Affordances of Large Language Models in Design Activity

Abstract Large language models and AI tools such as ChatGPT have possible benefits within design process and design activity across design courses in higher education. With the advent and rise in use of large language models (LLM's) we seek to better understand the effects of these tools. This technology has widespread application throughout the STEM field from writing/correcting code in any language to brainstorming ideas for the next big project, or even producing fully written research papers (although not this one). However, LLM's are not well understood and while many students hail them as a quick way to finish homework assignments, there are perceptions from students for both the positive and negative roles there may be in learning and understanding core concepts. This paper investigates the use of this technology and implications in and for the design process. In particular, large language models may be utilized to aid in generative forms of brainstorming as well as convergent synthesis of ideas, adding to its value in supporting design work. Additionally, the viewpoints of students who utilize LLM's and those who do not will be explored to see how they use LLM's to aid them. Policies pertaining to LLM's will be discussed to see if they align with student uses and what limitations are set. This includes interviews with students about how LLM's are commonly used. While many people in academia are working on how to limit the use of these tools as they can be seen as either 'cheating' or a 'short cut' this paper aims to investigate the possible benefits of this technology in the design process as well as to imagine how we might live with its effects in the future. We aim to fill this gap and discuss ways to improve the usefulness of LLM's as well as to successfully integrate it into project expectations, as we consider the future of design learning and practice.