It’s something we at the IDEO Play Lab have been exploring in our ongoing co-design sessions with Gen Z, honing in on the tensions around AI’s role and personality in the realms of creativity, mental wellbeing and relationships. If anyone understands how technology changed our humanity, it’s Gen Z. They’ve already lived through the unintended consequences of technologies like social media and will likely bear the brunt of AI’s impact on their job prospects and future career growth. That’s why we’re advocating for them to take their place in developing future iterations of AI, centering their needs and values (including all the hard no’s).
However, during our most recent prototyping phase, we started running up against the technical limitations of today’s generative AI tools. We noticed that loading time between responses and unpredictable outputs were distracting from the more future-forward explorations around how AI should support our closest connections and creative pursuits.
Whenever we explain the power of design-led research and prototyping to uncover inspirational possibilities, we use this diagram:
It’s pretty easy to observe what people say and do but it’s much more challenging to elicit what people think and feel.
Therefore, to prompt users to get in touch with complex emotions around AI, we orchestrated an interactive roleplaying experience at our annual d.camp (a weeklong summer camp at IDEO to inspire teens to use design thinking in their everyday lives to solve problems).
Of course, once you’re past second grade, role-playing can be awkward. And so to get campers into the mood to play, we designed cinematic-looking scripts that recreated imaginary group scenarios and props to create a futuristic setting, including over-the-top sunglasses for the camper playing the role of the AI.
Each discussion group (four to six campers) received two versions of a script, depicting the same overall scenario, for example: “You are planning a trip with friends.” Within each group, the two scripts showcased different roles that an AI concierge would play in the scenario varying from super friendly to super professional, efficient to user-focused, and more.
After playing through the original scripts, we paused to listen to campers surface their feelings and reflect on what felt fun and helpful, but also importantly, what moments in the interactions felt icky.
Afterward, campers jumped in to re-write their scripts to imagine their versions of how AI concierges would interact with users, with no fear about breaking any fancy tech or already-polished interfaces.
By acting out the scripts in a playful way, our campers and researchers were able to bring to life the discomfort of the interpersonal dynamics. In receiving permission to let loose and to fully embody the different roles in the chat scenarios, campers’ attitudes shifted to engage not just their intellect, but also their emotions.
Our campers flew through multiple versions of scripts in under an hour, producing clear opinions on nuanced elements of AI design that can be hard to articulate — even for seasoned designers. Because while most of us may theoretically support an AI-enabled concierge that can help you plan vacations with a group of friends or family — how do you define the limits of what it should know, and where it should intervene? For instance, how might AI show up in moments of tension or conflict — such as helping to navigate group friction around which restaurant or hotel to pick on your trip?
While these emotional insights were invaluable for direction, we were most delighted to see a cohort of high schoolers engaged in co-design, excited by the prospect that AI-powered tools aren’t a new technology that we “just have to deal with.”
We believe we must always start with and return to people throughout the design process.
For research and design managers building GenAI-enabled products, where the technology is deeply involved in crafting personalized interactions, roleplaying offers an agile solution to make tangible the most core experiential parts of an app’s personality or positioning. It serves as a sandbox for rapid learning, iterative exploration, and unhindered ideation before sinking dollars into developing platforms and tools to support.
Our exploration continues this Fall so stay tuned for more content on responsible AI innovation principles and behind-the-scenes looks at our research methods!
Want to try out this playful prototyping method? Reach out to [email protected] to access a free editable script and worksheet template.
As part of IDEO's Design for Play team -- the consulting arm of IDEO's 20+ year old toy invention studio -- Keren supports partners to connect the dots in applying the principles of Play not only to kids and toys, but also to today's toughest challenges, from finance to health, mobility, education, and more. Prior to IDEO, she launched two organizations in the education sector, collaborating with education bureaus, principals, teachers, parents and students across Greater China and the U.S. to innovate on approaches to building English literacy. She is an alum of World Economic Forum Global Shapers Community (2014-2019). She holds a B.S. in Biology & Society and a minor in Education from Cornell University and an M.P.A. from Tsinghua University where she was a Beijing Government Scholar with a focus on Urbanization.
Tweet
Share
Post
Share
Email
Print
Ideas In Brief
The article explores IDEO Play Lab’s innovative approach to understanding Gen Z’s perspectives on AI through roleplaying, shedding light on the emotional nuances and experiential aspects of AI design.
Your Design System Works in Figma. Does It Work in Code?
The article explains why many design systems don’t work well: designs made in Figma don’t translate well into code.
It introduces five practices: structure frames like code, use fewer components with more variants, organize by how both designers and developers actually work, let AI check your naming, and build documentation into your daily workflow.
The piece says that good design systems are the same in design and development, and when they match, everything just works.
