Citing Your Sources

Generative artificial intelligence (AI) is an emerging technology that is rapidly evolving. Tools such as ChatGPT and Google Gemini (previously Bard) offer new ways to interact with information but require a critical approach to ensure accuracy and reliability.

This brief introduction serves to assist you in developing a critical approach when engaging with generative AI tools, particularly concerning their relationship with information literacy and citing them as sources.

Where does the information come from?

As an AI language model, ChatGPT's responses are generated based on a mixture of licensed data, data created by human trainers, and publicly available data. The entire body of text that has trained ChatGPT is unknown.

When ChatGPT responds to a question, it doesn't immediately provide a specific reference for the source of information. This is because its responses are generated based on a diverse range of predictive language learned from various sources, rather than relying on a single origin. As a result, it is difficult to trace the response back to a singular source or determine its exact origin.

Can ChatGPT provide references?

ChatGPT gave the following answer:

"As an AI language model, I don't have direct access to external sources such as databases or the internet . . . I do not have the ability to provide real-time citations or references for specific claims. If you have specific information or require references, it's always a good idea to consult primary sources, conduct research, or refer to trusted external sources for accurate and verified information." [ChatGPT response to a prompt about references]. https://chat.openai.com/

In short, ChatGPT cannot provide accurate references. ChatGPT cannot provide accurate references. While it generates responses based on extensive data, it does not access specific sources for each response. Therefore, it may produce plausible but incorrect citations, or hallucinate citations. Always verify information from AI with primary sources or trusted external references. This can be an issue with other generative AI tools, like Google Gemini, as well.

How current is the information ChatGPT provides?

Training Data Cutoff:

• ChatGPT-3.5: The information this model uses is only current up until September 2021. Anything that happened after that isn’t included in the responses you get from ChatGPT-3.5.
• Newer Models: There are newer versions like GPT-4, which might have more recent data and even some real-time information capabilities through plugins. But to know exactly what’s included, you’d need to check the latest updates from OpenAI.

Best practices include always cross-verifying information from ChatGPT with current, reliable sources and referring to primary sources or official publications for critical information. For the most accurate and recent details, you can visit OpenAI’s blog.

Google Gemini continuously draws information from the internet in real-time. This allows Gemini to provide up-to-date answers by leveraging Google's search capabilities, making it a more dynamic tool for accessing current information​ [Gemini response to a prompt about training cut-off]. https://gemini.google.com/app. However, this does not mean Gemini will still be wholly accurate.

 This page was adapted by Erica Huff from AI, ChatGPT, and the Library Libguide by Amy Scheelke for Salt Lake Community College under a Creative Commons Attribution-Non Commercial-Share Alike 4.0 International License, except where otherwise noted.