Considerations when using ChatGPT and other generative AI: Home

We are seeing a dramatic increase in the abilities of artificial intelligence that is available to the public. Changes in the technologies we use change the way we do scholarship. It's necessary to make sure we aren't causing ourselves unwanted and unintended consequences. 

Generative AI is artificial intelligence capable of using an algorithm and a huge database of content to create new content. This may be text, images, audio, video, software code, even knitting patterns. It does this with varying degrees of success, but it is sometimes useful and sometimes hard to distinguish from work created by a person.

In general, use AI when you have approval from your instructor to use AI on an assignment, and are using the AI appropriately in the context of that assignment. 

• The results generated by an AI tool are unique and can't be replicated. This makes them unlike other sources that we typically cite, except personal communications. 
• The results generated by an AI tool can't be copyrighted, because US and international law do not allow copyright protection anything but content created by human beings. 
• AI tool terms of service often claim ownership of the copyright of the prompts input by users. 
• AI tool terms of service rarely have any guarantee that users' data will not be sold or turned over to a government entity. 
• Generative AI tools are known to "hallucinate" (come up with things that are incorrect.) At present, they do not have a way of self-correcting.
• Generative AI tools sometimes come up with scholarly-looking passages complete with citations, but the citations are incorrect.
• A generative AI is only as good as the content it is "trained" on. If biased, bigoted, or incorrect content was put in, biased, bigoted, and incorrect content will come out.