AI as Method, User & Critic: An Interior Urbanism Studio

"Context: a senior INT studio exploring interior privately-owned public spaces (INT POPS) in New York City.

Method: We have documented 22 case studies in the establishment of transferable AI-supported workflows as applied to research, conceptual development, presentation and assessment. These applications can facilitate engagement with the political and social dimensions of interior urbanism in academic settings. AI (including Claude, Midjourney, RunwayML, and HeyGen) was integrated across three studio projects, using an iterative process across platforms to both emulate and enhance the design process. Insights gleaned: prompt engineering shares key characteristics with a robust design methodology.

User: Students leveraged AI to craft optimized survey questions for visitors to POPS sites as part of their initial analysis, then fed those responses into Claude (an LLM) for analysis. Using photographs of site visitors, Claude generated user profiles. These AI-generated personas—complete with backgrounds, preferences, and specific spatial needs—enabled students to test their concepts acknowledging multiple user perspectives rather than defaulting to their own worldviews. Insights gleaned: the Claude-generated profiles most fundamentally represented the middle class, a data-driven bias.

Critic: The critics developed an AI-assisted process to incorporate visiting critics’ remarks into their assessments, hoping to check their own methods for bias. Insights gleaned: this promoted rigor regarding stated studio goals, but the critics’ feedback did not significantly challenge those of the professors themselves.

Our transparent documentation of both successes and limitations provides a critical foundation for further exploration of AI's role in design education and in particular, the design of public spaces.

team: Karin Tehve & Joette Jones (visiting assistant professor, INT)"