Real Talk on AI: Human Design, Not AI Drift

Summary

AI opens the door to more “adaptive” interviews; however, adaptive does not always result in better data.

When we built our AI Moderator, we made a deliberate decision to stay structured. Why? Because consistency, validated question design, and regulatory guardrails are crucial, especially in high-stakes environments.

The result: better data that’s comparable, reliable, and ready for real analysis.

Transcript

Hi everyone, I'm Christopher Farina, Director of Listening and Linguistics at inVibe. Welcome to Real Talk on AI. Today we're talking about why we've kept our interviews structured when we created our AI Moderator instead of semi-structured or unstructured formats.

Now, structured interviews mean that every participant gets the same questions in the same order, in the same way. And that's different from the more flexible approaches where moderators guide the conversation in real time.

Having used AI since 2021, we knew that there were parts of how we collect voices that we wouldn't or really couldn't use AI for. First, our questions are grounded in decades of best practices, from linguistics to life sciences, and in our own validated question library. We're not comfortable handing those choices to a system that might blur important distinctions or miss do-no-harm principles that are central to life sciences research. Second, we operate in a tightly regulated environment, so we need guardrails that stand up to legal and pharmacovigilance review. And third, and this one's a little bit hard to admit for folks using AI every day, LLMs are really still too unreliable for live moderation, you know, especially in heavily regulated settings like our own.

Even a single deviation can introduce risk and that risk compounds with every step off script. So, our Moderator just never goes off script. It preserves the exact language and structure of our original design for the interview. The result is a consistent participant experience and data that's clean, comparable, and ready for our in-depth linguistic analysis.

So, leave a comment or reach out to learn more about our approach to AI and how we use it to help us do the work more quickly and at scale. Thanks for watching!