Speaking Architecture: Designing a Shared Language Between Humans and AI

Abstract

Non-human intelligence is becoming increasingly capable of generating architectural designs. The architect’s role may shift toward reviewing and refining AI-generated design proposals, possibly through written or spoken prompts. This shift calls for a revisited focus in architectural education towards learning a shared, precise vocabulary for effective co-creation with machines. We present the first phase of our ongoing effort to develop an evolving catalogue of terminology, that maps the emotional and cognitive dimensions of architecture into six categories: (1) Structural Elements, (2) Geometries, (3) Compositional Operations, (4) Building Archetypes, (5) Social Parameters, (6) The Experience of Movement, Materiality, and Proportion in Space. The study draws from a range of sources, integrating historical and contemporary frameworks from Vitruvius to parametricism. It includes the traditions of building as a craft, as well as the most recent architectural works and design methods. Research confirms the absence of a unified vocabulary suited for architectural prompting. We present an initial stage of a matrix of terms, actions, including illustrative examples, designed for both interactive digital formats and print. As the transition from hand-drawing to digital modeling was gradual, the shift toward verbal or text-based interaction with AI will take time. This requires an adaptable, evolving catalogue—now under development. As methods of human–machine interaction change, prompt engineering is emerging as a core architectural skill, ensuring we can connect AI-assisted design to the embodied experience of space in all of its aspects.

Presenters

Janne Terasvirta
Professor, Architecture and Urban Design, GUC, Egypt

Details

Presentation Type

Paper Presentation in a Themed Session

Theme

The Design of Space and Place

KEYWORDS

AI, PROMPTING, EXPERIENCING SPACE, LANGUAGE