Introduction

This blog post is a response to the call to arms outlined by Jakob Nielson, who argued that UX needs more urgency in its response to AI. He argues that UX professionals need to move quickly to get up to speed with using AI tools. In this post, I outline some ideas about how UX researchers could do exactly that. This includes some ideas about how UX designers and researchers can incorporate AI tools within our work. This is both how we might incorporate AI into our toolkit as UX researchers and how we are setting the foundations to help clients transition to AI. My logic is that to authentically help clients transition to AI, we need to walk the walk, not just talk the talk. We are in a better position to help clients transition if we build up our own skill set in applying AI to our own work.

How AI Can Help Us Deliver Better Research

As design researchers, we are typically working under pressure to tight deadlines. Existing and emerging AI tools can help us deliver rigorous research within these constraints. AI can help at each step, including with planning, data collection, analysis, and communication.

AI can help through a combination of detection, prediction, and generation.

Planning 

Generative AI tools including Chat GPT can help us to plan our research. We can use such tools to help create documents such as research plans and discussion guides. It is important here that we use such tools to help and not to simply defer to them, they can help speed up the process by helping as a checklist and with formatting, but we need to ensure that we apply our expert knowledge here as well.

Data collection 

Detection functionality including the ability to turn speech into text has created an explosion of tools that can help with data collection by generating transcripts connected to videos, to help find specific clips. These tools can also help by creating summaries to highlight key points from long transcripts of interviews and meeting notes.

Analysis 

Detection functions in AI tools can also help with analysis, such functionality is starting to appear in virtual whiteboard/post-it note programs. These can create summaries across large amounts of text. They are also starting to be able to identify themes. 

Communication 

Slide deck presentation AI tools can help communicate insights to stakeholders. Communication can also be enhanced through AI image generators and tools that check, spelling, grammar, and readability. 

Documentation and research management 

Some tools provide whole ecosystems of interconnected tools that also help with documentation providing links between desk research and participant recruitment or between analysis and communication documents.

Overcoming Challenges and Barriers

Effectively making use of AI is however not without its challenges these include deciding which tools to use, overcoming bias, and securing buy-in for new ways of working.

The plethora of tools available 

Because AI is moving so fast there has been an explosion of AI tools that could be used in design research. 

From initial scoping, I identified forty-three different AI tools. Most of these require licenses for all collaborators. To enable team members to work together in research, most require everyone involved to have licenses and to understand how to use the AI tools that are approved. There is a danger that this could result in fewer collaborative research approaches emerging due to these financial and technical barriers. The challenge here, especially for tools that offer an ecosystem of connected tools is that documents and data for all of these would be locked inside the ecosystem and so everyone collaborating on a project would require a license. There is an organisational risk here, as time and money could be invested in the use of tools that cease to exist in a few years or become seen as clunky when superior products become available.

Overcoming bias

The outputs of AI are biased and not value-free. Bias exists due to prejudice within the training data that has been used to create AI. Where the training data is everything on the internet, there are cultural biases, including the dominance of Western, particularly the dominance of North American and European content available. As more internet content is created with the assistance of AI there is also the danger of self-reinforcing loops getting created. It is also commonly reported that generative AI is subject to hallucinations, creating content that sounds convincing but is fictional.

Securing buy In 

To gain the greatest possible value for AI, we may want to move away from simply enhancing existing methods towards creating a new UX research paradigm. Literature suggests that as routine tasks become partially automated, researchers will be able to make connections across wider datasets and so engage with clients at a more strategic level. Enabling such a shift, however, would require agreement from key stakeholders for a new way of working that is different from their current assumptions.

Overcoming Barriers

To address the barriers identified it is important that researchers apply their critical thinking and experience to spot and address bias. This will be vital to enable researchers to work symbiotically with AI tools whilst addressing the risk of distorted content.

At an organisational level, whilst there may be a danger of engaging with the wrong tools the risk of delaying is greater. If we delay until we know which the best tools to use are then we face the threat of being overtaken by teams of researchers who have already gained the efficiencies enabled by these them. Even if we start experimenting with AI tools that become redundant, we are still gaining core skills and understanding that could be transferred across to other tools. By doing so we would be in a much stronger position than we would be if we simply wait.

Another risk is that pressure from clients or the organisations we work for push us to use emerging AI technologies to simply conduct faster research rather than to use them to conduct better research within existing constraints.

Conclusion

From the overview outlined it should be apparent that existing and emerging AI tools have the potential to help us to conduct better research. It should also be clear that many technological and organisational constraints need to be addressed if we are going to harness these tools to produce better research to add better value for our clients. As discussed above, to gain the value that these tools it is increasingly important that we apply our research skills including critical thinking so that we continue to conduct good research rather than to simply defer to the technology, which would inevitably diminish the quality of research outputs. It might result in fast outputs, but it’s unlikely to address the deep-routed complex problems faced by clients. To deliver good research and better value, a symbiotic relationship between us researchers and the AI tools we use is needed.

As designers and design researchers, we are still very much at the beginning of our journey towards learning how to use AI tools to enhance the quality of our work. Whilst some of us are on the journey towards exploring how to use some of these tools, as yet nobody has all the answers. This makes this process both exciting and challenging. One of the aspects of design research that I love is collaboration. To work out the most effective ways we can harness these tools a collaborative approach is very much what is needed.

Do contact me if you have thoughts about which tools are likely to add the most value, or how we might harness them to enhance our work. I am certainly very excited about where these roads might take us. Join me if you would like to walk this path together.