Designing tomorrow: AI as a co-pilot for UX researchers

Streamline UX research by leveraging AI

Yesterday, I saw an IDEO ad for a course on AI for UX professionals. A comment on the ad said, “if you use AI as a designer, you shouldn’t have a job.”

I’d counter with, “if you’re not using AI, you need to start!”

AI is a mirror looking back at the UX professional: the more skilled you are at UX, the better results you’ll get when using AI for UX, in particular with regards to identifying AI weaknesses and applying UX judgment to winnow AI ideas into usable product design. Jakob Nielsen on Getting started with AI for UX

AI for researchers

Many tools have entered the market place that leverage AI to assist User Researchers in a variety of ways allowing you to scale your research efforts.

In my career, I’ve used unmoderated testing tools but have not had the hands on experience of bringing AI assistance into research planning or analysis.

My company has been evaluating tools and seeing how quickly research tools have pulled in AI features opens up a world that helps us streamline, automate, analyze faster, and have better collaboration and tracking of insights.

Workflow automation and data analysis

We have a foundation for automation and AI data analysis because most modern conference call software creates accurate transcripts. Microsoft Teams and Zoom not only create transcripts but both will summarize and take notes, and even create action items.

AI can perform sentiment analysisf, generate insights, summarize transcripts or portions of transcripts, pull out key moments in a session, and automated data tagging.

None of these features reduce the need for human oversight and input, but they can reduce analysis time by half or even more.

Content generation — creating research content

Large Language Models (LLMs) like ChatGPT excel in content generation. It’s easy to jump start moderated testing scripts, surveys, or other content inputs during research planning. LLMs also do a decent job of creating user journeys.

In the example below, I ask GPT to design a survey. Prompts should include the goals of the survey and the kind of information you’re seeking.

I haven’t added the entire output below, but it did a fair job of creating an outline that will still need human finessing to make it ready for use.

It addressed the need to distinguish between frequent and infrequent fliers by asking people to select a category of how often they fly, rather than to self-identify. This type of interpretation may facilitate the researcher considering the best way to phrase the question to get the results they need.

It also made some assumptions about a potential pain point by asking if it’s hard to compare flight options.

Content generation — creating personas

While there are some benefits to coming into a moderated research study knowing nothing and being curious, digging deeper and knowing where to ask deeper question requires a pre-existing understanding of the domain.

LLMs can help though should not be the only tool used to get up to speed.

You can use ChatGPT to explore personas but it isn’t a substitute for real user interaction that should be part of the follow up.

In 2022, I saw someone launching a persona generation tool that had the value proposition of not needing to spend time talking to real users. It was met with well-deserved derision, yet, there is still value in this endeavor so long as all stakeholders recognize that there are limitations.

LLMs do a brilliant job of helping understand core tasks, pain points, worries, and stresses for any given user as well as having the ability to create personas.

“What do optometrists worry about” brings a rich response about patient care, keeping up with technology, managing their business, optimizing revenue, and more.

Asking for 3 optometrist personas without any additional information leads to a persona with a specialty in pediatric eye care. That alone could bring up rich ideas about what an optometrist might need for their practice.

Predictive intelligence for UX research

This is a newer area in AI/ML and it will be interesting to see how it unfolds over the next 2–3 years.

I’m only aware of one company in this space. Neurons Inc’s value proposition is predicting human behavior including attention, emotional response, focus, click through rate, and conversion. Their focus goes beyond application design and includes ad campaigns, websites, and more.

Their predictive analytics are based on eye tracking studies, eeg research, and learning and memory research.

Their offerings include a remote research tool that collects information from actual users and integrates with their predictive intelligence allowing for quick iterations and tested solutions that have behavioral data to drive design iteration.

I don’t work in retail or eCommerce sectors but wonder how predictive AI might influence UX in other sectors in the future.

Summary

Emerging technology has blown up in the last 2 years. Slow movement into AI has turned into rapid iteration with huge breakthroughs being announced every month or 2.

We still need human expertise and judgement to refine AI/ML generated content and human analysis is still needed for decision making. Iterative refinement between human and AI may also drive to best outcomes.

I don’t see any near term future where UX humans are replaced by AI, but all of us should be seeking ways to strengthen our work while saving time, by taking advantage of emerging technology.

How are you using AI in your research practice?