How I use AI in UX
AI can’t do everything. At least not yet. What AI does well in the UX space is shorten timelines. It enables UX experts to:
Conduct research to validate assumptions
Analyze workflow designs
Ideate wireframe solutions to workflow problem areas
Identify patterns in customer feedback
Evaluate real-world usage data to prioritize areas for improvement
The thing I’ve found most fascinating in my usage of AI over the past few years is that the results you get depend entirely on how you engineer the prompt. While large language models (LLMs) have access to mountains of data, they’re still inherently dumb until you ask them a question. And depending on how you compose your question, the LLM may either shower you with shiny gems of useful data, or spew back a load of hallucinated malarky. To ensure the former, you have to define your desired outcomes.
This dovetails with the approach that any solid UX expert would take. That is:
First, clearly articulate the problem to be solved.
Then, ideate solutions to that problem.
I also occasionally use AI to create images or to help me write me some code for a prototype or micro-interaction. I don’t yet trust AI to write copy as it tends to get verbose – also it uses too many en-dashes. 😜
