Design Approaches to Creativity Support with Embedded Artificial Intelligence

This research investigates designers' attitudes and approaches towards using AI (Artificial Intelligence) to support their creative work. It consists of three studies with people working in creative roles in the design industry: a survey (n=45), a month-long diary study (n=30), and a 21-day digital probe study (n=5). Mixed-methods data analysis identified several factors that influenced participants’ preferences for the type and level of creativity support they desired for a particular task, and their willingness to accept support from an AI system. These factors were found to divide into three groups: Categories, Confines, and Competencies of support.

Three Categories of creative support were requested by participants: Information support, related to receiving the necessary data, references, or feedback need to complete a creative task; Generation support, related to direct help with the tools and processes of generating creative outcomes; and Situation support, related to organising and facilitating working environments, schedules and conditions for creativity. Of these, Information support was the most frequently requested.

The Confines of support related to the participants’ distinction between creative tasks which they considered of personal value, and which they were less likely to share with an AI system, and tasks which were not considered of personal value. This was found to relate to the perceived originality and creativity of the task experience. The Competencies of support related to the participants’ perception of the knowledge and abilities required to support a task, and how this related to their own knowledge and abilities. Participants were more likely to fully delegate tasks to AI that they already had experience of completing themselves, and preferred to work directly on tasks that were new to them.

These factors were tested across the different studies, and formalised into a Creativity Support Framework, which forms part of the contribution of this research, along with the design and implementation of an embedded AI digital research probe, used in the final study.