My thoughts on AI
Following Andrew Heiss and many others, this is my AI Manifesto. This is what I put in my course syllabi as of August 2026. Because I work at BYU, I have mostly integrated statements made at BYU and there is some religious connection. For previous versions, see here.
What has BYU said about AI?
Here’s what the linguistics department policies are regarding the use of AI:
The Department of Linguistics supports the university-level statement on the use of generative artificial intelligence (AI) for academic and professional purposes. The department also expects students, faculty, and staff to follow the Academic Honesty Policy. In addition, the department expects graduate students to follow the Graduate Studies Statement on Use of AI.
The department encourages its instructors to clearly delimit what their expectations are with respect to the use or non-use of generative AI in their classes, and their students are encouraged to follow those expectations. Use of generative AI that is not consistent with the expectations of a given instructor is considered a violation of the Academic Honesty Policy.
Please also see the Teaching With Artificial Intelligence document put together by the College of Humanities Task Force. It is more intended for faculty, but it is still worth a read for students.
In April 2026, BYU had a meeting for faculty about AI in teaching. This page has resources from that meeting. I will admit that my thoughts on AI have changed pretty drastically because of that meeting. I am in the process of integrating those ideas into my teaching.
What is responsible use of AI?
AI can be an extremely helpful tool and when used responsibly, can enhance your education rather than replace it. Proper use of AI means treating it primarily as a tool to augment your own work rather than to replace it.
A few years, ago I was in a position to create a new course. When deciding which topics to include in the course, I consulted potential textbooks to give me ideas and surveyed students who might be interested in the course before coming up with my own tentative list. Only then did I turn to AI to come up with a list of topics. As it turns out, it provided me a very similar list to what I had already which served as a confirmation to me that my list was good.1 I knew enough about the topic before turning to AI so that I could spot its bad recommendations. If it had provided me with a new idea, I would not have accepted it at face value but I would have done some additional research to decide whether I should include it. In no way would I have accepted the output at face value.
1 I don’t want to compare AI to the Spirit, but D&C 9:8 comes to mind here.
2 I thank Dan Dewey for the following example.
There are times when AI could be used to help with coursework.2 Let’s say you struggle with writing in an academic register, either because English is not your first language or because you’re just not comfortable with this style of writing. An inappropriate use of AI is to let it complete your assignment for you. An appropriate use of AI would be to give it a prompt like this:
“Here is a paragraph I am writing for a college term paper. [Provide the paragraph]. I am struggling with two things: English prepositions and English tense [or appropriate writing style, or whatever]. Please correct any errors you can find in either of these. Mark the corrections for me. Then give 3–4 tips for prepositions and tense based on the corrections.”
With this method, you are using AI to help you learn and get better—augmenting what you can already do—rather than having it do your work for you. Keith (2026)3 points out that having AI complete an assignment for you puts you in a passive role since you are just treating it as a production assistant. Having it teach you how to improve your writing puts you in a partnership role. He proposes there are far better ways to us AI that allow you to maintain your agency while letting it check, explore, stress-test, and reframe your work. I encourage you to at least skim Keith (2026) to learn more about that.
3 Mark Keith is an associate professor of Information Systems at BYU.
There are times when it is not appropriate to use AI. Unfortunately, students and faculty disagree on what is considered appropriate use of AI. Here are the results of a survey of BYU linguistics graduate students in Fall 2025. The survey showed that most students used AI at least once a week for academic work, that they expect professors to teach how to use AI, and that AI is appropriate in a variety of situations. While AI use varies among faculty, most of us are on the more conservative side of AI use in academic work and believe that students should not use it for any academic work.
Regarding using Large Language Models (LLMs) specifically to generate text in order to replace the work of writing, please do not do that. Andrew Heiss, a international policies professor at Georgetown University in Qatar, has a nice explanation about why not here.4 It basically comes down to this: writing is an essential part of learning and if you skip it, you’re not learning. Plus, I would 100% rather see sincere thoughts conveyed in your own voice (even if it has imperfect spelling, grammar, organization, and punctuation) than artificial thoughts in pristine writing. If you’re not going to spend the time to write, why should I or the TA spend the time reading and grading it?
4 He’s a member of my church too and has a sidenote explaining the colorful language!)
At this point, I assume that you, like me, regularly use AI. But one issue that I am now seeing with student work though is that even if they did not use AI for academic writing (or at least claim to), their writing style is starting to mimic it. Think about it: how much AI-generated content have you read about topics related to linguistics? Now think about how much actual linguistic writing you’ve read. If AI-generated writing makes up a disproportionately large amount of what you read about linguistics… well that’s not good for a lot of reasons. But one is that when it comes time for you to write academically, what you’re most familiar with is AI-generated writing and so that’s what you’ll mimic. I therefore again strongly urge you to not use AI for generating text, even if it’s unrelated to homework, and to read as much academic writing written by actual humans as you can.
So what is the policy in this course?
Here is a paragraph from BYU’s Generative Artificial Intelligence folks that I think best describes how you should use AI:
Before using GenAI on an assignment, you must ask your professor for permission if they have not already explicitly stated that it is permitted. If the professor has not said anything about GenAI use for either the assignment or in the syllabus, assume that using GenAI is not permitted. For each class, the professor is the arbiter of academic integrity. Use [your professor] as your first recourse when determining what is acceptable or unacceptable for an assignment.
So, in line with department policy of clearly delimiting what my expectations are with AI, and in line with students’ desires that we work with students to develop generative AI skills for academic work, here are my expectations:
You are welcome to use AI for some tasks in this course. These tasks from Keith’s (2026) document, “A Student’s Complete Guide to Thinking Better with Artificial Intelligence.” I encourage you to read through that guide for more ideas on what these roles look like and how you can use them. Of the eight ways to use AI, the six corresponding to his “Partnership” and “Agency” tiers are permissible in this course:
- a tutor (“Help me understand this. Don’t just tell me—ask me questions.”)
- a collaborative problem solver (“Here’s my thinking so far. What am I missing?”)
- a verification agent (“I think this is right, but check my work.)
- a creative expander (“Five approaches I haven’t considered?”)
- a critical challenger (“Attack my argument. Where is it weakest?”)
- a problem setter (“Am I even asking the right question?”)
While I probably won’t know otherwise, I do not want you to use AI for the “Passivity” roles:
- an oracle (“What’s the answer to this question?”)
- a production assistant (“Write this essay / Summarize this chapter”)
I want to emphasize that there is so much learning that happens through the process of reading and writing. By offloading that task onto a digital tool, you are denying yourself those learning opportunities.
In each homework assignments, I will clearly state how AI may be used for that assignment. I will remind you of these example uses and will even provide example prompts you could use to give you an idea of how to use AI at this level. If you do use AI, part of the assignment will be to then provide a statement stating how it helped you learn what the assignment was deisgned for you to learn.
This is a drastic departure from my previous stance on AI. Again, my views are constantly evolving. I hope you can be wise and honest in your use of AI in this course.