Reflection in co-action: AI joins the crit

This perspective explores how generative AI introduces new themes in architectural studio pedagogy and the meaning of “learning by doing”. Grounded in a reflection-in-action approach to design pedagogy, this perspective argues that studio work is increasingly shaped by interaction with AI systems, making “learning by co-doing” a useful lens for examining human–AI reflective practice in architectural education. The argument is supported by a scoping review in the Web of Science Core Collection, followed by title/abstract screening, full-text review, and finally, thematic exploration of 16 studies directly exploring GenAI and/or LLMs within the pedagogical framework of architectural education. Accordingly, four meta-themes were extracted through thematic synthesis: (1) phase-dependent integration across the design process; (2) necessity of new studio literacies in prompting, evaluation, and translation; (3) pedagogical scaffolds and formats that make human-AI collaboration teachable; and (4) governance questions around authorship, assessment integrity, bias, ethics, and boundaries. From this perspective, Schön’s paradigm may require new reinterpretation: “learning by doing” is increasingly enacted as “learning by co-doing,” in which agency and authorship are negotiated and design reasoning is mediated through ongoing interaction with AI. Ultimately, this perspective argues that the future of architectural education depends on cultivating a reflective praxis that negotiates the benefits and limitations of AI, while sustaining the central role of studio as a site of invention, dialogue, and ethical engagement.