ChatGPT, LLMs, and AI

ChatGPT (Chat Generative Pre-trained Transformer) is one of several generative AI tools that parse natural human language to generate detailed human-like answers to user questions through a conversational interface. ChatGPT is one of many generative AI tools available, and the number and sophistication of these tools grows daily. Other commonly-used generative AI tools are Google’s Gemini, Microsoft’s CoPilot, and Claude 2, as well as tools built on open source language models like Meta’s Llama 2.

ChatGPT and similar Large Language Models (LLMs) live at the intersection complex and interrelated aspects of artificial intelligence. Take all of the concerns about data on the Internet, compound them with algorithms that prioritize creating language that seems reasonable but is not necessarily factual, then multiply them by complex self-driven learning models that we don’t fully understand. The impact on our social systems and on our educational systems will likely be disruptive, if not transformational. How can faculty, staff, and students thrive amid these changes?

Starting with the basics

All Middlebury faculty, staff, and students should develop a basic understanding of LLMs and other generative AI. The landscape of generative AI is changing rapidly, and a foundational understanding of the tools can help us understand and navigate those changes.


Preparing for the semester

It is important to establish clear expectations for appropriate use of generative AI in the context of your classes. Whether you are banning use of generative AI tools, allowing limited or unlimited use, or encouraging students to decide for themselves what kind of learners they want to be, communicate your expectations and goals clearly to students. You may want to rethink assignments and assessments in your class to intentionally include AI tools or to mitigate their use. Design into your assessments opportunities for students to reflect on and share about their learning, and the role that LLMs are playing (or not) in their learning.

Using generative AI in your class

You can leverage different types of generative AI, including text, image, video, and audio generation, to support student learning in your courses and to support your work. AI text generators, like ChatGPT, can create personalized prompts and case studies for students to discuss. Image generators can quickly produce visuals for student projects or for course materials. Generative video and audio tools open up new multimedia possibilities for your classes. This section provides resources to help faculty implement generative AI as part of their classes and work at Middlebury.

Engaging critical perspectives

There is a lot of hype surrounding generative AI. The hype can obscure or drown out important critical voices that are raising concerns about the impacts of these technologies on humans and our world. We encourage everyone in the Middlebury community to learn more about environmental impacts, labor exploitation, data privacy, security, bias, and other issues pertaining to generative AI.