Open innovation when the crowd is a model
Open innovation and crowdsourcing appear on 71 recent papers, mostly in management, but only 3 of them carry an AI keyword. Firms ran idea contests and crowdsourcing platforms to get ideas from outside. If a language model can generate thousands of ideas in an hour, what is the crowd still for, and how do firms tell good ideas from fluent ones?
What prompted this#
Since October 2025, 71 papers carry an open-innovation or crowdsourcing keyword (39 in management journals, 11 in economics): "open innovation" on 42, "crowdsourcing" on 13, "open innovation ecosystems" on 5. Only 3 of the 71 also carry an artificial-intelligence, generative-AI, language-model or machine-learning keyword.
Questions#
- In idea contests, how do ideas from language models compare with crowd ideas on novelty, feasibility and what firms actually implement?
- Do participants in crowdsourcing platforms now submit AI-generated ideas, and does that raise the volume and lower the diversity of submissions?
- Does the value of open innovation shift from generating ideas to testing them with outside users?
Data that could help#
Submissions and outcomes from innovation contests and platforms, with timestamps around 2023, and experiments that mix model-generated and human ideas for blind evaluation.
Next step#
Run a blind evaluation in which managers rate a mix of crowd-sourced and model-generated ideas for a real problem, and track which ones they implement.