Are Academic Libraries Teaching Faculty to Use AI for Data Management?

p strongFaculty are experimenting with AI in data work, but relatively few report receiving relevant training from their academic libraries./strong/p p Primary Research Group's survey of 339 faculty&mdashdrawn predominantly from STEM fields at research universities and medical schools&mdashfound that about 11% said their academic libraries offered training in AI applications for data management. The report covers visualization, cleaning, discovery, summarization, analysis, tagging, upload and integration with statistical software./p p The training gap is important because these tasks involve more than prompting. Faculty need to understand whether a tool exposes data, invents values, changes formats, produces reproducible code or can document the transformations it recommends. Licensed datasets and human-subject information may introduce additional constraints./p p Libraries can begin with short, task-based sessions co-taught with research computing or data-services staff. Demonstrations should use non-sensitive examples and include verification, citation, provenance and export. Consultations for real projects need a clear boundary between methodological support and responsibility for the research result./p p A library does not need to endorse every AI application to play a useful role. It can help researchers compare tools, identify risk and build a workflow in which speed does not come at the expense of data integrity./p p  /p