AI cannot yet moderate a user research session as well as an experienced researcher, and its analysis is coherent but shallow. Nielsen Norman Group grants that these limitations may not last. The real argument is not about current capability gaps.
The core claim: even if AI matches researcher-level output quality end to end, designing, conducting, and analyzing studies, something irreplaceable is lost. The team observing users directly undergoes a transformation that a report cannot transfer. A researcher who hears a participant describe decades of shame around illiteracy does not just record a data point. That exposure changes how the team thinks, prioritizes, and advocates internally. That change cannot be outsourced.
The article is worth reading in full because it forces a sharper question than the usual AI-versus-human debate. It is not asking whether AI findings are accurate. It is asking what research is actually for inside an organization. If the answer is only outputs, AI wins eventually. If the answer includes the learning that happens to the humans in the room, the calculus is different. That distinction has direct implications for how teams justify headcount, structure research operations, and define what a researcher does.
[READ ORIGINAL →]