Why do people often ignore AI features, even when they're useful?

Most people aren't looking for AI features. They're task-focused, trying to get something done and move on. When Kate Moran shared her research at Future Product Days, the top three reasons people skipped AI features were that they had no reason to use them, they didn't see them, or they didn't know how to use them. And slapping "AI-powered" on something doesn't give people a reason. Solving their problem does.

Visibility is a big one, especially for established products. When AI gets bolted onto an existing app, it competes with every other feature already vying for attention. Microsoft Edge pitches its AI capabilities alongside browsing, shopping, syncing, gaming, personalization, and more. After years of being asked to adopt all that stuff, people have learned to tune it out, and the new AI features fare the same as everything else. Familiarity makes it worse: in interfaces people use a lot, they often miss new features introduced as a single action among many others.

Then there are plain old design mistakes. Amazon's shopping assistant tested well because it knew a lot about you, but people couldn't find it, partly because it was labeled "Rufus", a name nobody connects with shopping help. A lot of these failures come down to naming, icons, and presentation, which have nothing to do with AI.

And even when people find the feature, they often don't know what to do with it. Because AI models can do so many things (but not all things equally well), most people don't know what they can do or how to prompt them. I call this capability awareness, and it's one of three issues I keep running into.

An open text field also looks like a search box, so people treat it like smarter search and type short keywords instead of giving the model the context it needs to do well.

So what works? Smaller-scope features like automatic summaries that need no user interaction perform well because they fit into existing workflows. We learned this on Ask LukeW. Suggested questions started behind a sparkle icon, but once we saw how useful they were, we generated them by default and placed them in people's gaze path so they wouldn't be missed.

The good news is that all your existing design skills still apply. Put the value where people are already looking, and remember that obvious always wins.