The spirit of my policy on generative AI (genAI), large language models (LLMs), and related technologies is thus: I teach at a large public university because I care about the humans, craft, and philosophy of my discipline as much as I care about the technology of the discipline. Craft and philosophy are, like the humans of the discipline, inherently human. I have no interest in interacting with stochastic content generators, and will not knowingly engage them in the execution of my duties.
Teaching with genAI/LLMs
At the time of this writing I have no intention to use generative AI, large language models, or related technologies in my teaching in any way. All of my course materials are generated by me, personally, or drawn from sources that I believe are trustworthy.
Communicating with genAI/LLMs
I will never send you the output of a generative AI model in human-to-human communication. I believe that this is rude and demeaning. (To wit: if I can ask the LLM, so can you; why send you information you can retrieve yourself? What is it adding to the exchange? What is it detracting from the exchange?)
I will not knowingly read or respond to communication generated by a genAI or LLMs. This includes, but is not limited to:
- Emails
- Documents
- Assignment submissions
- Presentations
If I believe that a communication that I receive was produced using genAI, I will simply delete it without further consideration. In some cases, as I do have a duty to instruct, I may reply to students, specifically, telling them that I will not read their communication and explaining why (probably by linking to this policy).
On the Communication of Prompts
If you wish to communicate with me but do not believe that the quality or professionalism of your human communication is "up to [possibly my] standards," please reconsider. I teach at a public institution; as I frequently say in class, if you already knew all of this, we wouldn't make you do it. Instead of sending your communique through an LLM for refinement or generation, just send me the prompt. I would much rather see your human prompt, which is presumably responsive to your actual concerns, than see the slop that an LLM churns out of it. Your prompt is good enough for me; the genAI output produced by your prompt is not.
Building Knowledge
Remember that we are here to build knowledge. Knowledge lives in the minds and doings of people, not in bits on the wire. We may temporarily store knowledge in bits, or use bits to communicate knowledge, but those bits are sterile and inert. Our true goal is the changes we can make in other people.