• Rachel Walker, Author |
5 min read

Under normal circumstances, I am very good about ensuring that the train I am about to board is indeed going where I want it to go. Nearly every time I have boarded the wrong train it was because I thought the train was about to depart without me. You come down those escalators, see the doors about to close, and rush through them in a panic, only to discover that you’re now heading in the opposite direction than you intended. (Just me? Cool.)

We are experiencing this pressure on a much larger scale with Generative Artificial Intelligence (or GenAI). Under normal circumstances, successful organisations’ design processes are guided by user needs, not technological advances. The release of Chat-GPT and ensuing GenAI craze has companies in a panicked state, worried that they will be left behind if they don’t find a way to integrate GenAI into their products and services. In their haste, companies are asking what they can be doing with GenAI instead of asking what they should be doing. Designing with the goal of incorporating trendy technological advancements instead of addressing user needs can lead to suboptimal designs at best, and genuine harm at worst.

Proper implementation of user research can avert some of these negative consequences and ensure that this ground-breaking technology is used to make a difference in peoples’ lives. At the same time, GenAI has the potential to make user research more efficient, effective, and fun. Advancements in content generation, language translation, and information extraction have the potential to automate tasks and enable researchers to connect with previously underserved groups to create more inclusive products. GenAI and user research: a soon to be iconic duo.

How can user research improve GenAI?

User research can help address many of the shortcomings and challenges that GenAI faces. Focusing the design process around genuine user needs means you avoid shoehorning in trendy but useless features. By making the experience of users your top priority, you are less likely to implement a feature that makes that experience worse (or compromises their privacy).

One of GenAI’s main challenges is incomplete or biased data used in the model training process. This can lead certain groups to be better represented than others, which can perpetuate existing inequities. For example, the current image-based algorithms for detecting melanoma were primarily trained on moles from white individuals so they are not sensitive enough in people with darker skin. Ensuring that GenAI represents the needs of all users doesn’t just lead to better products and services, it leads to a more just society.

To ensure that everyone’s needs are being served, you first need to understand the needs of the relevant user groups. User research can help identify these needs and highlight certain groups more at risk of being underserved or harmed by the implementation of a product or service. This data can be used to enhance training. For example, introducing a voice-enabled chat interface might be genuinely valuable, but if the model was only trained on certain accents, then this introduces a disparity in the quality of service. I would know - my first user research job was collecting speech data to help Alexa better recognise accents besides just native-English speakers.

How can GenAI improve user research?

There is a plethora of AI-enabled tools that will make user researchers’ jobs easier, freeing up time to focus on the higher value parts of our work. Gone are the days of hand-transcribing recordings from user interviews and painstakingly coding each line. Speech-to-text transcription technologies have come a long way, including more supported languages and improved ability to decipher accents. Do I think the technology has advanced enough to rely on it exclusively? Definitely not, but it’s getting close.

Analysing free-response answers to large-scale surveys is excruciating. Nowadays, GenAI can support the end-to-end process, from generating a survey, extracting insights, and creating data visualisations. AI-enabled sentiment analysis has been around for a while, but it has improved in recent years. It’s not just about efficiency - GenAI-enabled tools can identify patterns in user behaviours and preferences over time. Going forward user researchers will likely focus more on quality-checking AI output as opposed to doing the analysis themselves, freeing them up to do the tasks they actually enjoy.

Focus-groups, workshops, and co-design research are also getting a facelift. GenAI-enabled prototyping will allow designers and participants to bring concepts to life in a more tangible and visual way. GenAI will allow designers to create interfaces in real time to iteratively develop ideas with users and stakeholders. Brainstorm sessions can be aided by chat bots to generate additional ideas and inform decision making.

Lastly, GenAI tools can make inclusive research easier by removing language barriers. Recent years have seen advancements in language translation technology, including previously underserved languages. This has the potential to be a game changer for user interviews and surveys. We are years, not decades, away from real-time conversation translation, allowing researchers to include previously ineligible participants.


I’m not generally one to get swept up in the hype of a new technology (which is easier said than done when you spend years working in Silicon Valley). While it is unlikely that GenAI will usher us into a labour-free utopia, I don’t think it is overly dramatic to predict that GenAI will change the way many people do their jobs and live their lives. User researchers will likely be part of the estimated 40% of employees that need to upskill due to the advancement of GenAI-enabled tools. I don’t think user research will ever be replaced by AI entirely, but I look forward to AI handling the portions of my job that are inefficient.

Due to the immense rate and scale of change that GenAI will bring, the stakes for getting it right are high. User research isn’t a perfect antidote for the limitations and risks that GenAI poses, but it can certainly help. Taking the time to centre users in the design process when incorporating GenAI into products and services isn’t just smart from a business perspective, it’s vital from a societal perspective. Failure to do so can present genuine harm and perpetuate existing inequalities in our society. While no one wants to miss the train, it’s worth taking the extra moment to make sure it’s going the right way.

Related Insights