1 of 75

Guiding Students on Appropriate AI Use: Collaborative Approaches

Anna Mills, College of Marin

A presentation for the University of Nebraska, Kearney English Department

August 19, 2025

Licensed CC BY NC 4.0

2 of 75

Welcome! What to expect: One writing teacher’s approaches to addressing AI

  • Yes to teaching with writing:
    • Frame the value of writing
    • Making writing assignments motivating�
  • No to AI offloading: set up layers of guardrails to discourage misuse

3 of 75

Housekeeping

  • Slides, open for comments: https://link.annarmills.com/guiding or scan the QR
  • I’ll pause for questions at the end of each section!

4 of 75

Yes to writing:

Framing the value of writing in an age of AI

5 of 75

Emphasize the purpose of each writing assignment. Why is writing in college important? Not for the product. For the thinking process.

“A fundamental tenet of Writing Across the Curriculum is that writing is a mode of learning. Students develop understanding and insights through the act of writing.

-- The Association for Writing Across the Curriculum

Presentation by Anna Mills, licensed CC BY NC 4.0.

6 of 75

I start the semester by discussing the purpose of learning to write in college. We read the introduction to my (free OER) textbook How Arguments Work.

7 of 75

“What makes writing so valuable?”

“I would argue that academia and the professions need writing because it is our best tool for sharpening our thinking. It helps us slow down and clarify our ideas.”

Introduction, How Arguments Work: A Guide to Writing and Analyzing Texts in College

8 of 75

Writing is collaborative human communication

“We can think of academia as a conversation of many voices that speak to each other across time and place, through the medium of writing…We need each other’s help and input.”

Introduction, How Arguments Work: A Guide to Writing and Analyzing Texts in College

9 of 75

I put the learning purpose at the top of each writing assignment

A sample purpose section from a Summary and Response Essay Overview:

Purpose

  • Practice slow thinking in relation to an interesting reading
  • Understand a short argument deeply
  • Develop your own opinion and response, building on that deep understanding of the argument

10 of 75

Why we need writing practice more than ever in an era when AI can produce text

  • It’s not just that writing is as useful as ever.
  • Writing, reading, and critical thinking are now more important relative to other skills.
  • AI can assist with technical details like coding languages now, but humans still need to be able to clarify and communicate our thinking.

11 of 75

What does it take to use AI effectively and “add value”?

  • You ask the AI system for what you want. So you have to be able to describe in words what you want. So you have to write well.
  • Then you question what it gives you. You revise, reject, add, start over, tweak. So you have to read critically and edit.

12 of 75

Yes to writing:

Designing assignments for intrinsic motivation and supporting writing practice

13 of 75

Let’s double down on what we know: research-based best practices for teaching writing have become yet more crucial

For one overview of principles in writing instruction, see the National Council of Teachers of English 2016 position statement Professional Knowledge for the Teaching of Writing.

14 of 75

Find ways students can make a writing assignment feel meaningful to them

  • Create assignments that have intrinsic meaning and are likely to be motivating.
  • Allow room for creativity, original thinking, and choice about the focus of the assignment.

15 of 75

Build relationships and community as the context for reading and writing

  • Invite peer responses not just on strengths and weaknesses but on what is interesting in each student’s writing
  • Hold conferences with students
  • Offer video feedback
  • Ask students to share audio or video notes about their writing and respond to classmates.

Presentation by Anna Mills, licensed CC BY NC 4.0.

16 of 75

Teach and support writing processes

  • Annotation of readings
  • Prewriting/brainstorming/outliningdrafting
  • Reflections on the thinking and writing process and on feedback
  • Peer review
  • Required tutor visits
  • Revision

17 of 75

Collaborative annotation/social annotation of class readings (try Hypothes.is or Perusall)

  • Students begin to form sentences in conversation with a text as they read.
  • Annotations can be short and informal: there isn’t as much temptation to auto-generate them.
  • Students write to peers in a social-media-like space where they may not feel as much pressure or self consciousness as in an essay draft.

18 of 75

Many of these strategies take time. Here are a few strategies for reducing the labor involved

  • Assign students to visit tutors
  • Assign social annotation of readings with automatic credit for participation in Perusall
  • Give completion credit for process assignments
  • Give feedback on process assignments only by request (with a warm invitation)

18

19 of 75

Three Mentimeter questions on writing pedagogy: what do you do, what might you add, what comments do you want to share?

Use the arrow key to advance through the next three slides in the Mentimeter poll at https://www.menti.com/alej2vex2pt8.�

Or go to Menti.com and enter 2985 9696 to rate each option.

20 of 75

No to AI offloading:

Why we need guardrails for accountability

21 of 75

I focus on the positive, preventative, supportive practices just described. But I don’t find them to be adequate.

“Trusting students not to cheat isn’t fair to the students who don’t cheat no matter what….For all other human behaviors, we understand that people need help sticking to their goals or keeping their promises.” –Dr. Tricia Bertram Gallant, “Crafting Your GenAI & AI Policy Guide”

22 of 75

Let’s acknowledge some reservations around focusing on accountability

  • Most teachers don’t want to focus our time on “policing” students.
  • We want to build trust, not an adversarial relation to students.

23 of 75

But even with the most motivating pedagogy in the world, students will sometimes be tempted to take shortcuts

This recent article “Everyone is Cheating Their Way Through College” in New York Magazine paints a dramatic picture. Even if it’s exaggerated, we’ve got to wonder, how much learning are students missing? How might the perception of widespread cheating will affect the value of course credits and grades?

24 of 75

We need ways to reduce potential learning loss and unfairness.

If we don’t know what is AI and what is the student’s, we can’t tell if the student is learning

25 of 75

Ignoring the impact of GenAI Tools on your Course Learning Outcomes undermines:

  • Teaching: you’re not teaching the skills/knowledge you think you are
  • Learning: the students aren’t learning what we intended
  • Assessment integrity: we aren’t evaluating what we think we are
  • Degree integrity: Students don’t have the knowledge & skills we say they do

This slide is from “Crafting Your GenAI & AI Policy:

A Guide for Instructors” by academic integrity expert Dr. Tricia Bertram-Gallant, shared under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license.

26 of 75

As you’re probably aware, there is no clear perfect way to ensure accountability and prevent AI misuse

  • But that doesn’t mean there’s nothing meaningful we can do.
  • Academic integrity researcher Dr. Phillip Dawson advocates a layered, “swiss cheese” approach. He argues on the AI in Education podcast and on LinkedIn that “many layers stacked (like layers of Swiss cheese) work better than any layer alone…the approach was first applied to academic integrity by Dr. Kiata Rundle with her PhD supervisors Guy Curtis & Joseph Clare.”

27 of 75

Rundle et al argue that multiple imperfect means can be quite effective at reducing incidence of cheating

27

28 of 75

I also find the metaphor of guardrails appealing because it reminds us of danger

29 of 75

What kinds of guardrails/swiss cheese layers might we set up? In the next sections, we’ll look at the following options:

AI Policies

Process observation

  • In-class assessment
  • Video and audio assignments
  • Process assignments
  • Process tracking?

AI Detection?

30 of 75

AI Policy Guardrails

31 of 75

“You should be able to expect clear guidance from your instructor… guidelines should make clear which specific systems or tools are appropriate for any given assignment.”

--Kathryn Conrad in A Blueprint for an AI Bill of Rights for Education, Critical AI (Duke University Press)

32 of 75

AI is a lot to process for students as well as for us…

AI policy can offer

  • Guidance
  • Anxiety relief as it clarifies
  • Empathy for the dilemmas students face today

33 of 75

We don’t have to have the final answers on AI to make a policy: seek help and keep iterating

  • There is plenty of support! I’ll share templates and examples.
  • We can seek input from students as well as colleagues and department or program leadership
  • We can plan to revisit and continue to revise our policies. We’re exploring this territory and seeing how things go…

34 of 75

Sample policies

–From Crafting Your GenAI & AI Policy: A Guide for Instructors by Tricia Bertram Gallant of UC San Diego

A large collection: Classroom Policies for AI Generative Tools, curated by Lance Eaton

35 of 75

Templates and worksheets

36 of 75

Consider a “Tools and Rules” section for each assignment

The idea and phrase come from ESL instructor Julie Carey of Cañada College. Here’s some sample language I use:

  • You are welcome to use grammar-checking software, just be skeptical of the suggestions.
  • No AI for coming up with ideas or text for your essay. The learning happens as you figure out what to say and clarify your points as you put the sentences together.
  • You can use AI to help with research or reading comprehension as long as you do the reading yourself and don't fully trust the AI summaries.
  • You can use AI for feedback on your writing or for ideas about writing strategies.
  • Did you think of another way AI could be useful for this assignment? Please run it by me! I am open to ideas as long as the AI use doesn’t get in the way of learning.

37 of 75

Activity: discuss Joss Fong’s Vox.com video interviews with teachers and students

“AI can do your homework. Now what?

Students and teachers grapple with the rise of the chatbots.”

38 of 75

It’s hard to get specific enough about all possible uses of AI. That will always be a work in progress.

Consider: There are many possible uses beyond auto-generating the whole assignment (i.e. AI for brainstorming, AI for feedback, AI for help with organization, grammar, or genre conventions)

How can we make a policy that accounts for that variety and the variety of ways such uses could impact learning for different assignments?

The AI Assessment Scale developed by Leon Furze and colleagues is helpful; I’d like to share my adapted version.

39 of 75

An AI Assessment Scale (adapted)

  1. No AI: Students demonstrate core skills that do not involve AI.
  2. AI Tutoring-Style Assistance: Student ideas, student words. Students use GenAI to help them think it through and improve it, while questioning what GenAI tells them..
  3. AI Collaboration: Students use some AI ideas and words and quite a few of their own.
  4. Unlimited AI: Students check AI output and are responsible for its quality but don’t necessarily change or add to it.

Note: This is Anna Mills’ short version. See also the version that explains each one. This scale is adapted by Anna Mills from the Perkins, Furze, Roe, and McVaugh (2024). The AI Assessment Scale. CC BY NC SA. I have edited it and incorporated aspects of the earlier version by the same authors.

40 of 75

Further guidance on developing AI policies:

“Reinforcing Academic Integrity in the Age of AI: A Guide for Instructors” by Tricia Bertram Gallant of University of California, San Diego

41 of 75

Are you aware of any college, program, or department policies or guidance on AI use? Describe where to find any and/or what they say.

Use the arrow key to advance to the next slide in the Mentimeter poll at https://www.menti.com/alej2vex2pt8.�

Or go to Menti.com and enter 2985 9696 to rate each option.

42 of 75

Process Observation Guardrails

43 of 75

Observing students’ process as they write can take various forms. Which, if any, are intrusive?

  • In-class writing
  • Video and audio assignments where students talk about their writing
  • Process assignments
  • Process tracking

44 of 75

Asking students to write in class at times can help but it’s not ideal to do all writing that way

We would have to give up too much class time to other essential learning activities.

Students wouldn’t have enough time and flexibility to experience an extended slow thinking writing process.

Handwritten timed work may not be a good sole measure for high stakes assessment.

44

45 of 75

Still, some in-class writing, including digital writing, could be helpful

In-class writing doesn’t have to be the sole mode of assessment to help reduce AI misuse.

Used judiciously, it may allow us to get to know student voice so we inquire when there’s a discrepancy.

Of course, many students have anxiety about performance in class under time pressure or may find writing in class distracting.

46 of 75

Proctoring centers

Another option is to proctor student writing processes in writing centers or other proctoring centers.

I’ve heard one anecdote of a professor organizing this: Luke Fernandez at Weber State University in Utah arranged to require his students to write their essays at the writing center.

47 of 75

As we discussed, low stakes process assignments are good pedagogy.

  • Each one can be autogenerated, yes.
  • But they do make offloading to AI more of a hassle.

48 of 75

Oral assessments can complement written ones

  • In person
  • Zoom meetings
  • Video and audio recording assignments. I ask students to record a 1-3 minute video in Canvas Studio talking about their initial ideas for an essay or their writing process. They share those to a Canvas discussion and comment on each other’s videos.

49 of 75

Caution: audio and video can be deepfaked pretty easily, for free

With platforms like HeyGen, anyone can upload a brief video of themself and create an avatar that will read AI scripts in AI-generated videos.

See this sample HeyGen X post where Wharton business school professor Ethan Mollick’s avatar speaks German (Mollick doesn’t).

50 of 75

Process tracking means ask students to share their document history

A user friendly version history can show how much time was spent, what was copied and pasted, what was edited, and more.

Note: process tracking is not foolproof. AI text can be retyped, either by hand or by free autotypers (alas).

One sample ad: “Need help typing your work? Duey.ai's Auto Typer for Google Docs helps you save time and effort by letting you create a Google Doc with a full Version History.”

51 of 75

Various apps and extensions allow replay and analysis of a document history, including cut/paste and time spent. Cost is $0 on up…

52 of 75

How much teacher labor does this involve? Some approaches involve some setup and then minimal labor.

If I ask students to share edit access to a Google Doc and enable a Chrome process tracking extension, I see a ribbon with process information across the top. No extra clicks or windows are needed.

I typically scan basic stats such as time spent typing in the document and large copy/paste events.

If something in those or in the style seems odd, I look at the full report, but this is more rare.

53 of 75

ProcessFeedback.org: nonprofit, educator created (Dr. Badri Adhikari, University of Missouri)

53

54 of 75

ProcessFeedback.org Chrome extension offers a ribbon across the top of a Google Doc that tells the number of edits

If you click “Explore Process” you get a lengthy report that shows time spent, the history of what was copied and pasted, and much more (see this sample).

The level of detail may send the message to students that the value of the assignment lies in the process of working in the document.

54

55 of 75

For a less adversarial setup, let the student generate and share/reflect on their process report

  • Students may feel a greater sense of control
  • They can review the report and be able to recognize and correct any inappropriate use or failure to cite.
  • They can also proactively explain any unsourced text they didn’t type.

56 of 75

Grammarly Authorship

  • Can work in Google Docs or Word or in the Grammarly App
  • Grammarly Authorship notes when Grammarly itself has been used to edit, revise, or generate text, so this becomes part of our class conversation on how to use Grammarly and be skeptical of its suggestions.

57 of 75

Screenshot from a Grammarly Authorship report

58 of 75

The user can decide if the authorship report includes a replay of the whole drafting process. I don’t require students to share this. I pay more attention to time spent.

59 of 75

Grammarly authorship reports attempt to label text by origin

60 of 75

Steps to try Grammarly Authorship

  • You need a Grammarly account. Free is fine.
  • Install the Grammarly Extension for Google Chrome
  • Open a Google Doc in Chrome
  • Click the thumbprint icon in the lower left corner:
  • Select “Track writing activity”
  • Click the thumbprint again and choose “See report.”

61 of 75

I ask students to reflect when they share a process report

  • If there is any AI-generated text, have you labeled it appropriately in the essay and cited or acknowledged its origin?
  • If there is any text labeled as “copied from an unknown source,” where did it come from?
  • If any text is copied from a website, have you quoted and cited it correctly?

62 of 75

Allow for alternative processes

My message to students: “Think of this as an online version of in-class writing that allows you more flexibility. It gives me a way to understand your writing process, both to encourage academic honesty and to encourage reflection on the writing process itself. If you have concerns or do not feel comfortable sharing your process in this way, please let me know and we can meet and work out another plan.

63 of 75

The discussions of whether process tracking is surveillance are ongoing…

See “What is process tracking and how is it used to deter AI misuse?” a blog post from the MLA-CCCC Joint Task Force on Writing and AI. In it, I explain why I’m using process tracking, and the other task force members disagree with me.

64 of 75

AI detection guardrails

65 of 75

We all have the experience of reading text and trying to decide if we think it was AI generated.

How well can we really distinguish AI text from student writing?

Our intuition about what is AI text may help us initiate important conversations if we know the student’s writing well. But let’s be wary of our own unconscious bias and possible overconfidence. AI can imitate student writing styles.

66 of 75

Studies on the extent to which instructors can differentiate AI from student writing suggest we are quite unreliable (Darn!)

67 of 75

Could software be better than we are at identifying AI patterns in text? If so, is it ethical to use such software?

  • Detection software could reduce AI misuse by deterring it
  • Does it create or reinforce an adversarial relationship between teachers and students?
  • Will it lead to false accusations? A false sense of security?

68 of 75

Is AI detection biased against English language learners? It’s unclear

69 of 75

Note: free software exists that rephrases AI text to “humanize” it and get around AI detectors

70 of 75

The accuracy of detectors is constantly changing as AI and detectors change. Right now, Turnitin seems to be ahead.

“Turnitin turned out to be the most accurate and consistent one, with a 100% AI score even with the adversarial techniques.”

It even did pretty well on AI text that was edited or paraphrased. Researchers used “three adversarial techniques (edited through Grammarly, paraphrased through Quillbot, and 10%-20% editing by a human expert).. The study found that the four AI-detection tools showed inconsistent AI scores - from very high by Turnitin (almost perfect) to very low by Writer AI. ”

AI vs AI: How effective are Turnitin, ZeroGPT, GPTZero, and Writer AI in detecting text generated by ChatGPT, Perplexity, and Gemini? (Malik and Amjad 2025)

71 of 75

“For those who decide to use AI detectors, please consider the following questions”

From the MLA/CCCC Task Force on Writing and AI working paper on policy development. (The Modern Language Association and the Conference on College Composition and Communication are the professional associations for writing, language, and literature faculty in American higher ed.)

• What steps have you taken to substantiate a positive detection?

• What other kinds of engagement with the student’s writing affirm your decision to assign a failing grade outside the AI detector’s claim that the text was AI generated?”

72 of 75

I am gravitating toward non-punitive, cautious ways of using detection. For more discussion of this, see

72

73 of 75

Three mentimeter questions on guardrails: what do you do, what will you add, what do you need to move forward?

Use the arrow key to advance through the next three slides in the Mentimeter poll at https://www.menti.com/alej2vex2pt8.�

Or go to Menti.com and enter 2985 9696 to rate each option.

74 of 75

To sum up, here’s my approach, and I hope it stimulates your process as you find your own!

  • Yes to teaching with writing:
    • Frame the value of writing
    • Making writing assignments motivating�
  • No to AI offloading: set up layers of guardrails to discourage misuse

75 of 75

Questions or comments?�Thank you, and feel free to get in touch!

AnnaRMills.com

LinkedIn

Slides, open for commenting:

https://link.annarmills.com/guiding�Or scan the QR code

armills@marin.edu

This presentation is shared under a CC BY NC 4.0 license.