Hiring for the AI-Ready Researcher

I recently wrote on the rise of AI-native roles in research — six graduates at Verve, hired into jobs that barely existed two years ago, none of whom called themselves researchers. The question that followed was harder: if AI is creating new kinds of research work, how do you actually hire and develop people, when the industry is still inventing the job?

It's a question I kept coming back to — and one Verve put to a panel at the MRS AI-powered Insights Conference, where we worked through what an AI-ready researcher actually looks like. Here's where I've landed.

The hiring spec is the first thing to fix

We posted an AI-focused entry-level role four different ways before we got it right. Too much technical language attracted software people. Too much research language attracted people who wanted to do traditional research. Neither was what we needed.

Liz Norman, who has recruited in market research for over thirty years, was on the panel, and crystallised why: job descriptions for AI roles routinely list technical prerequisites — tools, platforms, languages — that aren't actually required. They can be learned in weeks. But the right candidate has already clicked away. Isla Lees, one of our hires at Verve and also on the panel, would have ruled herself out if the ad had mentioned VS Code or Python. She now uses them every day.

At Verve we now ask applicants for a one-minute video and a short answer on how they currently use AI. Standout candidates aren't the most technically fluent — they are the ones who'd built something with AI in their own time, not because anyone asked, but because they were curious. And a generic answer to a question about AI is a warning sign. 

Development needs to be rethought

The conventional wisdom is that AI has broken the apprenticeship model — that the menial work which used to build instinct has gone. We're not finding that. Working with simulations replaces menial work with structured experimentation and the deliberate cultivation of judgement: seniors who ask why rather than show you how, real problems with no established answer, and a genuine expectation that a junior's instinct is worth interrogating.

Katie O'Connor, SVP at Behaviourally and also on the panel, was the clearest on what's changed here. Her own apprenticeship was fifteen years of fieldwork and tabs before anyone trusted her judgement. "That path doesn't exist anymore, and honestly, I don't miss it," she said. What's replaced it isn't one-way apprenticeship, but something more mutual — there's as much to learn from how someone like Isla works as there is to pass on from years of experience.

At Verve, cultivating that judgement means making the thinking visible, not just the output — what someone asked the simulation, where they pushed back, where they accepted the first answer when they shouldn't have. The process of deciding what matters is the exercise. The judgement is the point.

The window is shorter than you think

The industry is still debating what happens to traditional research, and who will carry first principles forward. Adjacent sectors aren't debating anything. Consultancies, media agencies and tech-adjacent businesses are hiring AI-native graduates directly, often paying more than research does, and building their own version of customer understanding without research rigour underneath it. Keep that up and insight risks losing ownership of the conversation — not because the work disappears, but because someone else ends up doing it.

The next generation of researchers — whether they come through simulation-led roles like the ones at Verve, or traditional research made faster with AI — is already out there. The industry has maybe two years to decide whether it's claiming them, or handing them to someone else.

Looking to make the move into AI-powered research - hear from three people at Verve who've made it.