1 of 119

For such a time as this

January 30, 2026

Tamara Tate, PhD

Associate Director, Digital Learning Lab

University of California, Irvine

2 of 119

3 of 119

4 of 119

5 of 119

6 of 119

7 of 119

Let’s start by introducing ourselves in chat

What do you teach & where?

How much have you used AI?

  1. What’s AI?
  2. I use it occasionally
  3. I use it regularly to create my curriculum and resources
  4. I have integrated it into my course

How do you feel about AI?

  1. I’m strongly opposed to its use by anyone
  2. I’m strongly opposed to student use in my class
  3. I’m open to considering use, but have some concerns
  4. I’m exhausted even thinking about it
  5. I’m curious
  6. I’m excited by integrating it into education

8 of 119

Why did I ask you that?

9 of 119

Let’s start by introducing ourselves in chat

What do you teach & where?

How much have you used AI?

  • What’s AI?
  • I use it occasionally
  • I use it regularly to create my curriculum and resources
  • I have integrated it into my course

How do you feel about AI?

  • I’m strongly opposed to its use by anyone
  • I’m strongly opposed to student use in my class
  • I’m open to considering use, but have some concerns
  • I’m exhausted even thinking about it
  • I’m curious
  • I’m excited by integrating it into education

Context matters

We start by assessing where our students are at when they come to us

We are seeing that this impacts if and how instructors use AI in their classrooms, along with pedagogical beliefs

We also need to hold space for exhaustion, overwhelm, grief, anger, fear …. ours & our students’

10 of 119

11 of 119

Why should I care?

    • Is generative AI already impacting your course?

12 of 119

Why should I care?

    • Is generative AI already impacting your course?
    • Are there equitable issues to consider?

13 of 119

Why should I care?

    • Is generative AI already impacting your course?
    • Are there equitable issues to consider?
    • Do I have an obligation to teach students how to use generative AI?

14 of 119

The “rich get richer” contradiction

AI is incredibly powerful for assisting communication

To get the most out of it, you need to know how to prompt it well, critically evaluate its output, and edit and incorporate it into your work

15 of 119

The “with or without” contradiction

If students never learn to use AI, they will be at a disadvantage in their study and careers.

If they use AI too much and too early, they will also be at a disadvantage as they will be robbed of foundational skills necessary to use it well.

16 of 119

Imitation contradiction

English learners are constantly told they need to imitate native speakers’ or expert writers’ language use.

But when they borrow exact phrases from sources, they can be accused of plagiarism.

17 of 119

18 of 119

If generative AI is unavoidable in the world

  • [All] Students need to be taught AI literacy
  • This is best done through authentic use of AI
  • Writing with generative AI is a common use of AI (24% of ChatGPT)
  • Using generative AI under the guidance of an educator increases the likelihood of students learning to use it productively and responsibly

a set of competencies that enables individuals to critically evaluate AI technologies; communicate and collaborate effectively with AI; and use AI as a tool online, at home, and in the workplace

-- Long & Magerko, 2020

19 of 119

But if we have AI do we even need to write?

In the chat:

Why do YOU write?

20 of 119

But if we have AI do we even need to write?

We write to communicate across space and time

Human connection may be a value (e.g., fan letter)

But not always (e.g., grocery list)

We write to help us think

Synthesize information (e.g., learning)

Precision of language forces us to refine our thoughts

We can string bundles of thoughts together that don’t otherwise fit in our short term memory

But not always (e.g., most administrative paperwork)

21 of 119

Tools Shape Thinking (and Writing)

From Socrates to AI

Every new tool transforms the way humans write and think.

In the chat:

    • How does writing with a pencil vs. a pen change your writing process?
    • Google Doc?

Affordances? Limitations? Sociocultural impact?

22 of 119

So we need to learn to determine

When humanness matters

When using AI offloads key skill building or scaffolds learning

What role AI might play in the writing process, without sacrificing learning or humanness

23 of 119

Back to basics

24 of 119

How do we begin?

25 of 119

Understanding the learning goals will drive the evaluation of if / when / how to use generative AI

Reflections from the Classroom

26 of 119

Understanding the learning goals will drive the evaluation of if / when / how to use generative AI

Reflections from the Classroom

Purpose of writing:

    • Is it to communicate?
    • Is it to support thinking?
    • Is it to show what we know (proof of subject matter knowledge acquisition)?

27 of 119

Have you looked at your SLOs lately?

28 of 119

Student Learning Outcome (SLO)

Related assignments

or activity

AI disruption (1-5 with 5 being high)

Should the SLO change?

Should the assignment change?

Example: Organize Oral Communication

Speech Outlines

5

Probably

Yes

AI Disruption & Revision Tables

Student Learning Outcome (SLO)

Rationale for change

Proposed SLO

Stakeholders

Starting Point

Example: Organize Oral Communication

Outlining speeches is easily done by AI. This will be how presentations are prepared in the workforce or public service

Evaluate and improve oral communication frameworks

Other instructors

CMST

CAB

Program Curriculum Committee and then CAB

29 of 119

30 of 119

Assignment

Effectiveness of AI at performing task (1-5)

What AI programs can do this?

Output Example

(example of what AI has produced when you input your assignment into an generative ai)

Example: Formal outlines for speeches

5

ChatGPT, almost any large language model

Assignment

Rationale for change

Alternative

Associated SLO Change?

Stakeholders

Example: Formal outlines for speeches

Outlining speeches is easily done by AI.

Use prompt engineering to create outlines. Students evaluate and improve outlines.

Yes

Other instructors

Map quizzes

Little point to ask students to remember things easily found via A.I.; like the calculator. Caveat: no tech, then what?

Scavenger Hunt - screenshot/s of places on map of relevant region/s with comments about the place/region

Yes

Students - value for tuition

Essay assignments

Strong likelihood that A.I. will be asked to generate written compositions

Podcast and/or video creation - Record a conversation with someone outside of class discussing course material; show what you have learned

Yes

Students - value for tuition

AI Disruption and Revision Tables for Assignments

31 of 119

Why we are writing?

If we know why we are writing, we can:

  • Make sure the student learning objective is the focus of the activity/assessment/assignment
  • Decide when, if, how AI might be a problem
  • Decide when, if, how AI might be a tool

32 of 119

Understanding the learning goals will drive the evaluation of if / when / how to use generative AI

Reflections from the Classroom

Purpose of writing:

    • Is it to communicate?
    • Is it to support thinking?
    • Is it to show what we know (proof of subject matter knowledge acquisition)?

33 of 119

Writing to show what we know

Susceptible to off loading with AI: Academic integrity challenge

Options:

    • Do it in person without computers?
    • Oral assessment
    • SRL:
      • Focus on Process > Product
      • Low-stakes, e.g., few points or yes/no grading, multiple checkpoints, plus EF scaffolding
    • UDL: Images, videos, podcasts, posters …..
    • Can you turn the tables?
      • Have them correct AI-produced summary, for example?
      • Can the AI prompt show the learning? E.g., instructions for the image or create a persona

Probably the hardest to make AI resistant

34 of 119

So if you need to revise some things ….

35 of 119

  • Authentic product
    • Workforce prep
    • Engagement
  • Authentic process
    • Workforce prep
    • Engagement
  • Agency
  • Modality
    • Engagement
    • AI resistant

36 of 119

  • Explicit instruction
  • Transparency
  • Modeling (I do, we do, you do)
  • Process over product
  • Smaller assignments over the course of the product creation

37 of 119

Principles of Instructional Scaffolding and Tutorial Assistance

  • Reduce instruction into manageable chunks
  • Concentrate the student’s attention on the task
  • Provide models of what is expected
  • Provide extended opportunities to practice in new situations

Jerome Bruner, 1978

38 of 119

Emerging best practices:

Building good habits

39 of 119

Think First

You are more creative, interesting, diverse, and aware of the desired outcome than any AI can be. Think about what you want to write before you start – you only have one chance to think before being influenced by the AI.

“Good Enough” Prompting

Harness prompting techniques like using personas, goals, and details + content knowledge, but if all else fails, use a conversational tone.

Be the Boss

Remember you are the boss of the AI. Ask it questions, make it clarify, expand, or revise things – iterative prompting allows you to define what is needed. Make the output fit your needs and desires.

Reflect

Taking a moment to consider what worked, what didn’t, and what could be done differently allows for improvement and agency.

Corroborate & Interrogate

Corroborate the information AI provides with other reliable sources. Interrogate the output: is there bias you need to address? The human is ultimately responsible for any output they use.

Human-Driven GenAI Use

40 of 119

Think First

“Good Enough” Prompting

Be the Boss

Reflect

Corroborate & Interrogate

Human-Driven GenAI Use

Reflect

Self regulation ends with reflection--did I meet my goals? What did I learn in the process? Should I do anything different? Do I want to adjust my goals next time?

Set Goals

Self regulation begins with an end in mind, what are my goals for using generative AI? What are my intentions?

Self-Monitor

While I’m using generative AI, I’m checking whether the output fits my needs, looking for bias, critically evaluating the content,

Self-Regulated Learning

41 of 119

Think first

Like in reading when we tap into background knowledge.

Strength based: respects what the student already knows.

Keeps the student in charge of the AI.

ALSO: Because of its fluency, language patterns, and good old psychological realities, once you start down the AI-driven path, you rarely diverge.

SRL: Setting goals/intentions needs to be a habit

42 of 119

Think first: What’s that look like in practice?

43 of 119

What if our SLO is related to revision?

AI-Assisted Peer Review

    • Read the writing. Make some notes on things you notice. Consider the assignment and rubric.
    • Prompt the AI to provide feedback.
    • Consider the AI output for accuracy and clarity.
    • Consolidate your thoughts and the additional information from the AI and give your peer consolidated feedback.
    • Consider what you did well, where the AI helped, and what the AI got wrong or missed.

Example

44 of 119

Good enough prompting

You and your students don’t need to be prompt engineers.

Models are improving quickly and dramatically.

Get started, when it doubt treat it like a conversation.

45 of 119

Where are the books on animals?

Where are the books on birds?

Where are the books on eagles?

Good enough prompting is easier if you think first about what you want

46 of 119

A couple of tips when prompting

  • Role or personas
  • Goal
  • Details
  • Context

What do you know that the AI doesn’t but might find useful?

  • Assignment requirements
  • Course content, level
  • What you already know
  • What things you find interesting

47 of 119

Example: Practicing prompting in class

  1. [Can be done whole class] Students receive (or create) a weak prompt (e.g., Write an essay about why AI is bad). They predict what kind and quality of output they will get. They enter the prompt into the AI and evaluate the output.
  2. In pairs or small groups, students revise the prompts using the information on the prior slide. They enter the prompt and evaluate the output, comparing it to the initial output. They discuss what made the second prompt more effective.
  3. Coming back together, the class discusses what they learned and what they will use going forward to improve the quality of the AI output. Teacher points out how they use language precisely, what content knowledge they brought to the task to improve the output, and reminds them how AI predicts based on language it is given.

Adapted from Karchmer-Klein, 2026, p. 84

48 of 119

Example: Asking better questions

  1. Choose a debatable topic (e.g., is AI beneficial for education?)
  2. Students write an initial prompt to get AI to provide an answer. They input it into the AI and evaluate the output. Is the response biased? Did it consider multiple perspectives? What could make the prompt better?
  3. Students revise their prompts to encourage a more balanced, evidence based response (e.g., adding a request for benefits and drawbacks, supporting evidence). They analyze the AI output to understand how it changed, whether the new response was more balanced, more evidence-based. They consider what helped reduce the initial AI bias.
  4. Students swap their prompts with a peer and try out the peer’s prompts. They analyze the output to determine whether the new prompt led to a balanced, thoughtful response; consider how the prompt could be further improved.
  5. Discuss as a class how the question you ask impacts the response you get. If you ask “why is AI bad?” you get a different output than if you ask “why is AI good?”--students need to think first about what a good, useful output will be and ask for it explicitly.

Adapted from Karchmer-Klein, 2026, p. 83

49 of 119

Be the Boss -- Iterative Prompting

  • Clarification
  • Correction
  • Constraint
  • Expansion
  • Concision

Some resources: DLL, PAIRR, Yost

50 of 119

Be the boss: What’s that look like?

  1. Summarize content knowledge or text
  2. Have students evaluate the initial input
  3. Students then push back by having AI change something using the prior sentence frames

51 of 119

Initial AI output

  • Thoughts?
  • On the surface, AI usually looks good at first
    • Why?
    • What can we do if we aren’t content area experts?

52 of 119

Try to ask a critical question of the output

Even if (especially if?) you’re not a content area expert

  • How would X criticize this document?
  • What would be the counterargument?
  • What is the weakest part of this argument?
  • Who is missing from this story?

53 of 119

Corroborate

Important habit to get into when using generative AI

  • They are built to chat, not gather facts
  • Good critical thinking anyway
  • Always tie back to original sources
  • Triangulate across multiple sources
  • Lateral reading on the internet like professional fact checkers

54 of 119

Corroboration: What’s that look like?

  • Same prompt we just used -- summarize content knowledge or text

Students work together or on their own to make the summary more accurate, precise, clear

  • Students use AI to generate a study guide, then “fact check” it
  • Student use AI to generate content with source citations, then look up the sources to fact check the evidence
  • Option: Students color code AI text with green = accurate and important; yellow = vague or overly general; red = missing or misleading information.
  • Consider: For some assignments, students must include a print out of the source circling where it says what they are citing it for.

55 of 119

Reflection

Building metacognition, self regulation, executive function, habits of mind

Student agency

Build transferable skills

Part of every assignment using AI:

What did you do and how did it work? What would you do differently in the future?

56 of 119

Students + AI requires AI literacy

  • How AI works
  • AI’s limitations
  • Good enough prompting
  • Pushing back
  • Staying in charge of the work

57 of 119

58 of 119

Should generative AI have a role in writing instruction?

59 of 119

Generative AI might play a role throughout the process of writing

Reviewing

  • Evaluating
  • Editing
  • Revising

Planning

  • Setting goals
  • Generating ideas
  • Organizing ideas

Drafting

  • Putting the writing plan into action

Graham, S., & Sandmel, K. (2011). The process writing approach: A meta-analysis. The Journal of Educational Research, 104(6), 396-407.

60 of 119

Scoring & feedback is decent and improving

61 of 119

Feedback Quality

62 of 119

Impact of AI on the Process of Learning to Write

Students were intentional users, minimalists, and offloaders, reflecting different levels of agency and self-regulation in their use of AI tools.

We found that students could make deliberate choices that reflected both their own agency and AI literacy.

Regular reflection supported both self regulation and AI literacy.

63 of 119

Impact of AI on the Process of Learning to Write

Students were intentional users, minimalists, and offloaders, reflecting different levels of agency and self-regulation in their use of AI tools.

We found that students could make deliberate choices that reflected both their own agency and AI literacy.

Regular reflection supported both self regulation and AI literacy.

But using AI didn’t just make things easier. It actually made me more careful. I started questioning whether the suggestions made sense, whether they fit what I wanted to say, and whether they were supported by good evidence. These tools helped a lot, but the biggest learning came from being deeply involved in the topic and thinking hard about how to communicate it well. [P41]

64 of 119

Course Implementations

Debate Partner

Elevator Pitch

Team Evaluation

Topic Exploration

Brainstorming

Outlining

Peer Review

Revision Feedback

65 of 119

Understanding the learning goals will drive the evaluation of if / when / how to use generative AI

Reflections from the Classroom

Purpose of writing:

    • Is it to communicate?
    • Is it to support thinking?
    • Is it to show what we know (proof of subject matter knowledge acquisition)?

66 of 119

Have you looked at your SLOs lately?

67 of 119

Student Learning Outcome (SLO)

Related assignments

or activity

AI disruption (1-5 with 5 being high)

Should the SLO change?

Should the assignment change?

Example: Organize Oral Communication

Speech Outlines

5

Probably

Yes

AI Disruption & Revision Tables

Student Learning Outcome (SLO)

Rationale for change

Proposed SLO

Stakeholders

Starting Point

Example: Organize Oral Communication

Outlining speeches is easily done by AI. This will be how presentations are prepared in the workforce or public service

Evaluate and improve oral communication frameworks

Other instructors

CMST

CAB

Program Curriculum Committee and then CAB

68 of 119

69 of 119

Assignment

Effectiveness of AI at performing task (1-5)

What AI programs can do this?

Output Example

(example of what AI has produced when you input your assignment into an generative ai)

Example: Formal outlines for speeches

5

ChatGPT, almost any large language model

Assignment

Rationale for change

Alternative

Associated SLO Change?

Stakeholders

Example: Formal outlines for speeches

Outlining speeches is easily done by AI.

Use prompt engineering to create outlines. Students evaluate and improve outlines.

Yes

Other instructors

Map quizzes

Little point to ask students to remember things easily found via A.I.; like the calculator. Caveat: no tech, then what?

Scavenger Hunt - screenshot/s of places on map of relevant region/s with comments about the place/region

Yes

Students - value for tuition

Essay assignments

Strong likelihood that A.I. will be asked to generate written compositions

Podcast and/or video creation - Record a conversation with someone outside of class discussing course material; show what you have learned

Yes

Students - value for tuition

AI Disruption and Revision Tables for Assignments

70 of 119

Why we are writing?

If we know why we are writing, we can:

  • Make sure the student learning objective is the focus of the activity/assessment/assignment
  • Decide when, if, how AI might be a problem
  • Decide when, if, how AI might be a tool

71 of 119

Understanding the learning goals will drive the evaluation of if / when / how to use generative AI

Reflections from the Classroom

Purpose of writing:

    • Is it to communicate?
    • Is it to support thinking?
    • Is it to show what we know (proof of subject matter knowledge acquisition)?

72 of 119

Writing to show what we know

Susceptible to off loading with AI: Academic integrity challenge

Options:

    • Do it in person without computers?
    • Oral assessment
    • SRL:
      • Focus on Process > Product
      • Low-stakes, e.g., few points or yes/no grading, multiple checkpoints, plus EF scaffolding
    • UDL: Images, videos, podcasts, posters …..
    • Can you turn the tables?
      • Have them correct AI-produced summary, for example?
      • Can the AI prompt show the learning? E.g., instructions for the image or create a persona

Probably the hardest to make AI resistant

73 of 119

So if you need to revise some things ….

74 of 119

  • Authentic product
    • Workforce prep
    • Engagement
  • Authentic process
    • Workforce prep
    • Engagement
  • Agency
  • Modality
    • Engagement
    • AI resistant

75 of 119

  • Explicit instruction
  • Transparency
  • Modeling (I do, we do, you do)
  • Process over product
  • Smaller assignments over the course of the product creation

76 of 119

Principles of Instructional Scaffolding and Tutorial Assistance

  • Reduce instruction into manageable chunks
  • Concentrate the student’s attention on the task
  • Provide models of what is expected
  • Provide extended opportunities to practice in new situations

Jerome Bruner, 1978

77 of 119

Emerging best practices:

Building good habits

78 of 119

Think First

You are more creative, interesting, diverse, and aware of the desired outcome than any AI can be. Think about what you want to write before you start – you only have one chance to think before being influenced by the AI.

“Good Enough” Prompting

Harness prompting techniques like using personas, goals, and details + content knowledge, but if all else fails, use a conversational tone.

Be the Boss

Remember you are the boss of the AI. Ask it questions, make it clarify, expand, or revise things – iterative prompting allows you to define what is needed. Make the output fit your needs and desires.

Reflect

Taking a moment to consider what worked, what didn’t, and what could be done differently allows for improvement and agency.

Corroborate & Interrogate

Corroborate the information AI provides with other reliable sources. Interrogate the output: is there bias you need to address? The human is ultimately responsible for any output they use.

Human-Driven GenAI Use

79 of 119

Think First

“Good Enough” Prompting

Be the Boss

Reflect

Corroborate & Interrogate

Human-Driven GenAI Use

Reflect

Self regulation ends with reflection--did I meet my goals? What did I learn in the process? Should I do anything different? Do I want to adjust my goals next time?

Set Goals

Self regulation begins with an end in mind, what are my goals for using generative AI? What are my intentions?

Self-Monitor

While I’m using generative AI, I’m checking whether the output fits my needs, looking for bias, critically evaluating the content,

Self-Regulated Learning

80 of 119

Think first

Like in reading when we tap into background knowledge.

Strength based: respects what the student already knows.

Keeps the student in charge of the AI.

ALSO: Because of its fluency, language patterns, and good old psychological realities, once you start down the AI-driven path, you rarely diverge.

SRL: Setting goals/intentions needs to be a habit

81 of 119

Think first: What’s that look like in practice?

82 of 119

What if our SLO is related to revision?

AI-Assisted Peer Review

    • Read the writing. Make some notes on things you notice. Consider the assignment and rubric.
    • Prompt the AI to provide feedback.
    • Consider the AI output for accuracy and clarity.
    • Consolidate your thoughts and the additional information from the AI and give your peer consolidated feedback.
    • Consider what you did well, where the AI helped, and what the AI got wrong or missed.

Example

83 of 119

Good enough prompting

You and your students don’t need to be prompt engineers.

Models are improving quickly and dramatically.

Get started, when it doubt treat it like a conversation.

84 of 119

Where are the books on animals?

Where are the books on birds?

Where are the books on eagles?

Good enough prompting is easier if you think first about what you want

85 of 119

A couple of tips when prompting

  • Role or personas
  • Goal
  • Details
  • Context

What do you know that the AI doesn’t but might find useful?

  • Assignment requirements
  • Course content, level
  • What you already know
  • What things you find interesting

86 of 119

Example: Practicing prompting in class

  • [Can be done whole class] Students receive (or create) a weak prompt (e.g., Write an essay about why AI is bad). They predict what kind and quality of output they will get. They enter the prompt into the AI and evaluate the output.
  • In pairs or small groups, students revise the prompts using the information on the prior slide. They enter the prompt and evaluate the output, comparing it to the initial output. They discuss what made the second prompt more effective.
  • Coming back together, the class discusses what they learned and what they will use going forward to improve the quality of the AI output. Teacher points out how they use language precisely, what content knowledge they brought to the task to improve the output, and reminds them how AI predicts based on language it is given.

Adapted from Karchmer-Klein, 2026, p. 84

87 of 119

Example: Asking better questions

  • Choose a debatable topic (e.g., is AI beneficial for education?)
  • Students write an initial prompt to get AI to provide an answer. They input it into the AI and evaluate the output. Is the response biased? Did it consider multiple perspectives? What could make the prompt better?
  • Students revise their prompts to encourage a more balanced, evidence based response (e.g., adding a request for benefits and drawbacks, supporting evidence). They analyze the AI output to understand how it changed, whether the new response was more balanced, more evidence-based. They consider what helped reduce the initial AI bias.
  • Students swap their prompts with a peer and try out the peer’s prompts. They analyze the output to determine whether the new prompt led to a balanced, thoughtful response; consider how the prompt could be further improved.
  • Discuss as a class how the question you ask impacts the response you get. If you ask “why is AI bad?” you get a different output than if you ask “why is AI good?”--students need to think first about what a good, useful output will be and ask for it explicitly.

Adapted from Karchmer-Klein, 2026, p. 83

88 of 119

Be the Boss -- Iterative Prompting

  • Clarification
  • Correction
  • Constraint
  • Expansion
  • Concision

Some resources: DLL, PAIRR, Yost

89 of 119

Be the boss: What’s that look like?

  • Summarize content knowledge or text
  • Have students evaluate the initial input
  • Students then push back by having AI change something using the prior sentence frames

90 of 119

Initial AI output

  • Thoughts?
  • On the surface, AI usually looks good at first
    • Why?
    • What can we do if we aren’t content area experts?

91 of 119

Try to ask a critical question of the output

Even if (especially if?) you’re not a content area expert

  • How would X criticize this document?
  • What would be the counterargument?
  • What is the weakest part of this argument?
  • Who is missing from this story?

92 of 119

Corroborate

Important habit to get into when using generative AI

  • They are built to chat, not gather facts
  • Good critical thinking anyway
  • Always tie back to original sources
  • Triangulate across multiple sources
  • Lateral reading on the internet like professional fact checkers

93 of 119

Corroboration: What’s that look like?

  • Same prompt we just used -- summarize content knowledge or text

Students work together or on their own to make the summary more accurate, precise, clear

  • Students use AI to generate a study guide, then “fact check” it
  • Student use AI to generate content with source citations, then look up the sources to fact check the evidence
  • Option: Students color code AI text with green = accurate and important; yellow = vague or overly general; red = missing or misleading information.
  • Consider: For some assignments, students must include a print out of the source circling where it says what they are citing it for.

94 of 119

Reflection

Building metacognition, self regulation, executive function, habits of mind

Student agency

Build transferable skills

Part of every assignment using AI:

What did you do and how did it work? What would you do differently in the future?

95 of 119

Students + AI requires AI literacy

  • How AI works
  • AI’s limitations
  • Good enough prompting
  • Pushing back
  • Staying in charge of the work

96 of 119

97 of 119

Should generative AI have a role in writing instruction?

98 of 119

Generative AI might play a role throughout the process of writing

Reviewing

  • Evaluating
  • Editing
  • Revising

Planning

  • Setting goals
  • Generating ideas
  • Organizing ideas

Drafting

  • Putting the writing plan into action

Graham, S., & Sandmel, K. (2011). The process writing approach: A meta-analysis. The Journal of Educational Research, 104(6), 396-407.

99 of 119

Scoring & feedback is decent and improving

100 of 119

Feedback Quality

101 of 119

Impact of AI on the Process of Learning to Write

Students were intentional users, minimalists, and offloaders, reflecting different levels of agency and self-regulation in their use of AI tools.

We found that students could make deliberate choices that reflected both their own agency and AI literacy.

Regular reflection supported both self regulation and AI literacy.

102 of 119

Impact of AI on the Process of Learning to Write

Students were intentional users, minimalists, and offloaders, reflecting different levels of agency and self-regulation in their use of AI tools.

We found that students could make deliberate choices that reflected both their own agency and AI literacy.

Regular reflection supported both self regulation and AI literacy.

But using AI didn’t just make things easier. It actually made me more careful. I started questioning whether the suggestions made sense, whether they fit what I wanted to say, and whether they were supported by good evidence. These tools helped a lot, but the biggest learning came from being deeply involved in the topic and thinking hard about how to communicate it well. [P41]

103 of 119

Course Implementations

Debate Partner

Elevator Pitch

Team Evaluation

Topic Exploration

Brainstorming

Outlining

Peer Review

Revision Feedback

104 of 119

AI-Assisted Peer Review

    • Read the writing. Make some notes on things you notice. Consider the assignment and rubric.
    • Prompt the AI to provide feedback.
    • Consider the AI output for accuracy and clarity.
    • Consolidate your thoughts and the additional information from the AI and give your peer consolidated feedback.
    • Consider what you did well, where the AI helped, and what the AI got wrong or missed.

Example #1

105 of 119

“I would say that it [AI] definitely encouraged me to get feedback because I wasn’t afraid of the stigma. That barrier to entry was much lower. It didn’t cost me anything to use the AI, if that makes sense.”��- UC Irvine Student

Feedback

AI as a Feedback Partner

106 of 119

Example: Student conversation with PapyrusAI on 2nd draft of Project 1.

“Going to office hours is helpful in determining what the professor wants to see in a paper as well as improving my structuring and see if I missed anything throughout my research. AI was helpful in revisioning too. It provides solid suggestions that sometimes matches the professor’s and peers’ review, and also captures minute details that sometimes may be overlooked.”�� - UC Irvine Student

Field Notebook Post Assignment Observations Reveal Blended Feedback

107 of 119

Reverse Outline

      • Ask the AI to outline your writing.
      • Look at the outline--what is missing? Are there logic gaps? Could you reorganize the flow?
      • Think about whether the outline helped you revise and strengthen your writing.

Example #2

108 of 119

Debate Opponent

      • Give the AI the debate topic and tell it to act as your opponent.
      • Tell it your position and arguments.
      • Have it counter you.
      • Then come up with counter arguments.
      • Ask it for suggestions of other things you might have argued or countered.

Example #3

109 of 119

“I would say it definitely made me think. It made me think about my argument a lot. It was a little bit daunting, a little scary because it provided really good counter-arguments. … I had to think a lot in order to provide a good argument.”��- UC Irvine Student

Brainstorming

AI Serves as a Debate Partner

110 of 119

How to Decide:

      • Align with learning outcomes.
      • Consider the student’s developmental stage.
      • Do risks of over-reliance outweigh benefits?
      • Do you have the time to do it well?

Skip It When:

      • Simply off-loading task without higher order effort required (task completion !=learning).
      • Human connection or imperfection is a value.
      • Students are not in a position to corroborate & that matters.

Use It When:

      • Productive struggle (ZPD) requires scaffolding.
      • Use requires critical thinking (prompting, reviewing, integrating).
      • Authentic.

111 of 119

How can instructors use AI ethically?

Think of a learning objective, activity, or assignment that your students struggle with. Use AI (we have prompts in the resources) to:

  • Get ideas on how to improve it using one of the two Student Support Ideas prompts
  • Use the assignment or active learning activity generator to create an alternative assignment/activity with the same learning objective
  • Use the lesson plan generator to create a lesson plan around the struggle
  • Use the “explanations” prompt to generate new ways to explain the content
  • Use our multilingual prompt suggestions

112 of 119

Multilingual prompts

Prompt Type

Description

Prompt Examples

Brainstorming Partner

Provides brainstorming assistance in the user's native language (L1), helping generate ideas or concepts, leading to the generation of a bullet-listed summary or table

Vocabulary Tutor

Supports vocabulary learning by offering definitions, examples, collocations, common errors, with opportunities to reinforce learning through quizzes or sentence writing.

Translation Tutor

Assists in translating text between languages, followed by vocabulary quizzes, writing exercises, or translation practices.

Multilingual Feedback

Supplements other PapyrusAI feedback prompts (focusing on content, organization, style, or mechanics), this prompt additionally identifies language and rhetorical errors potentially influenced by L1 and provides revision suggestions

Multilingual Feedback

RAG source materials:

  1. Swan, M., & Smith, B. (2001). Learner English: A teacher's guide to interference and other problems (Vol. 1). Cambridge University Press.
  2. Turton, N. D. (1996). Longman dictionary of common errors. Longman.

Revision Tutor

Using a dialogic feedback model, this prompt provides revision assistance through tutor-tutee interaction, based on the learner's identified needs or focusing on one prioritized area at a time, while scaffolding the revision process at both the sentence and text levels.

113 of 119

Good idea ✅

Nope, missed the point--same habits apply!🚫

  1. Drafted first
  2. Prompt inc. persona, context, expertise
  3. I did push back
  4. Corroborated & interrogated
  5. Reflected

114 of 119

If you only do 1 thing this term…

Pick ONE move. Keep it small. Make it modular.

1

Put one of your assignments into AI

  • See what is possible.
  • Make sure to push back a couple of times to get better, more appropriate AI output.
  • Should you plan some time to look at your SLOs and assignments?
  • Can you discuss the results with your class in a way that makes it a learning opportunity?

2

Use AI to modify ONE assignment

  • Use some of our suggested prompts to make it more active.
  • Modify the assignment to allow different modalities (UDL).
  • Have AI act as one of your students and give you feedback on the assignment. Make sure to tell the AI about your students.

3

Insert AI into one assignment

  • Adding AI feedback to a peer feedback cycle is low stakes.
  • Have AI summarize one resource and students corroborate it.
  • Can AI literacy be the topic of an activity?
  • See our activities handouts for other suggestions.

115 of 119

Resources you can use/remix/adapt for free!

  • Prompts--for you and your students
  • Activities
  • Foundational curriculum for students (and teachers)
  • Videos on AI literacy for students & AI-infused pedagogy for instructors
  • Research

116 of 119

Activity / Lesson

Learning Goal

Students will explore the capacities and limitations of generative AI.

Students will be able to use generative AI to efficiently generate feedback and understand their own role in ensuring that feedback is accurate and useful.

Students will be able to use generative AI to efficiently generate suggestions for peer feedback and understand their own role in ensuring that feedback is accurate and useful.

Students will be able to selectively use generative AI to produce concise summaries of a text, while also recognizing its supportive role and their responsibility to ensure the summary is accurate and meets their objectives.

Students will be able to use generative AI to receive feedback from an imagined audience and explore ways to improve their writing to meet the expectations and needs of different audiences.

Students will be able to use generative AI to create a reverse outline of their draft, analyze the structure of their writing, and identify areas where the organization can be improved. They will also learn to adjust the flow of their work to ensure clarity and cohesion.

Students will be able to use generative AI to effectively brainstorm writing topics, explore background knowledge, and refine their ideas to align with assignment goals. They will also learn how to evaluate each potential topic.

117 of 119

Let’s use AI only in ways that support learning and humanness.

118 of 119

PapyrusAI Team

119 of 119

Thank you!

Tamara Tate, PhD

Associate Director, Digital Learning Lab

University of California, Irvine

tatet@uci.edu

This material is based upon work supported by the National Science Foundation under Grant No. 2315294 and the California Educational Learning Lab AI Grand Challenge

© 2025 The Regents of the University of California

Want to spend the summer together?

UCI Continuing Education:

AI in Higher Education

Join me!