Exploring Fantasy vs Reality in AI
An account of the Service Design Tools & AI x Design Community Workshop
_________________________________________________________________
PSA: It is a key value for us at AIxDESIGN to open-source our work and research. The forced paywalls here have led us to stop using Medium so while you can still read the article below, future writings & resources will be published on other platforms. Learn more at aixdesign.co or come hang with us on any of our other channels. Hope to see you there 👋
The Research: Service Design & AI
At Service Design Tools, our design researchers have been exploring the context surrounding the design of AI-based services and uncovering methods to assist designers while maintaining a human-centered approach.
In interviewing designers, expectations became a major recurring theme that highlighted the common misconceptions surrounding AI capabilities, namely, that many people outside of the data/engineering field tend to severely overestimate the abilities of AI.
Stakeholders who are unfamiliar with the actual capabilities of Artificial Intelligence often think that AI is somehow magic — that it can do anything, and a lot faster, more accurate, and a lot ‘cheaper’ than a human can.
When working with AI, these fantasies can lead to false expectations and unrealistic projects. We were interested in exploring this friction, understand these expectations’ nature, and uncover best practices to navigate it effectively. To that end, we thought to crowd-source this exploration and organized a collaborative workshop surrounding this topic and our current findings. You can view the deck that includes the insights we shared during the workshop here.
By joining forces with the AIxDesign community, we were able to further enrich our research with more industry insights and experiences in a dynamic way.
The workshop: How to align on expectations?
On July 2nd, we facilitated a virtual brainstorm workshop with a focus on learning about different stakeholders’ assumptions when working on projects involving AI. Mapping both common and unique experiences with differing expectations, our research aims to create the basis for an expectation framework that will be published and made available on ServiceDesignTools.org in the coming weeks.
Intro
We kicked off the workshop by sharing some of our SDT team’s current research objectives and insights, followed by a quick locating icebreaker and introduced what we call the networking ocean (“Baby Shark” not included).
Fantasy vs. Reality
We started capturing people’s experiences by asking them to reflect on specific moments where “fantasies” (false expectations) showed up in AI projects, and connect them to a corresponding post-it where they wrote the “reality” of the situation. We mapped their experiences across three basic phases (Define, Build, and Implement), and made a distinction between expectations of clients and expectations within the team.
Bridging the gaps
In search of best practices, we picked out some of the challenges on the board and invited participants to elaborate. A range of people weighed in. The discussions that followed were a combination of humorously recounting each other’s struggles, sharing stories of best practices, and ideating future experiments to try out.
Closing
While we probably could have talked for hours, we were nearing the 60-minute mark, so we wrapped up with feedback and goodbyes.
Workshop Take-Aways: Lofty mental models & dumb data
Reflecting on the session, some of the takeaways were:
- The ease and effectiveness of building an AI is commonly overestimated by all stakeholders in the team as well as by clients.
- Emphasizing the importance and limitations of available data as early as possible to avoid delays is extremely important, as is including data scientists and ML engineers in meetings from the start.
- The need for a common language amongst designers and engineers, and the importance of learning the basics about each other’s practice in order to establish empathy and common ground.
- Setting concrete objectives and definitions of success with a client early on, in order to provide a valuable output and use time efficiently. Using an AI-specific value proposition template or worksheet (like the one here) could help.
Reflection: Next steps for the SDT team
The workshop gave us a better idea of the most frequent expectations and came across some we hadn’t seen before. If you’d like to take a look for yourself, you can check out the Miro board here.
As mentioned above, this workshop feeds into the larger research project focusing on how to better support Service Designers when designing AI-based services — perhaps for the first time. Once we’ve completed this guide, we will share it on ServiceDesignTools.org and add it here.
Thanks Renae Elkassih, Mariana Escalante, Eugenia Bifulco, and Elisa Palmieri — Service Design Masters students conducting research at Oblo and ServiceDesignTools — for hosting the workshop. Thanks to the AIxDesign Community for the opportunity and providing the platform.
About AIxDesign
AIxDesign is a place to unite practitioners and evolve practices at the intersection of AI/ML and design. Currently, we are organizing monthly virtual events (like this one), sharing content, exploring collaborative projects, and looking to developing fruitful partnerships.
To stay in the loop, follow us on Instagram, Linkedin, or subscribe to our monthly newsletter to capture it all in your inbox.
