Group leaders
Summary
The advent of large language models has lately and precipitously increased the demand for scientific publications for AI training, fine-tuning, and retrieval-augmented generation. However, these use cases are frequently proscribed by the terms of institutional content access licenses. While market solutions are slowly emerging and appellate court precedents still pending, in the meantime we may help our user communities locate fungible or alternative resources that are unencumbered for AI use cases, under frameworks such as open access, Creative Commons, Government Use Right, Fair Use, etc. The focus of this Working Group will be developing resources for libraries and users to support the utilization of unencumbered resources for AI use cases. Upstream, publishers may be interested in participating to better facilitate use of their open access publications.
In scope:
- How to locate and identify unencumbered publications, including use of aggregators and indices; finding unencumbered versions of specific publications; and finding unencumbered alternate, topical publications.
- How to collect unencumbered publications, via APIs or polite web scraping, aided by templates and workflows using readily available tools, and with consideration for available data formats.
- Prescriptions and best-practice interpretations of licenses and frameworks allowing for unencumbered use of publications for AI, including Creative Commons, other open access licenses, the Government Use Right, Fair Use, and other emerging community standards.
Out of scope:
- Post-processing of retrieved publications.
- Licensing content for AI use cases.
- Lobbying for government policy.
Deliverables
- Explanatory issue brief for end users.
- Resource guide(s) for end users on finding and retrieving unencumbered publications, including making sense of enabling license or permission frameworks.
- Recommendations for aggregators and publishers on how to facilitate discovery and use of unencumbered versions of publications.
- Recommendations for authors on how to facilitate discovery and use of unencumbered versions of their publications.
