top of page

Using AI for qualitative user research – opportunities and limitations

Updated: Oct 4, 2023

The democratization of AI tools such as ChatGPT, Bard, Midjourney or Dall-E has caused marketing and design professionals to quiver with excitement and immediately after shudder with fear.

While we are all knee-deep in AI wizardry, we're wondering if and when these digital marvels will render us humans obsolete.

This article explores ways AI can be used to enhance qualitative user research as well as a few tips on how not to get carried away by the magic.

Collecting users' needs and pain points ideas using AI

AI tools such as Chat GPT are great at analyzing huge amounts of data from all kinds of sources - industry reports, news sites and most interestingly forums, social media and reviews. It would take days to find and analyse this type of data for a designer, ChatGPT does in less than a second.

Below is an example of questions you can ask.

Once you have phrased - and rephrased - the questions in the right way, you can also prompt ChatGPT to give you more insight, just like you would do in a real user interview. You can also ask the sources of the insight, to check they are credible.

This data can be used to bring the researcher up to speed on users’ potential needs and experience and provide a base of data to be validated through further primary research such as user interviews.

Preparing qualitative research material

Provided with the right context, ChatGPT can help draft various types of research material such as user personas, user interview questions or online surveys.

Image generators such as Midjourney or DALL-E can also quickly produce research props such as concept illustrations. For storyboards, tools such as Storyboardthat or StoryboardHero save a lot of time.

AI generated storyboard mock up by Storyboardhero

AI can help generate solution ideas

AI can generate a wealth of solutions to previously identified needs and pains. In the example below on home retrofits in the UK, ChatGPT has come up with more than 10 ideas. I have personally done a significant amount of research in this sector, and I can tell all of them are very relevant!

Again, this material can be used to inform topics to explore and validate at future research stages, which saves time on some of the discovery activities.

Preparing usability testing

ChatGPT cannot review any image or URL of a prototype, but it can help you prepare your testing plan before conducting usability testing with real users. You can ask ChatGPT to give you a list of potential usability issues to watch out for in similar products.

The limitations of using AI for qualitative user research

AI is very good at analyzing huge amounts of data but sometimes data is just not available because no one else has surfaced the insight or the data is too recent.

Most AI chabots have been trained on data with a cut-off date, not real-time. Chat GPT would confess that “As of my last update in September 2021, I have been trained on a diverse range of internet text up until that point. I don't have access to real-time data or updates beyond that date “. As of September 2023, Google's Bard and ChatGPT both have access to real-time information and they are progressively rolling out this capacity to users...So watch this space.

Even if data produced by AI is up to date, it should always be handled with human judgement to stay away from errors and bias.

Research methods enhancing the credibility and validity of findings - such as triangulation - should always be used.

Overall, AI can do wonders for researchers but will not replace them. When mastered, AI magic can save time in some discovery tasks, which can be invested in other value-added activities that are often rushed such as observing, talking to users and making sense of all the insight. AI should be used wisely, as an additional, time-saving insight gathering method or data analysis tool, and always combined with other research methods to help come up with validated insight.


Join our next User Research Webinar on 11/10/23

24 views0 comments


bottom of page