Tom Woodnutt
Tom Woodnutt is Founder of Feeling Mutual, the multi-award winning agile online qualitative research specialists. He helps clients and agencies run global studies and offers training in the space. Tom has been a Digital Skills trainer for the Association of Qualitative Researchers (AQR) and is a regular speaker at industry conferences, including the MRS, MRMW and IleX.
Any qual researcher that has had a play with Chat GPT will know how you can ask it to act like a qualitative researcher and create documents like discussion guides, screeners and to come up with ideas for tasks.
How AI will affect the design of qual - video
Inspiring different ideas
I have to say ChatGPT is pretty impressive and occasionally suggests something that your usual pattern of thought hadn’t considered and so it can inspire a different line of questioning or tasks that you may not have come to without it. It can also be useful to generate hypotheses of what people might think, feel and do - this can also inspire questions you hadn’t considered.
It can make research design suggestions for different types of qual research (from groups to depths and even mobile ethnography and online text based qual) and it is clearly drawing on expert learning data from professional resources as it’s aware of how to maximise validity and maintain openness in the lines of questioning. So for a novice or untrained researcher I can see how this could now make a project possible - giving them a ‘good enough’ template to work with - in many ways offering something better than they would have had to work with.
Generic knowledge not tailored expertise
I wouldn’t say that it does a better job than a true expert researcher would. This is because it fails to adapt or tailor tasks to the unique circumstances of a given project. While you can brief it through a prompt you can’t give it the level of briefing that we get from knowing the client’s business, the real world, and all the political and human considerations that shape a great research design. For example, it doesn’t really know how long it takes to complete an online task. It will also default to direct questions (unless told not to). Whereas it’s often the more indirect questions which lead to the best insights in qual. But with some careful prompting (for example simply asking for open questions that elicit emotional disclosure) it can make reasonably good suggestions.
How much can AI help with design?
So in terms of design, Gen AI can perform as well as a basic qual researcher - albeit with less judgement, tailoring, intuition and creativity. So it can help non-researchers design projects that they would have not been able to before. But those projects will have better outcomes, the more expert the qual researcher who is using its design ideas.
Where I have found it particularly useful is with specific questions for recruitment screeners. It can pull in best practice in questionnaire design for example it can churn out a bespoke question with a Likert Scale or other recognised best practices in questions. Or if you need to create a quick list of brands from a category that you may not know well, or other multiple choice lists, it can very quickly suggest them.
For now generative AI represents a useful assistant for qual researchers - and one that can save time and effort. Although again, the better the researcher using AI - the better its input into design will be.
