Qualitative research is so intrinsically human that it seems like the antithesis of everything AI represents. Yet the two share one behaviour in common: both sometimes make assertions without direct links to observable data.

In qualitative research, these are often celebrated as ‘intuitive leaps’. When AI behaves in an ostensibly similar way, however, it is condemned for “hallucinating,” and the resulting content is seen as not just worthless but actively misleading.

My focus here is the qualitative virtue, not the AI sin, but the comparison helps make the case for the unique value of qualitative research in the face of seemingly, all-conquering AI.

Listening so hard you feel

Not everything we learn from fieldwork is heard, recorded, or captured in a transcript. Sometimes we simply feel it – whether in the room during the session or later, as an interpretation that arrives with a sudden flash of recognition.

This is a more significant part of the qualitative craft than many realise.

You might feel underlying tensions, ambivalence, suppressed emotion, defensive laughter, or quiet conformity to group opinion and social norms. You might sense status anxiety, hesitation, or even what people want to say but are unable – or unwilling – to articulate.

This is where the magic of qualitative research happens, but the sense of connection you feel reminds you that these dynamics are deeply human. The level of sensitivity varies across the population like other traits but where strongly present should be appreciated as a signifier of innate qual talent.

Defining qual data… qualitatively

If we are willing to leap beyond the data, we must first ask what counts as data in the first place. What is straightforward in quantitative research becomes complex and debatable in qualitative research.

Consider two broad positions: the qualitative positivist and the qualitative interpretivist.

A key difference between these perspectives lies in how they define qualitative data and the researcher’s relationship to it.

The positivist researcher sees a clear delineation: the data is ‘over there’. Moderation is data gathering; analysis is a separate act. The researcher remains conscious of that boundary.

Interpretivist researchers take a different view. For them, the full data set exists partly in the researcher’s own mind. It includes not just participant words or body language but also the thoughts and feelings these evoke in the moderator.

Insights, in this sense, are co-produced by researcher and participant working together. The qualitative experience is relational, and interpretation is ongoing throughout the research process.

Of course, most researchers are too busy doing research to agonise over which model they adhere to, but if we want to understand what is happening when we “feel the room” or make an interpretive leap, this distinction matters.

What is the ‘feeling in the room’?

We often describe these moments as intuition or empathy, but the academic world offers more specific explanations from multiple disciplines.

Some of the more prominent theories ideas include:

• Embodied perception – the idea that perception is not just a mental event but something experienced through the body and shaped by unconscious exchanges between people.
• Cognitive pattern recognition – the brain unconsciously recognising tone, pacing and micro-expressions based on personal  experience. This is perhaps closest to what we call intuition, and, indeed, how AI works.
• Countertransference (from psychodynamic theory) – the moderator’s own emotional responses to what is happening in the room.
• Affective sociology – the idea that people unconsciously read power, hierarchy and social norms.
• Mirror neurons – neural mechanisms that activate when we observe others’ emotions, enabling us to feel a version of what they feel.

There may be no single cause, but several overlapping ones but taken together, however, a few tentative conclusions emerge:

• Something happens between the people in the group, including the moderator.
• The researcher is not just an observer but a research instrument.
• Physical presence plays an important role.
• Background and life experience help shape researcher what is felt.

How human is AI?

It would be rash to claim anything is beyond AI. Yet these factors – especially in combination – seem quintessentially human.

They are tied to life experience, physical presence, and perhaps even sentience.

AI may one day beat us at cognitive pattern recognition, but even then, it will not feel the same as the resonance a researcher feels when recognising a pattern rooted in the detail of their own experience.

Looking after you leap

With or without supporting quotations, qual should not aspire to proof but to the higher goal of human understanding. It is, by definition, subjective, but never more so when it is felt rather than heard.  Coming from silence, the insights are potentially at their profoundest, but the risks of projection and bias are also at their highest.

In this context, the trinity of qual best practice are even more important; a reflexivity that acknowledges existing assumptions and preferences, open discussion with colleagues and a transparency as to insight sourcing.

If we want to work with AI on our terms, then we need to have faith in our human specialness, not automatically or on principle but when the difference is real.

Interpretative leaps, that ‘feeling in the room’ may not be wholly explained here but I do hope their legitimacy and value have been demonstrated. Have a (happy) future hallucinating.

Peter Totman is qualitative researcher at Jigsaw Research