This guide pulls heavily (with permission) from the Yale Poorvu Center for Teaching and Learning guidance on AI in the classroom.
The ethics of using AI are constantly evolving. Here are some links to general considerations:
Here are some links to considerations for ethical considerations for AI in Higher Education:
AI chatbots respond to user-submitted text with (sometimes) relevant content expressed in plain language. Chatbots remember all text that a user has entered in an ongoing conversation, and they are capable of revising their responses when prompted. While the media at large and most people refer to these as AI, they are actually much less sophisticated. It was more accurate to call these programs predictive text machines, or chatbots. These bots are not programmed to make logical or even informative responses. Instead, the chat bot is built simply to provide responses that keep the user engaged with the bot and submitting prompts to it.
AI chatbots operate on programs called large language models. LLMs are programmed by being fed large bodies of text (e.g. the entire internet) and trained to predict the most relevant sequence of words in response to a prompt. As a result their responses reflect the biases and limitations of the material they are trained on. Put differently, AI chatbots produce text by calculating the probability that it will be relevant to the prompt a user has submitted. As a result, the responses produced by AI chatbots tend to reflect consensus understandings, including any biases and inaccuracies that inform those positions. As an example if you were to ask ChatGPT or a similar program to give you the history of a specific disease it would be pulling the information from anywhere it is freely available on the web, which doesn’t include scholarly articles behind paywalls but would include all of the random internet conversation about said topic.
Chatbots do have a few areas in which they can excel however, because AI chatbots predict words based on probabilities, they are not capable of original thought. This means that chatbots can be good at responding to simple prompts, summarization, mathematics and scientific questions where answers are already widely known, or supporting one basic position over another. Asking the chatbot further questions can result in a slightly more nuanced response, as the bot will continue to learn and adapt to what the user is looking for by analyzing the prompts it is given.
The best way to learn about AI chatbots is to try them yourself. Signing up for an account at ChatGPT(link is external), Google Bard(link is external), or BingAI (requires Microsoft Edge browser) will allow you to explore how these programs work and help you better understand how your students might be using them. Try a couple of different prompts, try to place yourself in the mind of your students and make the sort of prompts you believe they would make. Here are a few ideas to get your started.
