MS Slide Designer: A Study on Human-AI Collaboration for Content Creation

Human-AI collaboration can be used to improve the competency of people in the creative content space. Nevertheless, users struggle to understand what strategies can lead to a better partnership that can help them in creating aesthetic, creative, and coherent content. As part of our study, participants are requested to create slide presentations and asked to use the Microsoft Slide Designer AI tool as part of their design process. Thereafter, our team conducts comprehensive user-consented interviews to understand the users' experience with the tool, learn the kind of AI suggestions users would accept, and acquire knowledge about AI teaching good slide design. The study highlights that MS Slide Designer performs quite well when it deals with different forms of text data such as recognizing the context of words or recommending aesthetic themes for title or subtitle slides. However, the user's control over the presentation process is limited, resulting in a highly iterative slide creation process. This limitation becomes evident through user experiences, such as the user's inability to revert to previous slide versions after selecting a design, and the lack of consistent theme maintenance for the presentation. Based on the findings collected through the user interviews conducted, our research suggests strategies through which humans can perform more effective human-AI collaboration in the creative content space. One of our strategies highlights the need for MS Designer to be more interpretable seen through a new feature suggestion asking for tags with each design recommendation.