We’ve already made a start at using AI agents to review our code for accessibility issues, and to help us fix them then and there - so what if we took this concept and started to apply it within our design tools, our research tools, and even our operational tools? Just recently I reviewed some written ideas for digital experiences with Gemini. It evaluated the ideas against dark UX patterns, assessed inclusivity and accessibility gaps, looked at psychological impacts and then made recommendations for more ethical UX. We were then quickly able to work through the issues and improve our ideas. Using AI enhanced our thinking and helped us to progress the work before going on to test it out with a diverse range of people.
More complex tasks such as reviewing visual design and interaction patterns with AI are trickier but not impossible, and tools are advancing all the time.
If we remain open-minded to the possibility that we may have unintentionally designed something that makes people feel unwelcome, and then embrace new ways of spotting those issues - it feels right to consider AI powered tools as just one of the ways we could do better.
I do have a number of concerns however - can AI really represent people from all different backgrounds and in all situations? The answer surely has to be ‘No’. Plus there’s the risk of giving us a false sense of security - if the AI says there’s no issues then it must be fine - this rings alarm bells to me. Traditional inclusive design methods such as involving relevant communities in the design process simply cannot be replicated by machines. And giving people ownership over the design of something they will be using cannot be faked.
So it would be easy to come to the conclusion that AI and Inclusive Design are at odds with each other and should never combine forces. However, I think it’s worth considering that given the speed of AI, and the ease of use of AI powered tools, to not use them at all feels like such a missed opportunity. We’ve seen first hand the value these tools can bring and even if a tool catches just one issue, that’s one more issue caught than otherwise.
Maybe the answer is more about identifying specific areas where AI can truly add value, areas where it does not distract or misdirect us. Plus, AI tools can also live alongside best practice methods, enhancing our human-led expertise. On another recent project we tested some communications with a panel of real customers and conducted an AI powered audit. This combination allowed us to feel confident that the insights we were gaining were both human and at scale.
It’s about getting the balance right and supercharging inclusive design in this new era of technology. We can use AI to help us spot issues, but not to sign off on inclusion. We can use it to support inclusive design, but not to replace the people who need to be part of it.
