or breakthrough insights that upend existing thinking. To tap their full potential, the authors recommend using LLMs not as a substitute for human creativity, but as co-creators—tools that help humans think better, broader, and deeper.

The paper introduces a practical toolkit for using LLMs in ideation. For instance, you can
increase productivity by prompting the model with a few great examples (few-shot prompting),customizing it to your topic (fine-tuning), or connecting it with external databases (retrieval-augmented generation). You can boost semantic breadth—the diversity of topics the LLM draws from—by varying your prompts, using persona-based framing (e.g., “Imagine Steve Jobs is answering this”), or guiding the model step-by-step (chain-of-thought prompting).

Still, there are tradeoffs. While individual users may get more creative ideas, the collective
originality of a field may decline if everyone starts using the same models in similar ways. The paper warns that, over time, this could create a kind of AI echo chamber, where we all start generating—and publishing—slightly different versions of the same ideas.

To help researchers avoid this trap, the paper introduces a set of metaphorical ideation roles that LLMs can play in the ideation process:

• As Designers, LLMs can generate diverse and balanced experimental materials,
  improving internal validity.
• As Writers, they can express ideas more clearly and persuasively—important in a field
  where creativity is judged by other humans.
• As Interviewers, they can ask probing, Socratic questions to help humans discover their
  own ideas.
• As Actors, they can roleplay consumers in qualitative research to help researchers
  anticipate reactions, find new segments, or surface overlooked hypotheses.

Each of these roles harnesses the strengths of LLMs while preserving the essential contributions of human creativity. The authors emphasize that using LLMs wisely means knowing both when to guide generative AI and when to let it guide your own imagination.

Looking ahead, the paper raises provocative questions. Could LLMs one day help researchers not only generate ideas, but also select the most promising ones—maybe even predict which papers will have the greatest real-world impact? Might they help the field of consumer research break out of its paradigmatic grooves and tackle long-neglected but socially important questions?

Ultimately, “Ideation with Generative AI” is a call for thoughtful experimentation. Rather than letting AI homogenize our thinking, the authors show how to use it to stretch our creative muscles—and perhaps even see our work with new eyes.

Read the full paper:
De Freitas, J., Nave, G., & Puntoni, S. (2025). Ideation with Generative AI—in Consumer Research and Beyond. Journal of Consumer Research, 52(1), 18-31.