Micro-foundations of Effective Curatorial Practice in Human–AI Collaboration

Identify the specific cognitive and behavioral techniques that constitute effective human curation in AI-mediated knowledge work with large language models, including which prompting strategies, review protocols, and collaborative workflows most effectively transform users from passive consumers into active curators.

Background

The paper argues that generative AI has dual effects: it homogenizes information via an AI Prism while also enabling recombinant innovation through a Paradoxical Bridge, conditional on active human curation. The proposed contingency model hinges on curatorial practice—human phronesis guiding and vetting AI-generated content—to achieve positive epistemic outcomes.

In the conclusion, the authors emphasize that while curation is the decisive variable, the concrete practices that make curation effective remain insufficiently characterized. They call for experimental and observational research to determine which human-AI interaction patterns (e.g., prompting strategies, review protocols, collaboration workflows) best convert passive acceptance of AI outputs into active, critical curation.