Fishing in the Archive

Abstract This essay considers the effects of the advent of Large Language Models (LLMs) on research in the humanities—specifically in the field of the history of music theory—by linking these transformative developments to the momentous rise of relevance-ranked full-text search over the preceding quarter century. It considers both the extraordinary potential of LLM-assisted archival work and its risks, examining AI-enabled archival searches and new research paradigms on the one hand, and opaque mediation, “AI slop,” and above all the dangers of deskilling and cognitive surrender on the other. It argues that music theorists and humanists more broadly must actively engage with these tools rather than avoid them, since scholars trained in earlier research regimes are uniquely positioned to recognize what stands to be gained—and lost—as close reading gives way to new methods (and habits) of research.