Can AI Become the New Design Buddy for Architects?

New artificial intelligences that generate images are reshaping how architecture and interior design students work. In a recent study, 42 students took part in hands‑on workshops with visual‑generation tools and then answered a questionnaire. Most showed enthusiasm: AI tools are seen as a valuable aid in the early stages of a project, when experimenting with shapes, colors, and atmospheres. Students also noted that AI can boost creativity and give a competitive edge in the future job market. However, their acceptance drops as the design progresses, because originality and critical judgment become more crucial. From an ethical standpoint, students consider AI use acceptable as long as its presence is openly disclosed, avoiding plagiarism or misleading attribution. The findings suggest universities should embed AI into design curricula, but with clear guidelines on when and how to use it, balancing innovation with creative integrity. This way, students can harness AI’s potential without losing the ability to think and design independently.

Image‑generation tools built on deep‑learning models such as DALL·E, Stable Diffusion, or Midjourney rely on convolutional neural networks trained on massive image‑text datasets. The diffusion process starts from random noise and iteratively reconstructs an image that matches the given prompt, leveraging the model’s learned semantic links between words and pixels. Pedagogically, integrating these tools can enhance design thinking by enabling rapid visual prototyping, shortening the gap between conceptual ideas and concrete representation. Yet, reliance on AI raises originality concerns: models tend to remix patterns already present in their training data, potentially limiting genuine innovation if students do not develop a critical eye to assess the novelty of generated outputs. Ethically, transparency is essential: crediting AI as a co‑author prevents copyright infringement and the misrepresentation of AI‑generated work as entirely personal. Moreover, dataset bias can reproduce cultural or aesthetic stereotypes, influencing students’ design choices. To mitigate these risks, institutions should introduce training modules that teach how to interpret, contextualize, and edit AI‑generated images, fostering a “co‑creation” practice where artificial intelligence serves as a tool rather than a replacement for design thinking.