- Human-AI Co-Creation
Over the past years, the role of AI systems in creative fields has made a shift from being a tool, to becoming a co-creator. There is an abundance of HCI research that covers the principles of human-AI interaction. They cover opportunities and challenges of co-creating with AI from a theoretical perspective. However, many co-creation studies are predominantly digital and quantitative in nature, often employing a lab-based methodology that fails to include the reality of co-creation as design practice. It remains unclear how practicing designers experience cocreation with AI, and how it shapes their design journeys, particularly when taking co-creation output from the digital into the physical world through textile sample making. This studio-based study investigates co-creating with ChatGPT3.5 in practice by taking a first-person perspective. A template was designed tonconduct diary studies. It focuses on the effect of natural language communication as a form of interaction and the resulting perceived dynamics of AI role-switching. The results of the study show that AI roles described in existing literature were insufficient to map AI agents’ dynamic adaptations according to the communication style, attitude, goals, and expectations of designers. Furthermore, the co-creation patterns that emerged between humans and AI in co-creation processes were much richer than expected. Findings are visualized in a co-creation protocol and a mapping of human’s and AI’s roles during design iterations. They are intended as a reference for designers who aim to collaborate with AI within the soft wearables field.