Generative AI at SSU: Citations and Ethics of ChatGPT

What is Generative AI?

Generative artificial intelligence (AI) is a collection of algorithms that can generate content such as text, images, audio, video, and more. A user inputs a prompt, and the AI will serve them a response that fits the context of the words fed into it. To craft these responses, AI typically references a bank of content created by humans - like text from published books, newspapers, websites or images from photographs, paintings, and sketches or audio from music and recorded speeches - that was previously fed to it.

Recent advancements in the amount of content an AI program can be taught and how have resulted in drastic and widespread changes to the availability and capability of AI. However, because recently released AI tools are still new, controversial, and frequently prone to mistakes (also known as AI "hallucinations"), it is best to proceed with caution and transparency when using them in academic and professional settings.

Some generative AI tools include ChatGPT, DALL-E, and Midjourney.

Where Does AI Find Its Information?

Again, AI is trained on a large body of text from across the internet. This could include:

• blogs
• news articles
• social media posts
• e-books
• scanned documents
• anything else you can think of

Even this page could be scanned by AI. So far, developers have not been fully transparent about how sources are selected to train AI and how thoroughly they are incorporated. Once a piece of text is entered as a prompt, it is safest to assume that the AI incorporates that text into its learning and will use it to answer prompts for future users, so be careful not to enter any private or identifying information like your passwords, email, Social Security number, or home address.

When prompted, the algorithm predicts the response that is statistically likely to be the answer the user desires. It then pieces together various parts of this bank of content to craft the response. Because this process incorporates so much information, it is difficult to pinpoint where a piece of text or idea originated. Even if you ask AI to cite its sources, because the program is first and foremost producing the answer it thinks users want instead of the one that is most accurate, these sources may be incorrect. They may not even exist.

Always make sure to double-check the information AI gives you.

You can view a list of some of the sources AI has scanned and incorporate into its processes here.