When AI is certain, we should be sceptical – Uncertainty, friction, calibration and qualitative analysis
In qualitative research with AI, uncertainty should not be eliminated but made visible, because doubt, ambiguity, and even absences in the data can constitute relevant information.