By now, you may have already started experimenting with OpenAI's ChatGPT (and now GPT 4). ChatGPT (and BING AI) are large language artificial intelligence systems that use collected data to create "human-like" responses to prompts. These systems are considered to be one of many emerging generative AI. Generated content may include things like essays, speeches, song lyrics and songs, graphics and images, jokes, scripts, poems, dissertations, and even computer code. It does things like score a 90th percentile pass rate on the bar exam and generate nice course outlines complete with learning outcomes. It also does things like generate fiction as fact (in small details to the entire prompted result) and presents the results as authoritative. Academic libraries have many growing stories about interlibrary loan requests for publications that don't exist and trace the citation back to ChatGPT. Unlike a system that pulls narrative from the Internet, this system uses the data in its neural network to generate content and to "learn" and improve the responses provided. The system allows for a "conversation" where the requestor can refine responses by adding criteria to the prompt. By doing this, you are actively training the system. There are AI content detectors but these tools cannot do the heavy lifting of detection that you may want and it is a moveable feast that changes daily. The U.S. Copyright Office has ruled on several cases about copyright ownership of AI created content. (Not human, no copyright). Great discussions will take place on campus this year about how teaching and learning will change because of the development of generative AI. Librarians look forward to being part of this conversation and in creating AI Literacy modules for our community. Stay tuned.
Consider the source is a good thought when getting references and citations from ChatGPT and other generative AI sources. Academic libraries are receiving more and more requests for resources that do not exist and the citations are coming from the confident and authoritative "sounding" generative AI systems. We'll be working with faculty and students to build AI Information Literacy modules and some checks to ensure that we all don't wind up chasing down these ghost citations. Our liaison librarians and resource-sharing staff stand at the ready.
No, ChatGPT Can't Be Your New Research Assistant
There once was a model called ChatGPT
Whose responses were often quite apt
It could chat all day long
And its knowledge was strong
But it never got tired or sapped.
Several of the resources linked here have ideas on pedagogical approaches and tools to help navigate a world with ChatGPT and new large language AI bots yet to come. Several of the sources we reviewed discussed the importance of including digital and media literacy in courses and programs as a way to counter the desire to abuse ChatGPT. Here are a few tools and tips. The UIW Libraries and the Center for Teaching & Learning invite your new found treasures and tactics that can be added to this guide.
