AI in UX Design: Where It Makes a Real Difference

Picture a designer at the start of a new project. A blank canvas. A folder of user interviews nobody has had time to read. A developer waiting on a handoff. And a constant drumbeat telling them AI is about to change everything.

So what is really happening where AI and UX meet? That question drives Marginally Better, the latest episode of SKOP, the SkillUp Online Podcast. It follows a design project from the first user interview to the final handoff, to find out where AI is genuinely helping designers today, where it still falls short, and what that means for anyone wondering whether to start learning it.

Two verdicts on AI and UX design

In May 2025, Nielsen Norman Group, one of the most respected names in usability research, published its latest review of AI design tools. The title said it all: AI Design Tools Are Marginally Better. The tools had improved since the year before, but they still couldn’t match the quality of a human designer, and they were nowhere near what had been promised.

Just weeks later, the State of AI in Design report from Designer Fund and Foundation Capital surveyed 400 designers. Of those, 89% said AI had improved their design process in the past year.

It is tempting to ask which one is right. But look closer, and they are answering different questions.

Nielsen Norman measured the tools against the promise: could AI do a designer’s job? Designers measured them against their own week: did AI make some part of the work easier? Both answers can be true at the same time. In fact, Nielsen Norman found that the tools designers actually use every day are narrow, each doing one small task well. That is exactly the kind of help a working designer would notice.

Where AI is already making a difference

Research is one of the clearest examples. In User Interviews’ 2025 State of User Research report, published in September 2025, 80% of researchers said they use AI, up 24 points in a single year. Research platforms such as Dovetail now transcribe interviews, pull out key quotes and surface themes across sessions automatically. Hours that once went into transcripts and sticky notes can go into the part that needs a human: deciding what the findings mean.

Design itself is changing too. In March 2026, Google rebuilt its Stitch tool around what it calls “vibe design”: an infinite canvas, voice conversations with a design agent, and screens that link together into a clickable prototype. It is an experiment from Google Labs, free to try, which means the barrier to exploring it is curiosity, not budget.

Neither tool replaces design judgment. Both remove some of the slow, repetitive work around it.

The part the episode saves for you

There is a third shift in the episode, and it may be the most surprising. It involves the line between the person who designs a product and the person who builds it, and one number that nearly doubled in a single year.

Then there is the bigger question. If AI is only marginally better, what do designers actually need to do about it? Retrain from scratch? Wait for the hype to settle? The answer in the episode is far smaller than either, and far more encouraging.