AI Won't Moderate Your User Interview (Yet) — But It Transforms the Prep
By 2026, roughly 80% of UX researchers already use AI, and a wave of tools now sells AI-moderated interviews — promising 50 parallel sessions instead of the bottleneck of recruiting and moderation. Sources below.
My take is blunt: auto-moderated interviews are weak. Meanwhile other uses of AI in research are phenomenal right now — and those are the ones to lean on.
Where AI in research is phenomenal
Auto-transcription and auto-coding are simply excellent at the moment. That is where AI gives real leverage: working across a large dataset that a human cannot process by hand. The second area is preparation. Brainstorming and preparing for an interview — even on topics I do not know well — is superb.
This is exactly how I work today at CortexMine: I research deeply every company and every entity I meet, and I scrape the internet with that in mind. It lets me prepare extremely well for an interview or a sales conversation.
Where research ops actually breaks
The biggest problem in research ops is not the stage of listening to the user. It is the stage of forming assumptions. You have to focus hard not to bias your interlocutor with your own thinking — just let them speak. This is enormously important, especially for startups: to understand the user early and to have repeatable practices for continuously improving the product.
Why it is not the day for AI moderation
In my view, AI does not let a person fully open up today. The interviewee does not see that a human and a company want to devote their time to them. And as a company you have to delegate people who will spend that time with the user — so they truly open up. AI can serve surveys and analysis, but not conducting interviews.
On top of that, you have to be deeply curious about the company or person you are talking to. To really understand their pains, you have to be an enthusiast of the project. That is why, in my opinion, it is not yet the day for AI moderation.
How to set it up today
- AI for preparation: deep research on the interviewee and the context before the meeting.
- AI for surveys and analysis, and for transcription and coding of large datasets.
- Humans for interviews: delegate people who genuinely spend time with the user.
- Guard the assumptions stage: do not lead the answer, let them speak.
If you want research to become a repeatable part of product decisions, start with the process, not the tool. See Research Operations, or the Research Ops pillar.
Sources
Get the DesignOps Maturity Self-Assessment
A short PDF to help you spot bottlenecks in your design process.