Designing inclusively with AI
Summary
We know there’s massive opportunity for AI to be used for greater speed, growth, and insights, but how can we avoid the pitfalls? Head of Accessibility Innovation at Fable Kate Kalcevich will share how to best leverage current AI tools in your day-to-day work with inclusion in mind.
Key Insights
-
•
AI-generated images of people with disabilities often show racial and representation biases, such as all-white subjects and inaccurate assistive device depictions.
-
•
Disability representation in AI images is frequently inaccurate, e.g., guide dogs shown with leashes instead of harnesses and blindfolds over eyes.
-
•
AI-generated personas for users with disabilities can contain improbable feature combinations, such as a MacBook user reliant on Windows-only assistive tech like JAWS.
-
•
Personas created by AI lack the depth and nuance gained from real user research and testing, especially for disability accessibility considerations.
-
•
AI-generated mockups can have accessibility issues like poor contrast and unlabeled UI elements, requiring human accessibility review.
-
•
AI tools like Stark Sidekick improve design accessibility through automated issue detection, though they are not yet full replacements for experts.
-
•
AI-powered task management apps can enhance productivity and accommodation for neurodivergent users by organizing schedules and reducing cognitive load.
-
•
AI can help identify potential biases in accessibility automation projects, but requires careful prompt crafting and domain expertise.
-
•
Designing AI-powered interfaces comes with inclusion challenges, as users unfamiliar with prompting may struggle to get effective results.
-
•
Combining AI output with continuous user testing, especially involving people with disabilities, is essential to ensure accessibility and inclusion.
Notable Quotes
"If you have questions, I will take questions throughout."
"AI-generated images often have no racial diversity and show inaccurate assistive device representation."
"You are either navigating with a white cane or a service dog because you are not gonna want both of your hands full."
"Personas tend to not give you those deep insights that you would get from user research and user testing."
"Automation might not be able to capture the context of an image if you're trying to describe an image."
"AI is actually a pretty good productivity hack as a starting point for accessibility reviews."
"A lot of people will try a prompt and not get what they were hoping to get and maybe miss out on all the benefits of AI."
"For people who identify as autistic, AI tools that review writing tone can be especially helpful."
"If the code doesn’t match the label on screen, voice controls might not recognize the command."
"There always has to be a human option if you’re not getting what you need from the AI."
Or choose a question:
More Videos
"We have never played together before and never rehearsed, yet we pulled off a great rendition spontaneously."
Jim KalbachJazz Improvisation as a Model for Team Collaboration
November 6, 2017
"Inclusion, collaboration, and iteration are the three pillars of how we work with authors to make a book."
Louis RosenfeldCoffee with Lou: Should You Write a (UX) Book?
March 7, 2024
"The difference between director and principal often lies in whether you’re inspired more by leading people or focused on outputs and craft."
Catt Small Micah Bennett Brian Carr Jessica HarlleeWhat's Next for ICs: Exploring Staff and Principal Designer Roles
February 22, 2024
"Less than a third of respondents feel like they work somewhere with a clear investment in customer-driven decision making."
Marieke McCloskeyUser Science: Product Analytics & User Research
March 11, 2021
"DeepFakes can look too realistic and might introduce new privacy risks if donor faces come from real people."
Llewyn Paine[Demo] Deploying AI doppelgangers to de-identify user research recordings
June 5, 2024
"You don’t want to overdesign just to make something more data interpretable because that can lead you down a dark path."
Joshua NobleCasual Inference
October 6, 2023
"Stakeholders are users too; the product we’re sharing is the research and learnings for their decision-making."
Sara LogelYour Colleagues are Your Users Too
March 29, 2023
"We'll have a centralized Slack channel for all discussion to come together as one community."
Bria Alexander Louis RosenfeldWelcome
January 8, 2024
"Consistency is so important that sometimes even consistency in failure works if it means I only have to learn the workaround once."
Sam ProulxOnline Shopping: Designing an Accessible Experience
June 7, 2023
Latest Books All books
Dig deeper with the Rosenbot
How effective are AI-generated follow-up questions in usability testing compared to human moderators?
Why is prior authorization such a persistent problem and how can patient-centered solutions be designed around it?
What design strategies can support community-led rebuilding of maternal healthcare in rural maternity care deserts?