Skip to Main Content

Himmelfarb Headlines: Director's Desk

Director's Desk - Generative AI

Director's Desk, Laura Abate

Generative AI has been a frequent topic of discussion since last November when ChatGPT launched.  ChatGPT is a ‘large language model’ that allows a user to ask questions to which ChatGPT will create and immediately display responses.  Text that is generated by ChatGPT is difficult, if not impossible, to distinguish from human-generated text.   And, while ChatGPT can generate text, it can also generate inaccurate information, too - by fabricating information (including citations) when it doesn’t have the right information or by providing outdated information.  ChatGPT was initially trained on data through September 2021 so its responses reflect that limitation; premium ChatGPT subscriptions now use a version of ChatGPT which will continuously browse the Internet for updated information. 

Does GW have a policy for ChatGPT and similar technology?  In April 2023, GW released Guidelines for Using Generative Artificial Intelligence.  This policy covers all GW schools except the School of Medicine and Health Sciences and the School of Law.  GW Law has since published its own policy.  The School of Medicine and Health Sciences does not currently have a policy.

How can I learn more about ChatGPT?  The best way to learn about this technology is to try it.  Ask ChatGPT questions about a topic in your field of expertise.  You can ask it to provide information, or for something more specific, e.g. a one-page summary, an outline of a 5 page paper, citations to support that paper.

What should I tell my students about generative AI?  Dr. Ryan Watkins, from the GW Graduate School of Education and Human Development, has developed several resources to support GW faculty who want to learn more about generative AI and updating course materials to reflect its availability:

What about using ChatGPT for generating patient notes?  Generative AI tools have promise for productivity gains in the future, but at present pose a risk to the confidentiality of patient data.  This article may be useful for understanding this topic more broadly:

AI Chatbots, Health Privacy, and Challenges to HIPAA Compliance
Marks, M., & Haupt, C. E. (2023). AI Chatbots, Health Privacy, and Challenges to HIPAA Compliance. JAMA, 330(4), 309–310.

How should information from ChatGPT be cited?  Himmelfarb Library’s Scholarly Communications video series provides a brief video on Artificial Intelligence Tools & Citations to show you how to accurately cite generative AI as a source.