AI in UX Research and Design – A personal reflection

As a UX researcher and designer, I find myself reflecting on how AI has transformed everything from how we analyse user behaviour to how we sketch out new interfaces. In this post, I’ll share my first-hand perspective on using AI in digital product design, what’s working, what’s on the horizon, and where human judgment still reigns supreme. My goal is to provide an honest and practical examination of the opportunities and challenges that AI presents to UX research and design.

AI-Powered User Research: Understanding User Behaviour at Scale

In user research, AI has become an incredible accelerator. I remember when analysing user behaviour meant manually watching countless session recordings and painstakingly noting patterns. Today, much of that heavy lifting is automated. Platforms like Hotjar, for instance, use AI to crunch through thousands of user sessions and highlight recurring behaviour patterns or “friction points” that deserve my attention. Instead of spending hours scrolling through heatmaps, I can focus on why those patterns occur and how to address them.

An example of an AI-augmented UX research dashboard consolidating heatmaps, sentiment analysis, and drop-off rates. Such dashboards use machine learning to surface where users struggle and what they feel, enabling researchers to pinpoint issues quickly.

Smarter Usability Testing with AI

I’ve also embraced AI-driven tools for usability testing. Askable platform provides its AI-based capabilities for running AI-moderated discovery interviews. I can set up an unmoderated test in the evening and wake up the next morning to a dashboard of automatically generated insights, complete with heatmaps of where testers clicked, task success rates, and even sentiment analysis of their written feedback.

For example, when my team was iterating on the onboarding flow of a health tracking app, we conducted a study to observe new users. The AI analytics quickly flagged a drop-off point in the signup process that many users struggled with. Thanks to those instant insights, we redesigned that screen within days, ultimately improving signup completion rates and user confidence in the app. It's a level of rapid, evidence-based iteration that wasn't previously possible.

Turning Data Into Actionable Evidence

‍Crucially, these AI research tools not only save time but also often reveal patterns that we might otherwise miss. I've seen AI cluster dozens of open-ended survey responses into themes, or detect subtle behaviour trends like hesitation on a specific UI element, which a lone researcher might overlook.

The result is that far more data now backs my research recommendations to stakeholders. Instead of saying, “I think users are confused by this step,” I can pinpoint the exact step and show heatmaps or transcripts that illustrate the issue. In a sense, AI enables us to truly listen to all users, not just the loudest ones, ensuring that no critical feedback is overlooked. This breadth of analysis gives design teams and executives greater confidence that we’re focusing on the correct problems.

Personalisation at Scale: Tailoring Experiences with Machine Learning

The most exciting impact of AI in UX design has been the rise of personalisation. As a design professional, I’ve always aimed to create experiences that feel intuitive and relevant to each user; however, doing so for thousands or millions of users was notoriously difficult. AI has changed the game by enabling real-time, data-driven personalisation on a massive scale.

At its core, AI-powered personalisation means the interface can adapt to each user’s behaviour and preferences without me having to hand-craft every variation. A familiar example is Spotify: the design community often points out that Spotify doesn't just recommend music with AI; it also adjusts UI elements, such as playlists and home screen layouts, based on an individual's listening habits.

When I open my Spotify, the app's layout and featured categories feel like they were designed just for me, and in a way, they were by an AI that knows my patterns. I strive to bring a similar level of intelligent customisation to the products I work on.

In practice, personalisation can take many forms. On an e-commerce website, for instance, AI might analyse a user’s past browsing and purchase history to rearrange the homepage on the fly, surfacing products they’re likely to be interested in. Instead of a static “one-size-fits-all” page, the content prioritisation becomes fluid and user-specific. 

I’ve worked on a project where we used machine learning to predict what category a shopper was most likely to click next, and we dynamically adjusted the navigation menu for that user. The shoppers never knew an algorithm was at play; they just found what they needed faster. Similarly, on a news platform we experimented with, the AI would learn which topics a reader engages with and then gently promote more of those, creating a personalised news feed. In both cases, we observed an improvement in engagement metrics, validating that users appreciated the more relevant experience.

Predictive Personalisation: Anticipating User Needs
‍One compelling aspect of AI is predictive personalisation using models to anticipate user needs. UX teams are leveraging predictive analytics to stay ahead of users. For example, consider a travel booking app that learns a user's preferences and can predict when they might plan their next trip [classic example of booking.com].If the AI predicts you're likely to start looking for a summer holiday destination soon (based on your past behaviour or even broader trends), the app might proactively highlight a flight sale or a travel guide on your next visit. Done right, this feels helpful, not creepy like the product "gets" you. I am pursuing a Master's degree in Artificial Intelligence, and I've observed that our learning platform also employs this approach. By analysing students' study habits, the system can serve up tailored study tips and resources when the student is likely to need extra help, thereby boosting their engagement and outcomes.

Of course, effective personalisation demands a careful balance. As a design researcher, I need to ensure these AI-driven customisations remain subtle and welcome. Users should feel understood, not watched. Transparency plays a role here when appropriate; giving users some insight or control over personalisation (such as "Recommended for you" labels or the ability to refine their content preferences) can help maintain trust. The last thing we want is to cross the line into a "creepy" territory where the interface feels invasive. When handled thoughtfully, however, AI personalisation can delight users with convenience and relevance, all while driving key business metrics like engagement, conversion, and retention.

‍From Experiment to Everyday Practice
‍Reflecting on my journey with AI in UX research and design, I’m struck by how quickly these tools have moved from being experimental add-ons to becoming integral to my daily practice. They help us see patterns in user behaviour we might miss, personalise experiences in ways that were once unimaginable, and speed up iteration cycles without sacrificing depth. But even as AI expands what’s possible, I’m reminded that the real value lies in how we, as designers and researchers, interpret and apply those insights with empathy and care.