Artificial Intelligence: OpenAI, ChatGPT, LLMs, and More

ChatGPT is not connected to the internet, and it can occasionally produce incorrect answers. It has limited knowledge of world and events after 2021 and may also occasionally produce harmful instructions or biased content.

These models were trained on vast amounts of data from the internet written by humans, including conversations, so the responses it provides may sound human-like. It is important to keep in mind that this is a direct result of the system's design (i.e. maximizing the similarity between outputs and the dataset the models were trained on) and that such outputs may be inaccurate, untruthful, and otherwise misleading at times.

--What is ChatGPT? Openai help center. (n.d.). Retrieved June 15, 2023, from https://help.openai.com/en/articles/6783457-what-is-chatgpt

ChatGPT is fine-tuned from GPT-3.5, a language model trained to produce text. ChatGPT was optimized for dialogue by using Reinforcement Learning with Human Feedback (RLHF) – a method that uses human demonstrations and preference comparisons to guide the model toward desired behavior.

--What is ChatGPT? Openai help center. (n.d.). Retrieved June 15, 2023, from https://help.openai.com/en/articles/6783457-what-is-chatgpt

There are many products that promise reliable detection of AI-generated work, such as the one created by Turnitin. Instructors and students should be aware that the accuracy of these services has not been confirmed, and AI technology is advancing and changing at an incredibly fast rate. There have been many reports of false positive detections, and there are many sources available online that instruct students how to avoid true positives. 

Understandably, the NU community--and higher education in general--is having anxious conversations about authentic student work in light of AI software, and these conversations should continue with open and honest dialog. Reliable AI detection, however, may never become a realized solution to this issue, and it is recommended that any instructor using AI detecting software does so with care and understanding of its shortcomings.  

ChatGPT and other generative AI tools can and--sometimes do--give users citations (or what appears to be citations). 

Occasionally, ChatGPT may tell users that, as an AI language model, it can't give back accurate citations due to the limitations of its training data. In other instances, ChatGPT may give generally helpful but not very specific websites. And in other cases, ChatGPT has given users very convincing citations that turn out to be completely made up. 

So what does this mean for using ChatGPT to find citations? 

Always approach citations and facts produced by ChatGPT and other AI text generators with critical skepticism. Understand that not everything that these tools generate will be factually true, and that it is up to users to verify the accuracy of anything they receive from these tools themselves. 

How can you determine if a citation you got from an AI tool is real? 

If you do get a citation from an AI tool, you can quickly verify whether it is real with a few search techniques: 

1. Copy and paste the whole citation in a Google search, in a Google Scholar search, and in a library database search. Can you find the citation in a different location? If not, you should be very skeptical. 

2. Find the author on a cite like ResearchGate, Google Scholar Profiles, or try to find their page on the university website where they work. Can you see the title of the work you found from your AI tool on any of these profile pages? If not, you should be even more skeptical. 

3. Collaborate with a librarian! Your NU librarians are information professionals, and we are happy to help you determine whether what you are seeing is an actual resource or not. There are many ways of contacting the library with your questions.