The annual Information Architecture Conference (IAC) brings together hundreds of IA and UX professionals to discuss key topics of the day and progress the field through educational presentation and workshops. This year’s conference tackled AI (of course), but also centered on the continued importance of understanding humans. While I presented a session on including a broad range of humanity throughout the product design lifecycle, I also had the pleasure of hanging out with industry icons and learning how they are thinking about this topic.
IAs are no longer ignoring AI
While past conference publications and discussions primarily centered on the shortcomings of machine learning and large language models for auto-classification of information, this year many attendees could speak first-hand on how they were using automation and AI tools to improve the experiences they were building.
Most retained a healthy skepticism and emphasized the need to get beyond “the hype,” but there was a growing curiosity — and some practical examples — of how to leverage large language models (LLMs) to do some of the more tedious tasks such as information gathering, sorting, tagging, etc. A favorite quote, oft repeated, was the wish that AI could be trained to do the dishes and laundry and leave art and creative writing to humans, who enjoy it (when they have time). I left asking myself, “Can I identify the ‘dishes and laundry’ equivalent for any users which I’m currently designing for?” For example, could automating tedious tasks — like finding the right form and transferring information from this PDF to that form — free up clinic staff for more meaningful interactions with patients and providers?"
I left asking myself “Can I identify the ‘dishes and laundry’ equivalent for any users which I’m currently designing for?”
The pressure to leverage AI solutions was also a common point of discussion. Three presenters shared helpful case studies of working with leaders who wanted them to provide AI-powered experiences. By grounding leaders in 1) the real needs of humans, 2) articulately sharing the boundaries that humans had, and 3) demonstrating the boundaries of the technology, they found smart (and quite lightweight) ways to weave a small bit of AI in where it actually made sense. For example, structured, thoughtfully presented information about hours for different campus libraries solved a UX problem much better (and cheaper) than an AI chatbot. The chatbot gave misinformation at times and competed with the “ask a librarian” chat feature, which was staffed by a human to answer student and faculty research questions.
Inclusivity matters more now than ever
Accessibility of information and inclusivity of design are foundational to our IA practice, but I heard more and more people talking about the limitations of our traditional methods for getting input from all types of users.
Many folks, myself included, are exploring new methods and re-doubling our efforts to achieve more inclusive designs. We are still learning how to not be "the heroes" and how to really share space and co-create with other constituents. What I love is all the people I met who are sharing freely what is working for them in terms of finding contributors from diverse backgrounds and allowing them to participate in ways that foster authentic participation. Writing inclusive screener questions, having people with disabilities test experiences for accessibility, and asking yourself who you might be unintentionally excluding were topics of discussion.
This ongoing theme around inclusivity happened despite of — or in some cases in response to — some practitioners who currently must navigate doing what's right for all users despite the removal of support for long-held accessibility standards. Others face the challenge of how to advocate for equity and inclusivity in a moment where speaking openly about those topics is (formally and informally) suppressed.
Could AI actually help with Inclusive Design?
I initially walked away thinking about all I had learned about these two themes, but as I reflected more, I started to also see an important intersection between them.
AI can actually help provide insights into what people from hard-to-recruit groups might think or need. It’s certainly no replacement for truly inclusive recruiting, but it could potentially be a stepping stone to further the case for inclusive research or help shape up the areas to explore with different groups.
I had a chance to try out a prompt engineering technique called “panel of experts” — when you instruct an LLM like GPT4 to represent multiple people using the backgrounds you provide. You can then ask it what it “thinks” about different topics or ideas, and have it discuss multiple viewpoints for your benefit.
In addition, one presenter shared an intriguing case study on using an LLM to improve the employee review experience. The LLM could review feedback a colleague or manager had drafted for an individual and point out potential biases that had crept into their thinking or writing — such as commenting on someone's "youthful behavior" or a woman’s cheerfulness or assertiveness.
AI can actually help provide insights into what people from hard-to-recruit groups might think or need. It’s certainly no replacement for truly inclusive recruiting, but it could potentially be a stepping stone to further the case for inclusive research or help shape up the areas to explore with different groups.
Conclusion
Time together with other practitioners is always energizing for me. I left this particular conference impressed by the ways people are finding to stick to our convictions about staying human-centered, show the value of our work, and support each other and our users during this inflection point in our practice.
If you’re interested in learning more about my presentation on Designing Safe, Usable, and Inclusive Medical Devices or want to chat about an IA and AI-powered project, drop us a line at info@consultfathom.com!
