Eric Rodenbeck argues AI should be treated as a design medium for inquiry rather than a tool for efficiency

Rodenbeck, a lecturer in architecture at Harvard GSD and founder of Stamen Design, proposes treating AI as material to be tested and interrogated in design practice, where prompts become sketches and outputs become sites of critique.

In his course "Re-imagining the Archive" at Harvard GSD, students use large language models (AI systems that generate text) to work with museum and library collections—not to visualize or make them legible faster, but to reveal their gaps and seams as sources of meaning. One student, Roy Zhang, found that when prompting ChatGPT repeatedly with the same question, early responses were wide and exploratory, but over successive iterations the system settled toward narrower, more conventional descriptions.

Students Kevin Tang and Yuanqing Xie used AI to atomize a podcast archive from the Institute of Black Imagination by breaking conversations into sentences, analyzing their meaning, and reorganizing the archive around themes rather than episodes—treating the AI output as material to work on rather than accept as final.