Generative artificial intelligence is rapidly reshaping the academic landscape, and it'll likely be a heated topic of debate (if it hasn't been already) in courses you take or teach during your time at URI. This guide will give you a brief introduction to AI tools, covering how they work and how to use them effectively as well as their potential pitfalls to help you be a savvy and ethical user of AI.

ChatGPT 4, response to “create a small banner image to illustrate artificial intelligence,” January 4, 2024, https://chat.openai.com.

AI tools are not search engines. While they might feel similar when you use them - you can type a question and get an answer from both - they actually function very differently.

AI tools generate output based on training data (the images and text used to train the tool enables it to shape a likely response), while search engines crawl the web to find sources that contain potential matches for a search query. The crucial difference being that a search engine connects you to published material that exists online while generative AI creates a new result based on the data it was trained on. Because AI tools are creating new content rather than quoting or referencing information from a specific, verifiable source, it's essential to critically evaluate AI output before using it in any way.

To break things down further, Wikipedia provides a brief definition that is helpful to understanding artificial intelligence tools:

Generative artificial intelligence (generative AI, GAI, or GenAI^[1]) is artificial intelligence capable of generating text, images, or other media, using generative models.^[2][3][4] Generative AI models learn the patterns and structure of their input training data and then generate new data that has similar characteristics.^[5][6] [see "Generative artificial intelligence," Wikipedia, The Free Encyclopedia, accessed January 4, 2023]

AI tools for writing, generating images, or creating music work fairly similarly.

First, developers of these tools collect text, images, music, or audio from various sources to use as training data. Sometimes, this collection is called a "corpus."

Next, developers create programs and interfaces that allow users to explore the contents of the corpus. Some popular interfaces are chatbots such as ChatGPT, DALL-E, and Midjourney, where the user provides instructions as though conversing with another person.

Finally, the user prompts the tool with what they're looking for, for example, asking for ideas for topics to research for a class or creating an image that would illustrate a point for a presentation. The AI tool then provides output for the user based on the prompt, the corpus, and the algorithms used in the chatbot program.