People are messy. We change our minds halfway through a conversation. We say one thing and then immediately qualify it. We develop workarounds for products that designers never anticipated. We make decisions based on habit, emotion, convenience and context, then explain them later as if they were completely rational.

For researchers, that messiness has always been valuable because it’s often where the most meaningful insights come from.

Yet at the same time, many of the technologies reshaping research are designed to do the opposite. They organise, summarise and smooth complexity into patterns that are easier to understand and scale.

AI is making it easier than ever to identify what is most common. It can process large volumes of data, identify patterns quickly and give us a clear sense of how people are generally feeling or behaving.

One of the patterns that keeps emerging in conversations around AI, however, is how often outputs converge around the average. You get a strong sense of what is most common, with less visibility into the long tails where people are more emotional, more conflicted or reacting in ways that don't quite fit.

In my experience, that’s often where the more meaningful insight sits.

What starts to surface in real conversations

In more open or conversational settings, the initial answer people give rarely tells the full story.

On the surface, responses can sound clear and well-considered. But as the conversation unfolds, additional context emerges: internal pressures, constraints, competing priorities or nuances people hadn’t initially articulated. It’s these layers that reveal how decisions are actually made, beyond the polished, surface-level rationale.

The same thing shows up when you observe people in their own environments. Watching someone move through a product experience or a daily routine tends to bring forward moments of hesitation, small workarounds or behaviours that have become second nature. There’s often a level of imperfection in how people navigate those experiences, and that’s part of what makes the insight useful (and real).

How that fits alongside AI

AI is becoming essential to how research is done. But it does tend to smooth things out in ways that, as researchers, we should be concerned about.

When you’re working with large datasets or synthetic approaches, you often end up with a very strong view of what is typical. What you don’t always see as clearly are the responses that sit outside that, the ones that are more instinctive, more situational or shaped by context that isn’t immediately visible.

Those responses can be harder to capture, but they often explain why something is happening, not just what is happening. That’s why maintaining a connection to real people remains so important, even as AI becomes more embedded in our work.

Why this feels bigger than research

As AI-generated content becomes more common, people are becoming surprisingly adept at spotting what feels manufactured. Whether it’s social media posts, marketing copy or everyday interactions, there is a growing sensitivity to content that feels overly polished, overly uniform or disconnected from real experience.

That doesn't mean people are rejecting technology. Most of us use AI in some form every day, but as more of what we see and consume becomes automated, polished and increasingly similar, society is left craving the real: experiences that feel tactile, imperfect, grounded in nature or connected to something authentically human.

You can see it in the renewed interest in more ‘analogue’ experiences. From vinyl to board games, people are gravitating toward things that feel tangible and less mediated. As a recent Fortune magazine article puts it: “As technology distracts, polarizes and automates, people are still finding refuge on analog islands in the digital sea.”

The same dynamic has implications for research. As researchers, our job isn't just to identify patterns. It’s to understand the people behind them. If people are placing greater value on experiences that feel genuine and human, we need to think carefully about whether our methodologies are helping us capture that authenticity or inadvertently filtering it out.

Designing research with that in mind

Scale and efficiency remain important, but there’s a growing need to be more intentional about creating space for those kinds of responses.

A few things I’ve found helpful:

• Create space for people to think out loud: When someone is still working through what they think, you tend to get more context than when they’re trying to give a final answer.
  
  
• Capture reactions closer to the moment: People process experiences in real time, and designing for that tends to surface a different layer of insight.
  
  
• Pay attention to the responses that don’t quite fit: Those are often the ones that point to something more meaningful beneath the surface.
  
  
• Use AI to organise and connect, not flatten: AI is incredibly effective at identifying patterns, but the value increases when those patterns are grounded in real, varied human input.

Where this leaves us

AI will continue to improve how we work, especially when it comes to speed and scale.

At the same time, it can be easy to lose sight of the nuance that doesn’t show up in the average. That broader view is useful, but it doesn’t always explain why people make the decisions they do.

The insights that tend to move things forward often come from understanding where that pattern breaks and what’s behind it.

As more of the process becomes automated, those moments can be easier to overlook and more important to capture. Our role as researchers is to make sure they aren't lost.

Paige Waldie is senior vice-president at Angus Reid