Input describes the prompts you type into a GAI tool's chat interface, files you upload, and any other content or instructions you supply to generate content.

Copyright law in generative AI contexts is far from settled. As you do with the other data, information, and source materials you use in your research, always verify the accuracy and appropriate use of the materials you intend to input into GAI tools, use to train GAI models, publish, or make other use of.

GAI tools generally run on servers managed by the tools' providers, rather than on your own computer. This means that anything you input may be saved, stolen, or passed on to others. In a world of increasingly effective predictive machines, remember that even apparently neutral data can be linked up with other data to discover more private information.

The provost has asked members of the Yale community not to submit medium- or high-risk data to GAI tools, which includes most human data. Yale-provided tools may allow higher risk data; AI at Yale offers a table of provided tools and their permitted use.

All of the same obligations that exist for researchers working with human data apply to use of GAI tools. Additional steps or requirements may apply. Researchers with questions should reach out to Yale's HRPP (hrpp@yale.edu).

Yale Library provides access to data, scholarly journals, and other resources by contracting with outside vendors. The use of these resources is bound by licenses that specify how they can and cannot be used. Any resources that require your Yale login, a login provided by or through Yale, or use of the VPN or campus wifi, should be presumed to be covered by a license.

Researchers interested in using AI with Yale Library resources should refer to the Using AI with Library Resources Guide. The guide is written and maintained by Sandra Aya Enimil, Director of Scholarly Communication and Collection Strategy at the Library, and Lindsay Barnett, Scholarly Communications Librarian.

As a note, researchers may also agree to similar terms when they login to resources with a paywall or free account, accept clickwrap agreement prompts, or otherwise agree to terms and conditions or other policies, even if these resources are not provided by or through Yale. Sometimes, simply using something constitutes agreeing to its terms.

Prompt engineering describes the intentional design of a prompt to a GAI tool intended to improve the quality, accuracy, or usefulness of an output. A number of companies offering GAI tools provide prompt engineering guidance, which may support more effective use of their tools.

For researchers interested in the theory of prompt engineering, Cliff Anderson (the Director of the Divinity Library at Yale) recommends:

Justin DeMayo, System and Application Specialist at the Medical Library, and Caitlin Meyer, Research and Education Librarian at the Medical Library, offer a general audience prompt-engineering workshop that may be of interest to researchers.

If you have any questions about methods, resources, projects, or more, feel free to book a consultation.