AI in Usability Research
AI, AI in UX

AI in Usability Research

We all love a sidekick: Chewbacca, Ron Weasley, Samwise Gamgee. In films and movies, a sidekick supports the protagonist by helping them achieve their goals, overcome difficulties, and develop as a character. Sidekicks can also offer high-level observations that move the plot along and add a hue to the narrative.

We all love a sidekick: Chewbacca, Ron Weasley, Samwise Gamgee. In films and movies, a sidekick supports the protagonist by helping them achieve their goals, overcome difficulties, and develop as a character. Sidekicks can also offer high-level observations that move the plot along and add a hue to the narrative.
Part of what makes the sidekick lovable is that he or she isn’t meant to be the story's protagonist. In a sense, he or she is meant to be an offshoot of the protagonist and act as a point of refraction for various themes in the plot.

AI is like a sidekick for researchers. It isn’t meant to play the starring role in our work; it cannot be expected to hold the weight of the narratives we weave as researchers. But it can be helpful for ideation, support, summarization, and organization. It can help researchers gather and organize the information they need to develop their research and assess their findings.

Samwise was never meant to be the one to save Middle Earth. For all of his lovable, industrious traits, he couldn’t have been trusted with the ring. AI similarly can’t be trusted to give accurate insights on all of the data researchers deal with during testing. But it has use cases that make it worth having on the journey.

How AI is showing up today in UX research at large

As it turns out, the majority of researchers have some version of an AI sidekick helping them in one way or another: 92% of those asked in a Nielsen Norman Group survey said they had used at least one type of generative AI tool in their research efforts. The majority of usage was around generating types of copy, such as microcopy for modals, drafting examples of UX OKRs, summarizing benchmark studies, or generating definitions. Others reported using it for gathering general information or getting resource recommendations, and others said it was helpful for some UX research tasks (generating recruitment text, summarizing insights, and drafting research reports). Thirty-one percent of those surveyed said they used some form of generative AI to help jog ideas when drawing a blank around scenarios and solutions for a study.
AI can also help with the design side of research. Nearly a quarter of researchers surveyed in the study above said they used AI tools for design-related tasks. For example, ChatGPT can create storyboards and user flows to help researchers expand their approach (remember that prompts need to be highly detailed and informed to generate helpful responses). It can also generate basic wireframes and prototypes that can suffice in certain types of testing. 

It’s crucial to remember that these generative AI tools will only take researchers as far as their own knowledge and expertise can take them. UX demands a thorough understanding of products, users, and domains—you can’t fake this type of knowledge, and AI can’t build castles out of the air around an ill-informed prompt. The more complicated a task is, the more sophisticated the prompts will have to be, and even then, there isn’t a guarantee that the results will be accurate.

How AI is (and isn’t) showing up in usability research

Usability research has always relied heavily on the involvement and participation of humans, and this hasn’t changed. However, AI can be a helpful sidekick here as well.
With usability testing, it’s all about human observation. Research participants attempt to complete tasks while researchers watch, listen, and take notes throughout the duration of the usability test (some forms of usability testing are unmoderated, and participants might be asked to voice their thoughts out loud as they experience a product). The goal is to identify issues in usability, gather participant feedback, and use that information to enhance the design.

Usability testing is essential – it’s where the rubber meets the road, where concepts encounter the reality of human experience. It is also often very time-consuming, and depending on the scope of the project and the type of usability tests being conducted, it can be costly. AI can assist in a few ways, particularly in the fundamental stages of testing for usability. AI can support researchers by:

  • Transcribing the recordings. UX researchers have to be excellent listeners. In usability tests, they must pay attention to everything the user conveys – whether in words or actions. It’s difficult to take detailed notes when you’re observing so closely. AI-powered services can virtually compile notes (with good accuracy rates), allowing researchers to focus on the task at hand and/or jot down hyper-detailed impressions throughout testing. 
  • Video editing. AI can help identify key areas in usability testing video recordings by transcribing speech activity and providing timestamps. This saves researchers the arduous task of combing through hours of footage.
  • Eye tracking. AI tools – specifically, predictive analytics –  can track where users' eyes spend the most time during testing. This can help researchers hone in on the more subtle cues about where a user's attention is going during testing.

A note of advice on how to get started and why 

Based on the stats shared above, most researchers have already found ways to use some version of AI in their research processes. My advice to anyone in the field is to regard AI as an assistant who’s pretty good at keeping things organized and who can save you a bit of time when it comes to gathering information or documenting processes but who shouldn’t be expected to know how to tie the threads of any research narrative together (because it doesn’t know how). But also, don’t blow it off. AI is here to stay, and so it is incumbent upon researchers to learn how to use the tool well. AI won’t replace you and your job as a researcher, but somebody who knows how to use it well in research might. 

Consider AI as the Dr. John Watson to your Sherlock Holmes in the realm of research. Just as Watson brings diligent note-taking and a reliable perspective that often opens up overlooked avenues of investigation, AI helps organize information and process data quickly, allowing you to dig in a bit deeper in your research and analysis. It won’t make the conclusions for you, but it’ll help you get there.

At Usability Sciences, our aim is to provide the best research methods, tools, and strategies on offer to guide companies to the insights that will allow them to connect with their users on a deeper level. Visit us today at www.usabilitysciences.com