This guide provides an overview of generative artificial intelligence, explores resources for the ethical use of AI, and offers suggestions for effectively integrating generative AI tools into your research.
Generative AI, especially large language models that generate text by statistically predicting strings of words, can contain false information that is stated to be fact. The New York Times reported on one study by the tech start-up Vectara that found that "chatbots invent information at least 3 percent of the time — and as high as 27 percent."
Check the various large language model chatbot hallucination rates using Vectara's Hallucination Evaluation Model.
Asking ChatGPT About Davidson College
When asked to describe Davidson College and it's most famous alumni, the output included at least 6 factual inaccuracies and one completely invented person. See the full conversation for the prompts.