Take into consideration the following when using ChatGPT or other AI tools:
Academic Integrity. Make sure to check your syllabus and ask your professor for guidance. We also encourage students to review UGA's Academic Honesty Policy before using AI tools for coursework.
Inconsistency. Because generative AI is probabilistic (it predicts the most likely next word in a sentence), it is liable to be inconsistent and provide different answers for the same prompt. This makes it difficult to evaluate, and hinder its reproducibility, which is a crucial feature of scientific knowledge.
It is not always accurate or reliable. In fact, AI can hallucinate or fabricate information, presenting imaginary or nonsensical statements as facts. As librarians, we regularly come across fake references created by generative AI that simply do not exist. AI is not suited for information that would have dire consequences if it was incorrect (such as health, financial, legal advice, etc.)
Lack of transparency: It does not cite sources or otherwise indicate what in training data has led to the response
Citing AI: It is difficult to know how to cite generative AI, which is predictive based on its training data, and does not generate original ideas. It can also be difficult for your instructor to verify the citation since the same generative AI tool may not always answer the same question the same way every time.
Bias. Tools like ChatGPT are highly sensitive to slight wording changes in the prompts, which can skew responses. They are also limited by their training data, which may have inherent biases and over or under-represent certain voices, perspectives, and accounts of the world. They may have been trained on older and potentially outdated data and may not represent all topics or perspectives equally. For example, ChatGPT 3.5's training data goes up to 2022. ChatGPT4 and other alternatives have access to more recent information.
Further Reading:
Comparing Google Bard with OpenAI’s ChatGPT on political bias, facts, and morality - Brookings Institute
These new tools let you see for yourself how biased AI image models are - MIT Technology Review
Students' Guides to Navigating College in the Artificial Intelligence Era - Elon University & American Association of Colleges
and Universities
