More than one client that already knew the quality of our transcription sent us work that had been AI transcribed. They sent it asking us to prepare transcripts as well. They knew firsthand that the AI transcription fell short of capturing context, tone and nuances; the three big virtues of conducting qualitative research in the first place, and those AI transcripts included a tremendous amount of AI-generated clutter.

While it’s common even in a quality recording to come upon inaudible words, skilled transcriptionists, i.e., humans, can often figure out a muffled word. The same challenge arises when speakers talk at the same time; and again, a good typist can sometimes sort out who says what, but AI is likely “hear” nonsense.  AI also cannot follow speakers with pronounced accents. AI transcription will improve, but as of 2026, human transcription remains superior.