A few weeks ago, I wrote about a moment from a qualitative research conference that stuck with me: the sense that much of the industry’s conversation about AI has become a defence of qualitative research itself.
That’s understandable. But it’s also the wrong conversation.
Here’s my take: pitting qualitative research against AI is the wrong fight.
AI isn’t sitting outside the qualitative research process, waiting to be let in or kept out. It’s already woven through it. So, the conversation worth having is a narrower one:
- What, specifically, do we mean when we say “AI”?
- Which AI tool is right for which research task?
- And how do we get it pulling its weight alongside the human judgment and research expertise our teams already have?
What does responsible AI look like in research?
That’s exactly what our first AI Hackathon set out to explore. Looking back, it was a genuinely brilliant few days.
Over three days, people from Escalent, C Space and Hall & Partners built and stress-tested AI agents against real problems from the business. No presentations about “the future of work.” No theoretical exercises. More than 130 ideas, 23 teams, 118 people, heads down building something that had to actually work.
There was a real buzz to it, the kind you don’t often get from an internal event. People who’d never worked together before, from different brands and different time zones, were huddled around laptops, arguing happily about the best way to solve a problem that really mattered to the work they deliver every day. It felt genuinely mixed, genuinely global. Watching teams present on the final day, you could feel how much energy had gone in. Every single idea had something genuinely exciting in it, and had personality, too: there were unicorns, LEGO and pirates showing up which emphasized the human touch on the technical solution.
What I noticed pretty quickly was that the best outcomes weren’t coming from the strongest AI prompts. They were coming from the strongest combinations of people.
The most valuable AI solutions don’t emerge from the smartest prompts. They emerge when diverse expertise comes together around real business problems.
Every team needed more than one kind of contributor:
- Someone who lives close enough to the work to know where it gets slow and repetitive
- Someone who could turn a rough idea into a working prototype
- Someone fluent in the workflow who could spot what was actually worth automating
- Someone who could hold the client’s perspective
No single contributor held all of that judgment alone. Take any one of those roles away and the idea doesn’t survive contact with reality—but put them together, and a team with plenty of junior people in it produced ideas as sharp as anything a lone senior expert could have called.
What AI does for research won’t be decided by technology. It’ll be decided by how well we choose what to point it at, and who we build around it.
Handled well, AI removes the friction that was never really where our value lived in the first place—tidying up messy inputs, getting a first draft moving, giving us a stronger place to start across qualitative research workflows.
Handled carelessly, it flattens things. It can narrow the range of language and ideas rather than widen it. None of that is a case for staying away from it. It’s a case for staying close to it.
The future of qualitative research won’t be determined by how much AI we use, but by how intentionally we combine AI efficiency with human judgment.
Why human judgment matters more than ever
The value of qualitative research has never really lived in the process.
It lives in human judgment. It lives in knowing what actually matters, what rings true culturally, what will genuinely shift how a business thinks. And it lives in the language people use without meaning to be memorable: the strange, specific line that lands in a room and changes it.
That’s the part no tool gets to own. We do.
So rather than spending our energy defending human insight from AI, we should spend it getting sharper about what human insight is actually for, then pick up every tool available to do more of it, better.
That’s what the Hackathon looked like to me. Not a threat creeping into the work, but an exciting look at how the expertise we have can be elevated even more through thoughtful AI adoption.
