Can we GenAI be a Design Collaborator?

P3: Converse to Inspire

 

Through their conversation with the AI, P3 began narrowing the design space around a specific use case problem. How could Canvas represent performance in a class with few assignments? Narrowing and framing the design problem to something more specific like this is a move more often associated with advanced designers (Tang et al. 2012). This line of thinking came as a result of the conversation with the AI, and demonstrates the power of having a conversation within design contexts in framing activities and generating creativity. 

 

P3 was an interesting case, after asking only two questions of the AI they said, “And then this is where I would stop asking it questions and then start implementing it in the designing and experimenting.” A potential to critique design ideas using AI was suggested by the interviewer, which proved to be a fruitful avenue for P3. They proceeded to investigate the specific problem of measuring performance on Canvas when an instructor has few assignments. This led to regulatory utterances like, “One thing that really like jumped up to me that I liked was the early warning system within the grade book.” and “The AI mentioned something…  it didn't really sit with me that much.” Even though they are critiquing the AI in the latter example, P3 is demonstrating self-regulation. They also moved towards a more specific design framing, “I guess now I would think of another way to phrase this question that's more specific to what I'm thinking about. So I would say something like, ‘What if the professor doesn't keep track of participation or attendance?’” Ultimately, P3 presented design ideas that directly addressed this problem, scoring high in specificity (Table 3), whereas, in independent brainstorming, their ideas were somewhat vague. 

It might be helpful to think of the process of designing as a conversation with a design problem. In this framework, we can understand how AI tools might assist students in expressing their thoughts in a design conversation. AI tools might act like an excellent conversation partner, listening and providing their own input on the topic at hand. Much as a good conversation partner might make you feel more willing to bring up thoughts that would have otherwise gone silent, the AI in this case could help novice designers learn to investigate new ideas.

 

“Creative confidence” is additionally helpful for understanding this case. IDEO founders, Tom and David Kelley said, “Creative confidence is about believing in your ability to create change in the world around you. It is the conviction that you can achieve what you set out to do” (Kelley 2013). Internal dialogue can be a mess of self-doubt and anxiety. By expressing views and ideas to an AI that speaks only kindly and helpfully, students may begin to find their creative confidence growing. 

 

This might be what happened in the case of P3, through the conversation with the AI, they felt more confident to focus on a design space they otherwise might have avoided. Because the AI represented a zero-consequence sounding board, P3 explored their unique use case. Creative confidence boosting like this would be valuable for our chatbot. We might replicate it by demonstrating true listening attributes, like looping–repeating what someone has said in a different way to show understanding–, and reactive tiles that suggest where users could go next. 

 

​

Informed design attributes from P3: Converse to Inspire

• Reactive tiles that suggest where users could go next. 

• Good listening attributes. Looping: “Is there anything more?” 

P9: Broadening Perspectives

​

"So this is a unique thing for me because I'm in a different time zone... So whenever I pull up my calendar, all of my assignments look like they're due the next day and I can't edit that. And from a user standpoint that would be really nice for me to say, okay, well, I want to edit this, or if I want to give an arbitrary due date like being able to bump that up on the calendar."

 

P9’s conversation centralized around this specific problem they encountered in Canvas. Throughout the design tasks, they moved across different scopings but continued to come back to this one issue. This interaction was noteworthy because it demonstrated a potential thought pattern of lower-experience designers and inspired the design of an introductory interaction with the AI for novices. 

 

Beginning the interaction with the AI, P9 asked what problems users are encountering with calendar events, explaining, “I’m just seeing if there were any other like problems that other users were having other than ones that I’ve already experienced.” Having validated their initial thinking and expanded the scoping of the problem, P9 then moved into an ideation space, asking for methods to smooth out the problems with calendars. Brainstorming solutions expanded the mindset of P9, looking over ideas to connect syllabus information to the calendar they thought about lecture speakers in the calendar which prompted the following: “another key thing in one of my classes, right now we had to have, like our final presentation proposal, and by a certain time so like that was something that would be good to have that pull from.” This expanded thinking also included teachers, “That could be more work for the teacher because they'd have to go in and like pull the link from the calendar and put it into the syllabus.” These two lines conjunctionally inspired P9’s final idea, a personal due date alongside the instructor’s. This solution’s complexity was a result of P9 analyzing the situation from a different perspective (pulling information from the teacher’s syllabus) and looking at the work required from the teacher's side of things. 

 

The value of P9’s interaction is best explained in their own words: “Yeah, it gives a broader perspective, cause, like, I have my limited perspective.” Perhaps the most common failing of the novice designer is assuming they are the user themselves–a narrow perspective generates narrow solutions. Through their conversation with the AI, P9 broadened the platform upon which they developed solutions. Whereas in independent brainstorming they generated solutions that would address their experience, interacting with the AI prompted ideas to address multiple. This arc of broadening perspectives is something we aim to reflect in the design of Chatbot conversations. 

 

P9 started the interaction with their own perspective, moved to enlarge it, and embraced solutions of capacious application. An introductory conversation could be designed this way to help those unfamiliar with design methodologies. This proposed conversation would start with the AI and participant finding a personal problem of the user followed by a guided process along the same path taken by P9. The ultimate goal would be to initiate designers into the process of including perspectives beyond their own, and generating ideas for those perspectives. 

 

Informed Design Attributes from P9: Initiating Designers

• Introductory conversation pattern. 

• Image/visual-based responses and user simulation with submitted images.