AI in Research: A Hyper-Intern or Complete Replacement?
Why the future of qualitative insight lies in collaboration—not automation.
The research world is buzzing with AI. Everywhere you look, new tools promise faster, deeper, cheaper insights. But if you peel back the marketing gloss, a more nuanced question emerges: Where does AI genuinely help—and where does it fall short?
In a recent Early Adopters session in collaboration with CoLoop, Hazel Haskayne, Director and Qualitative Research Practitioner at Purdie Pascoe, opened up a whiteboard-style forum to unpack exactly that. What followed was an honest, sometimes provocative, insightful conversation among researchers experimenting at the frontier of AI-powered analysis.
From Curiosity to Caution to Confidence
Hazel opened the session by reflecting on her own AI journey. Back in 2022, her team at a global qual agency asked a simple but powerful question: What would make our lives easier? The answer? A research companion—not a black box, but a tool that could reduce time spent wrangling transcripts, so they could focus on real thinking.
That companion turned out to be CoLoop. And like many teams, they approached it cautiously. The fear wasn’t just about accuracy—it was about expectation. “I didn’t want teams to think they could plug in data and magically get beautiful slides with perfect insights,” Hazel said. “That’s not how it works—and it’s not how it should work.”
The Myth of the Magic Machine
Several participants echoed that point: AI is only as useful as the thinking behind it. “The worst prompt you can use is something like ‘Give me insights about X,’” said one attendee, “It’s lazy. You have to give the tool context, structure, something meaningful to work with. Then it can really shine.”
That theme—AI as a sparring partner rather than a sage—ran throughout the discussion. For many, tools like CoLoop aren’t replacing researchers; they’re extending their capacity. One participant called it a “super intern.” Another: “It’s like having a second brain to challenge my own assumptions.” But all agreed: the real work still belongs to humans.
What AI Can’t Do (Yet)
While AI can cluster responses, surface patterns, and speed up segmentation, there’s a growing recognition of its limits. Hazel pointed out that AI lacks emotional nuance and cultural context—both essential in qualitative work. “You can’t walk in someone’s shoes with a machine,” she said. “You can’t read the room. And in qual, that matters.”
Others raised concerns around “lazy analysis”—the temptation to take AI output at face value, especially when deadlines are tight. “It’s easy to go down the rabbit hole,” said a researcher in pharma. “You ask for insights, and boom—there they are. But if you don’t challenge that output, you’re not doing your job”.
Smarter, Not Lazier
Despite those risks, the consensus was clear: AI can help researchers be better—if they use it well. From checking bias to stress-testing conclusions, tools like CoLoop are increasingly being used not as replacements, but as reflective aids.
One attendee, put it simply: “It’s like having someone in the room to bounce ideas off—someone who never gets tired.”
This is where AI finds its sweet spot: freeing researchers from repetitive, time-consuming tasks so they can spend more time thinking, questioning, and connecting the dots.
Building Trust with Clients
A critical part of the discussion focused on how researchers are presenting AI use to their clients. With new regulations requiring transparency, several teams now include AI tools like CoLoop directly in proposals and consent forms.
But it’s not just about compliance—it’s also about framing. “We position it like this,” said Josh from CoLoop. “Would you rather I spend hours cutting clips, or focus on delivering deep insights? CoLoop lets me spend more time on what actually matters.”
The Real Value: Thoughtful Use
What made this session different from most industry webinars was its honesty. No one claimed AI had all the answers. No one pretended it was perfect. And yet—almost everyone agreed it’s becoming indispensable.
The reason? Not because it replaces expertise, but because it amplifies it. When used thoughtfully, tools like CoLoop don’t automate insight—they unlock more time to find it.
So maybe the real future of research isn’t AI vs humans, but AI with humans—working together, asking better questions, and thinking just a little bit deeper.