Published August 21, 2026 · Written for Library-school faculty, academic libraries, professional associations and AI-training providers
Library-science faculty are experimenting with AI prompt libraries, but systematic creation, sharing and institutional stewardship remain limited. Primary Research Group found that about 31% of surveyed library-science faculty had used a prompt library. Use is therefore established enough to matter, yet far from routine. Only 4% had purchased a prompt, suggesting that faculty still rely mainly on freely available collections, personal experimentation or informal exchange.
Contribution rates were lower than use. About 9% had added prompts to a shared archive, and a similar share had developed reusable ChatGPT templates. Seven percent had participated in proprietary AI development. These findings point to a familiar knowledge-management gap: many people consume examples, but relatively few convert their own successful practices into documented, reusable assets.
Where sharing did occur, 13% used professional or trade associations and 7% used GitHub. Strikingly, no respondent in the sample reported using an academic library or institutional repository to share prompts. That does not prove such activity is absent everywhere, but it highlights an open role for institutions that already organize, describe and preserve scholarly resources.
A useful prompt library should contain more than clever wording. Entries need a stated task, intended audience, model and date tested, input-data restrictions, expected output, verification steps and known failure modes. Without that context, a prompt can be copied while the judgment that made it effective is lost.
For library schools and academic libraries, prompt collections could become teaching infrastructure: curated examples for reference, metadata, research support and administration, paired with guidance on privacy, bias and factual checking. The commercial opportunity for associations and training providers is equally clear. The market may value trusted, field-specific prompt workflows more than another undifferentiated list of commands.