2 Teaching Writing in the Context of Generative Artificial Intelligence

This section of the playbook is designed to help instructors navigate the changing GenAI landscape as they design and implement public health coursework. Acknowledging that AI literacy varies, the section will start with a primer on AI and GenAI. The playbook will then cover teaching writing in public health in the context of AI, providing guidance on designing AI-assisted or AI-resilient assignments.

This playbook will not cover the complex intricacies of AI, environmental ethics of AI, AI literacy, or detail applications of AI to non-writing assignments in public health (i.e., writing code, data analysis). This guide will also not cover the use of AI to design a course, instead focusing on how AI can be used or prohibited to enhance teaching and learning writing.

2.1 What is Artificial Intelligence?

Artificial intelligence (AI) is a set of technologies that allow computer systems to identify or classify, predict, recommend and generate material, which simulates human cognition.^27–29 Artificial intelligence technology or the accompanying research field are not new. For example, text prediction on iPhones and Gmail was introduced in 2014 and 2018 respectively.^30,31 Advances in the field have continued to integrate AI into our daily lives (e.g., online customer service chat bot; wearable fitness trackers). Despite its ubiquitous nature, especially after the introduction of OpenAI’s ChatGPT in 2022, a Pew Research study found that although over half of Americans were aware of AI, only 30% were able to correctly identify six common daily AI encounters such as email spam filters.³² These proportions have likely increased and will continue to do so as the AI landscape continues to evolve and become more pervasive, such as Google’s “AI Overview,” an AI-generated summary that appears at the top of every Google search.

This playbook will focus on generative artificial intelligence (GenAI or GAI), a subset of AI which specializes in generating content (e.g., text, images, video).³³ Specifically, this playbook will focus on GenAI models that rely on large language models (LLM), which are trained on large bodies of text and specialize in generating written content.²⁹ Drawing on the large body of training text, LLMs use known patterns and next-word probabilities to generate text responses to user prompts. LLM responses appear to mimic human cognition by appearing conversational and authoritative in tone. In essence, LLMs, such as ChatGPT, Claude, Gemini and Microsoft Copilot, are “extremely powerful forms of autocomplete”.²⁹

2.2 GenAI and higher education

GenAI is influencing how students, staff, and instructors approach teaching and learning in higher education. It is important to note that GenAI’s relative novelty and ongoing evolution mean that researchers and educators are still working to understand its complexities, including the potential benefits and challenges it presents for higher education and student learning. The majority of the existing popular literature and scholarship examining the impact of GenAI use in higher education have focused on writing and composition assignments and coursework.^27,34–39 This increased emphasis is likely attributable to the iterative nature and creative aspect of writing,⁵ both of which are essential to learning and may be threatened by student GenAI use. Student use of GenAI can vary, from helping write a catchy title or rephrasing a sentence to remove biased language, to writing large portions of an assignment.

2.2.1 Public health education

Public health education and practice is not exempt from the benefits and challenges of GenAI use. GenAI may help personalize education, rapidly process large volumes of data, as well as culturally and linguistically tailor health communication interventions.^40,41 Simultaneously, GenAI may present ethical challenges and impede student learning by potentially eliminating critical thinking steps (section 1.3.2) and providing students with written output.^40,42–44

2.3 GenAI in the classroom

Students are already using GenAI. A 2023 survey of public health students found that over half (63%) used GenAI and only 9% of students reported having never used any form of AI (e.g., spell check, citation software). Student AI use ranged from clarifying difficult concepts and brainstorming to summarizing readings and drafting original text for a paper.⁴⁴ GenAI has also been found to exceed instructors’ expectations when responding to existing writing assignments⁴⁵ and has prompted instructors and higher education institutions to reconsider writing assignment design to maximize student learning within the context of GenAI.^45,46

It is increasingly clear that an “AI-proof” assignment may not exist. Additionally, methods to identify plagiarism and AI use are inadequate (section 2.3.5), often reporting false positives⁴⁵ and exhibiting significant bias towards non-native English speaking students.^35,39,47 In response, recent efforts to design writing assignments have focused on limiting AI use (AI-resilient) or intentionally incorporating AI as a student-collaborator (AI-assisted). These approaches require a basic familiarity with AI and GenAI, rely on clear expectations and policies, and maintain a focus on student learning (e.g., critical thinking skills).

2.3.1 Getting started with GenAI: Building AI literacy

The best way to understand GenAI, build AI literacy, and determine the potential inclusion or exclusion of GenAI from a course is to explore the tools firsthand. Using GenAI tools increases familiarity with its abilities as well as helps to identify its limitations relative to a given course or assignment. Most GenAI tools and platforms have a free version as well as a paid version, making it easy to try different tools for various tasks. The University of Washington has also provided a paid version of Microsoft Copilot for all UW ² staff, students and UW Medicine employees. The paid version has commercial data protection enabled when signed in with your UW credentials. Harvard’s AI Pedagogy Project also has a “Compare LLMs” tool where you can compare Claude and ChatGPT side by side (https://aipedagogy.org/guide/comparison/) without a free or paid account.

Consider the following questions^35,48 while exploring GenAI tools:

• How easy is the GenAI tool interface to use?
• What can GenAI do?
• What steps or components of a typical assignment can the GenAI tool perform?
• What are the GenAI tool limitations?
• What steps or components of a typical assignment seem beyond GenAI tool capabilities?

A summary of several common GenAI tools is listed in Table 5.

Table 5. Comparison summary of AI tools.
	Copilot	ChatGPT	Claude AI	Gemini	Notebook	Perplexity	Research Rabbit	Zoom AI companion
UW approveda	X
Parent Company	Microsoft	OpenAI	Anthropic	Google	Google	Perplexity	Litmaps		Zoom
Free (limited) version available	X	X	X	X		X	X		X
Paid version for UW	X								X
GenAI	X	X	X	X		X
Research AI tool					X		X
Chat bot (conversational)	X	X	X	X		X
Saved chats
File uploadb	X	X	X	X	X	X	X		X
Image generation	X	X	X	X		X
a Commercial data protection is enabled for UW instructors, staff, students and Medicine employees. Must log-in with UW Net ID. Tools with this selection must be used when protecting student privacy and complying with FERPA requirements. For more information please see: https://uwconnect.uw.edu/it?id=kb_article_view&sysparm_article=KB0034403 b Platforms vary by number and size of files that can be uploaded.

2.3.2 Empowering student AI literacy

AI literacy is valuable for students to understand the basics of AI, determine appropriate and ethical use, and critically evaluate AI output.⁴⁹ Although many students may already be using AI, the AI literacy of students in your course may vary. To start building AI literacy, especially within a classroom setting, set expectations (i.e., GenAI use policy and syllabus statements) and communicate with students early and regularly.²⁸ Many of the suggestions below for communicating with students about AI are sourced from the University of Washington “AI + Teaching” resource.

2.3.2.a Discuss AI and GenAI

Despite increased use of AI, many students may not know how AI analyzes patterns and contextual information using large datasets and algorithms.⁵⁰ Ensuring that students have a foundational understanding of what AI is and is capable of will help students use AI more effectively in addition to better understanding the benefits and risks associated with AI-use.

2.3.2.b Discuss the importance of critical thinking and learning

Students are often stressed and taking full course loads, which may lead to students using GenAI to help save time, reduce overall stress, and subscribe to the belief that GenAI may improve grade outcomes.⁴⁴ Reinforce with students that although GenAI may save time, it may also remove the inherent struggle that is critical to learning and especially learning to write.²⁸

2.3.2.c Discuss ethical and social issues related to AI

Students may be using AI and GenAI but may not be as familiar with the ethical and social issues related to AI and AI use. Conversations around data privacy, bias, accuracy, equity, environmental impact, intellectual property, and academic integrity can help students make informed decisions about when and how they interact with and use AI in the classroom. For more information about each of these issues, please see the University of Washington page on Ethical and Social Issues Related to AI: https://teaching.washington.edu/course-design/ai/ai-ethical-issues/.

2.3.2.d Provide students access to additional resources

Additional campus resources can help students navigate the ever-evolving AI landscape and provide additional insight into appropriate AI use relative to each student’s individual learning. AI-related resources tailored to student audiences can be found in the Resources section R.2.e.

Figure 4. Framework for AI literacy.⁵¹
A more comprehensive approach to developing AI literacy can be found in the first three of four steps of Barnard College’s Framework for AI literacy (Figure 4).⁵¹ The fourth and final step of the framework is to create AI, which is one step further than necessary for the context of teaching public health writing.

2.3.3 Developing a GenAI use policy

AI and GenAI are omnipresent and most of us use AI daily. Whether you are curious about using AI in your course or are against AI use, having clear guidelines can help students navigate whether they can use GenAI as a resource and if so, understand the parameters around that use.

The University of Washington (UW) has an overarching GenAI use policy, which can be found here: https://it.uw.edu/guides/security-authentication/artificial-intelligence-guidelines/. The UW GenAI use policy allows for the use of GenAI when it is “helpful to further [the University’s] mission, the preservation, advancement and dissemination of knowledge.”⁵² In addition to sharing a university-wide policy, the UW encourages instructors and staff to create course specific policies pertaining to student GenAI use, which should be included in syllabi and assignment prompts.²⁸ Remember that not every writing assignment is the same; although you may have an overarching course GenAI use policy, you may need to consider what each assignment entails (e.g., task, learning objective) and create assignment specific GenAI policies.^45,53

The University of Washington has also provided guidance for syllabi statements regarding student use of AI, which is summarized below and can be found here: https://teaching.washington.edu/course-design/ai/sample-ai-syllabus-statements/.

Key considerations when developing your AI course policy are detailed below:^28,54,55

• Type of policy: start by clearly stating the course policy on GenAI use. There are four common types: 1) GenAI use prohibited; 2) GenAI use not allowed unless certain conditions are met; 3) GenAI use allowed but limited to specific circumstances, tools, or assignments; 4) GenAI use allowed/unlimited. If an instructor is uncertain about their stance on GenAI in a given class, see the next section, “Framework for developing technology guidelines for public health writing assignments”.
• Gen AI tools: when applicable, it is helpful to list specific GenAI tools that students are or are not allowed to use. This reduces any confusion around what tools may or may not fall under the course policy.
• Reasoning: share the rationale behind the GenAI use policy. This helps students understand the purpose and importance of the course policy and how it impacts their learning in the course.
• Conditions and contexts (as applicable): providing specific conditions or contexts allows for parameters around limited use. Examples include: can only use GenAI for non-graded assignments; only with your data; only for reflection and studying but cannot generate content for assignments.
• Color-coded policies: using “green, yellow, orange, red light” color coding can help students visualize when your GenAI use policy. Green is representative of unlimited GenAI use, yellow and orange are limited GenAI use, and red is representative of prohibited GenAI use.
• Consequences: clearly stating the consequences for GenAI use violating the course policy reduces misunderstanding, allows for consistent application of the policy, and empowers students to consider potential outcomes of their actions and make choices accordingly.
• Questions: the landscape of GenAI is rapidly evolving and students are likely to have questions. Providing language that encourages them as well as details how they can reach the instructor or other university personnel about questions can help provide additional clarity and support where needed.
• Privacy reminder: it is helpful to remind students that Microsoft Copilot is the only UW-supported GenAI tool as of 2025. The standing agreement helps protect user data and privacy.

Stanford University has also provided a guide to developing AI course policies, including a worksheet for creating an AI course policy,⁵⁴ which is located in the resource section of this playbook.

Table 6. Example syllabi AI use policies
Type of AI use policy	Source/ Institution	Example syllabi statement	Key considerations included
AI use prohibited	University of Washington55	All work submitted for this course must be your own. Any use of generative AI tools when working on assignments is forbidden. Use of generative AI will be considered academic misconduct and subject to investigation The assignments in this class have been designed to challenge you to develop creativity, critical-thinking, and problem-solving skills. Using AI technology will limit your capacity to develop these skills and to meet the learning goals of this course. If you have any questions about what constitutes academic integrity in this course or at the University of Washington, please feel free to contact me to discuss your concerns.	Policy type Reasoning Questions
	University of Pittsburgh Writing Institute Workshop: AI and the Teaching of Writing56	The use of generative AI writing tools (such as ChatGPT, GrammarlyGO, GPT-3, GPT-4, BERT, or others) is prohibited in this class. Assignments for the course have been designed to help you develop as a writer without the use of these technologies. You will generate ideas, read, revise, and write on your own and/or in consultation with peers, me, or Writing Center tutors and you will not use AI at any stage of your writing process. You are the author of your work for the course and authorship means you take responsibility for your words and claims. Any use of AI technologies in your work will be considered an academic integrity violation and addressed accordingly: [INSERT UW WEBSITE]. Please see me if you have any questions about this policy.	Policy type GenAI tool Reasoning Consequences Questions
AI use limited	University of Washington55	In this course, students are permitted to use AI-based tools (such as UW’s version of Copilot) on some assignments. The instructions for each assignment will include information about whether and how you may use AI-based tools to complete the assignment. All sources, including AI tools, must be properly cited. Use of AI in ways that are inconsistent with the parameters above will be considered academic misconduct and subject to investigation. Please note that AI results can be biased and inaccurate. It is your responsibility to ensure that the information you use from AI is accurate. Additionally, pay attention to the privacy of your data. Many AI tools will incorporate and use any content you share, so be careful not to unintentionally share copyrighted materials, original work, or personal information. Learning how to thoughtfully and strategically use AI-based tools may help you develop your skills, refine your work, and prepare you for your future career. If you have any questions about citation or about what constitutes academic integrity in this course or at the University of Washington, please feel free to contact me to discuss your concerns.	Policy type GenAI tool Reasoning Conditions & contexts Consequences Questions Privacy reminder
	University of Pittsburgh Writing Institute Workshop: AI and the Teaching of Writing56	The use of generative AI writing tools (such as ChatGPT, GrammarlyGO, GPT-3, GPT-4, BERT, or others) is allowed in this class within specific contexts and only if such use is properly acknowledged. Assignments for the course have been designed to help you develop as a writer, and some of them may call on you to practice writing with the help of such tools. As your instructor, I will assume that any use of these tools will be only within the contexts the assignment allows (for instance, you can use ChatGPT for brainstorming if the assignment asks you to do so). You must acknowledge the use of AI in your assignment in an “Acknowledgement of AI Use” statement that: Specifies which technology was used and on what date (ChatGPT, GPT-3, etc.) Includes explicit descriptions of how the information was generated Identifies the prompts used Explains how the output was used in your work Examples of such Acknowledgments of AI Use can be found on Monash University’s website. The use of AI outside of contexts where the instructor specifies its use, or failure to acknowledge any use of AI technologies in your work will be considered an academic integrity violation and addressed according to [INSERT SCHOOL] Academic Integrity policies: [INSERT UW POLICY WEBSITE]. You are the author of your work for the course and authorship means you take responsibility for your words and claims, regardless of which tools you use. Please see me if you have any questions about this policy.	Policy type GenAI tool Reasoning Conditions & contexts Consequences Questions
AI use unlimited	University of Washington55	Students in this course are encouraged to explore the use of AI-based tools (such as UW’s version of Copilot) when completing assignments. All sources, including AI tools, must be properly cited. Use of AI without proper citation will be considered academic misconduct and subject to investigation. Please note that AI results can be biased and inaccurate. It is your responsibility to ensure that the information you use from AI is accurate. Additionally, pay attention to the privacy of your data. Many AI tools will incorporate and use any content you share, so be careful not to unintentionally share copyrighted materials, original work, or personal information. Learning how to thoughtfully and strategically use AI-based tools may help you develop your skills, refine your work, and prepare you for your future career. If you have any questions about citation or about what constitutes academic integrity in this course or at the University of Washington, please feel free to contact me to discuss your concerns.	Policy type GenAI tool Reasoning Consequences Questions Privacy reminder

2.3.4 Framework for developing technology guidelines for public health writing assignments

August and Anderson (2024) designed a 3-step framework for developing technology guidelines for writing assignments in public health.⁵⁷ In the framework, August and Anderson guide instructors through identifying the learning objectives, assessing how GenAI may assist or detract from students’ achieving the learning objectives, and finally, creating assignment-specific guidelines. The framework is applicable to all public health assignments, although the example that they provide is for the introduction of a standard scientific paper in an epidemiology course. The 3-step framework is detailed below in Table 7 and an applied example from the August and Anderson article can be seen in Figure 3.

Table 7. August & Anderson framework for developing technology guidelines for public health writing assignments.
Framework recommendation	Steps	Rationale
Articulate the assignment learning objectives.	Write out all learning objectives using behavioral verbsa	Strong and well written learning objectives should clearly provide the specific knowledge and/or skills that students will obtain by completing the assignment. By clearly articulating the learning objectives, you will be able to assess whether the use of GenAI will enhance or inhibit a student’s ability to achieve the objectives.
Identify how students could use technology for the assignment, and impacts on the learning objectives.	Write a list of tasks a student needs to go through to finish the assignment Assess what GenAI tools students may use to help them complete the assignment; in doing so, consider accuracy, bias, or equity-related issues that may stem from GenAI use in the assignment Consider whether students benefit or are hindered in achieving their learning objectives when using GenAI Decide whether there could be ripple effects from GenAI use – are subsequent parts of an assignment impacted by early GenAI use	By considering the benefits and challenges a student might encounter by using GenAI to complete portions of a given assignment, an instructor is better positioned to develop corresponding GenAI use guidelines.
Develop technology guidelines.	Review step 2 and decide whether use of a given technology is acceptable, should be limited, or should be prohibited based on student learning outcomes. If GenAI use is allowed, consider equity impacts. Create guidelines for the assignment, and if GenAI use is limited or prohibited, provide how learning is compromised by the use of GenAI.	Clear technology guidelines, which account for the entirety of the assignment, will help set parameters for student GenAI use.
This table was created from the framework outlined and explanations provided in August & Anderson (2024).57 a Behavioral verbs can be found in Bloom’s taxonomy and are used to help classify, and measure student progress on objectives: https://sph.washington.edu/faculty/academic-resources/course-learning-objectives.

2.3.5 Detecting student AI-use

Students may continue to use GenAI despite instructors’ best efforts to mitigate use through policies and assignment design. AI-detection tools have been created in response to help identify and thereby mitigate student GenAI use. However, these tools, including some developed by GenAI companies, have been found to be ineffective and report false positives, mis-identifying student writing as AI generated.^39,45,58 The number of false positives increases for multi-lingual students’ writing.⁴⁷ Additional caution should be exercised when using AI-detection tools, as GenAI is constantly evolving, making GenAI output increasingly difficult to distinguish from student writing. This challenge may be compounded by growing student GenAI use, through which students may inadvertently learn to mimic GenAI content and writing styles. Instructors may be better served if deciding to develop AI-resilient and AI-assisted assignments rather than trying to “catch” student AI-use.

2.4 Developing AI-resilient writing assignments

GenAI will continue to evolve and in doing so will present myriad known and unforeseen benefits and challenges. A common concern in higher education is increasing student use of GenAI to streamline and complete coursework, which may hinder the development of critical thinking skills and long-term learning. Prohibiting AI in classrooms is unrealistic; however, instructors can mitigate student GenAI use and promote student learning by designing and using AI-resilient assignments. For the purposes of this playbook, AI-resilient writing, which can also be referred to as AI-resistant writing, is defined as writing coursework that is less susceptible to AI assistance.

2.4.1 Partnering with students in AI-resilient learning

As end-users, students are an integral part of every assignment. Despite instructors’ best efforts, students may still willfully use GenAI to help them complete their academic work. Instructors can partner with students to mitigate the use of GenAI by sharing their reasoning for limiting its use. This can occur through a variety of modalities (e.g., written explanation within a student assignment prompt or in a class discussion) and should reflect the AI policy and reasoning found in the course syllabus (section 2.3.3). Explicitly and transparently sharing the reasoning behind limited or prohibited GenAI use with students creates a climate of trust within the classroom and allows students to approach their assignments with an understanding of how GenAI may hinder their learning.⁵⁹ From a student perspective, this transforms assignments from tasks to be completed, to an experience that benefits them and their learning.⁶⁰

2.4.2 Assignments as extensions of the classroom

GenAI is limited to the information on the internet, and as such cannot adequately address prompts that engage with or respond to classroom activities or discourse. This creates a unique opportunity for instructors when developing AI-resilient assignments. Writing assignments with prompts that ask students to reflect upon, respond to or build on ideas from class discussions requires students to be present and engaged with class material, as well as have a fundamental understanding of the class conversation, which GenAI is not immediately privy to.⁶⁰ This encourages students to take a more active role in class (inclusive of listening and not just speaking) as well as in their writing, in which they have to wrestle with the concepts covered as well as how they can expand the discussion. Furthermore, it asks students to move beyond remixing and referencing ideas, which GenAI is capable of, and build upon ideas to create something new.

2.4.3 Focus on the writing process

Designing assignments to focus on the writing process may help reduce students’ use of GenAI by demystifying how to approach writing and actively apply their learning through review and revision. There are three approaches that can help emphasize the writing process and simultaneously help reduce student GenAI use: 1) scaffolding, 2) incorporating multiple skills, and 3) review and revision. While these approaches are described individually below, they can also be used together to enhance student learning.

2.4.3.a. Scaffolding assignments

As described in section 1.2.4.a,²² scaffolding creates smaller manageable tasks that incorporate opportunities for feedback and revision early in the writing process. This allows students to engage with and understand the writing process rather than focus solely on the final product.^1,61 This shift in focus allows students to reduce their use of GenAI as assignments may be perceived as more manageable and students may feel more confident in their ability to write, especially as they develop sustainable writing skills (e.g., how to outline, develop an argument, cite a source), which in turn can be used in future coursework as well as in practice. Reflection and metacognition (section 2.4.4) can be built into scaffolded assignments, helping to further solidify student learning. Lastly, for instructors, scaffolded assignments may provide helpful insights into student writing, and especially student voice as they provide incremental drafts throughout the course.⁶² This allows instructors to observe and facilitate gradual student learning, as well as potentially identify areas of concern or a sudden shift in writing that may indicate GenAI use. Examples of scaffolded assignments can be seen in sections A.1.1. and A.2.4.

2.4.3.b. Incorporate multiple skills

Assignments that require the use of multiple skills, such as a project that requires data collection, data analysis, and writing, may discourage student use of GenAI^63,64 for two reasons. First, although GenAI has the capability to produce large volumes of text, it cannot perform tasks that require human engagement (i.e., data collection). This limits GenAI use for specific stages of a larger assignment, which may encourage a more balanced approach to GenAI use.⁶⁵ Similarly, by incorporating tasks that require critical thinking (i.e., data analysis), instructors are encouraging students to synthesize and connect ideas, which necessitates deeper engagement with course material than GenAI summaries can produce without additional context and prompting. Second, by requiring skills that leverage active participation, instructors may increase student engagement with and ownership of the overall assignment, potentially reducing the temptation to use GenAI throughout the assignment.

2.4.3.c. Leverage review and revision

Revision is an essential stage of writing (section 1.2.2.) and is often neglected.⁴ In the context of AI-resilient assignments, it is important to re-emphasize the distinction between revision and editing. Revising is the process in which a student examines and improves the ideas and concepts of a written body of work.⁶⁰ This is an expansion beyond the traditional examination of grammar, phrasing, and formatting, which are hallmarks of editing. It is important that both instructors and students understand this distinction. Incorporating revision into scaffolded assignments (section 2.4.3.a) may dissuade student GenAI use for several reasons. First, revision requires a level of critical thinking that GenAI tools may not yet have and encourages the incorporation of human-driven insights.⁶⁵ Second, written assignments generated by AI are often free of grammatical errors but may lack overall substance or may contain hallucinations. This can lead to a substantiative revision process, which increases the effort needed for revision and may discourage student GenAI use in the future.⁶⁰ Additionally, writing assignments in the review and revision stage with a lack of grammatical errors and substance may also serve as a potential indication of student AI-use for instructors. Examples of incorporating review and revision can be seen in Example Assignments section A.1.1. and A.2.4.

2.4.4. Emphasize reflection and metacognition

Reflection and metacognition are important strategies for acquiring knowledge and as such are integral parts of Writing Across the Curriculum.^60,61,66 Both processes ask students to consider and critically examine their experiences including their approach, strengths and areas of growth, assumptions, and knowledge.⁶⁰ Reflection is a retrospective experience, focused primarily on experiences within the context of past assignments.^60,61 Metacognition is a prospective process in which students are asked to think about how past experiences can inform their future actions and approaches. Reflection and metacognition can be used individually as well as together, and throughout the writing assignment. Emphasizing these strategies helps to reduce student GenAI use by focusing on the writing process. Students who use GenAI to complete a writing assignment will have a harder time reflecting on their process or how it can be applied in future scenarios. Furthermore, although AI can generate a vast amount of content, it cannot yet reflect on personal experiences, making it difficult for students to use GenAI to complete reflective assignments.

2.4.5. Active, experiential, and collaborative writing

GenAI use requires access to a computer and the internet. Writing assignments or other writing-related assessments (e.g., presentations, peer-review activities) that are completed in-class may help deter the use of GenAI tools through limited computer use. That would also be the case for active (e.g., brainstorming, concept mapping, root cause analysis), experiential (e.g., outside the classroom, internships, field studies), and collaborative (e.g., project-based) learning activities.⁶⁷ Adding a local focus (e.g., local public health issue) can also reduce the use of GenAI tools as GenAI tools may be more limited in terms of hyper-local knowledge as well as assignments might hold more relevance for students and be directly applicable to their communities.^35,66

2.5 Developing AI-assisted writing assignments

Although limiting student GenAI use in classrooms and writing-based coursework can promote critical thinking and learning, there is also value in teaching students how and when to use GenAI effectively without compromising their learning. This is especially prudent as students will graduate into a workforce where GenAI use will be increasingly common and potentially expected.^3,68 Designing and using AI-assisted writing assignments provides student exposure to GenAI tools, building AI literacy, as well as helps students understand the accuracy, capabilities and limitations of GenAI, and begin to examine ethical and political questions that GenAI use prompts.

This section of the playbook will share several considerations for designing AI-assisted assignments. When reading through the suggestions below and when considering incorporating GenAI use into student learning, it is important to remain mindful of accessibility and ethical concerns. Lastly, for all suggestions below, it is important to remember and remind students that AI can hallucinate and requires human oversight.

2.5.1. Idea generation and development

GenAI can be used as a powerful brainstorming tool for students throughout the writing process, without minimizing the process itself or subsequent learning. By drawing on large training datasets, GenAI can help students by brainstorming topics, directions, or ideas for writing assignments that align with student interests and assignment prompts. Similarly, GenAI can be used to help students overcome writer’s block by providing additional ideas, suggestions for starting sentences, or by asking guided questions.⁶⁹ In doing so, GenAI serves as a brainstorming partner while still allowing students to continue writing on their own. These considerations can be incorporated into class settings by explicitly allowing brainstorming in limited AI use policies or by asking students to use GenAI to brainstorm as an initial stage within a scaffolded assignment. See Example Assignments section A.2.1. for an example of how to use GenAI to help students brainstorm.

2.5.1.a Citation hallucinations

Students may use GenAI to help identify literature that supports a position or argument within their written assignment. However, it is important to remember and remind students that GenAI can provide information that is inaccurate, biased or outdated. This includes reference literature or quotes from suggested articles. Students should always review GenAI output to make sure the information they are using is factual.

2.5.2. Providing writing structure

GenAI can help students outline or organize ideas for writing assignments.^63,69 This can support students by providing suggested structure for a writing assignment and ensuring logical flow, allowing students to focus on the process of writing and connecting individual sections.

2.5.3. Review and editing

Review and revision are an essential part of the writing process (section 1.2.2.);⁴ proofreading and editing written work can be enhanced by GenAI.^63,69 Allowing GenAI to assist in correcting grammatical errors and improving writing style can help students focus on the writing process, minimizing the amount of time spent on making a written piece grammatically or syntactically perfect. This may be especially beneficial to students first learning to write in the discipline or students who may be non-Native English speakers,⁶⁹ as it will allow them to practice and understand components of the writing process, and discipline specific requirements or practices, without experiencing additional pressure to craft perfect written assignments.

2.5.4. Tone and formatting

Public health students and the public health workforce are often asked to communicate findings and recommendations to diverse audiences. To be effective, public health written communication must be linguistically and culturally tailored to the needs, values, and expectations of each audience. This often means adjusting the tone, language complexity, format and framing of each written piece to ensure audience relevance, engagement and maximize impact. GenAI should not be used to assess culturally appropriate phrasing or relevance; however, GenAI can assist students in tailoring the tone (e.g., professional, academic, conversational) of written communication to a given audience.⁶⁹ For example, GenAI can be used to increase professional tone and use of strong action verbs in a resume or could be used to adapt a paragraph from passive to active voice. This use of GenAI still requires that students compose initial drafts but supports writing skill development by providing immediate feedback and modeling different tones, which may help students understand how different word or phrasing choices affect tone and clarity. An example of an assignment of using GenAI to help adjust tone and formatting can be seen in the Example Assignments section A.2.2.

2.5.5. GenAI as a peer editor

GenAI can help students by serving as an editor, helping to examine grammar, phrasing, and formatting in student written work. This is distinct from revising, which is discussed in section 1.2.2 and 2.4.3.c. Many applications may already have built in editors (e.g., Microsoft Word grammar and spell check) or plug-ins (e.g., Grammarly) which help students improve grammar and clarity throughout written documents. GenAI may be able to assist students further by providing explanations for suggested revisions; however, the onus to read, understand, and evaluate GenAI suggestions falls upon the student in order to allow for sustained learning. To help this process, it may be best to recommend GenAI as a peer editor is used at a sentence level, rather than on an entire document at one time.⁷⁰ Then students can easily compare their original draft and the GenAI edited version to identify differences and work to understand the suggested edits. An example of an assignment of using GenAI as a peer editor can be seen in the Example Assignments section A.2.3.

2.5.6. Emphasize reflection and metacognition after AI-use

As discussed in section 2.4.4., reflection and metacognition are critical steps for acquiring knowledge and cementing learning.^60,61,66 Asking students to engage with reflective and metacognitive strategies after using GenAI in AI-assisted assignments can help illustrate for students how they are using GenAI tools, and ultimately how GenAI contributed to their completed work.⁵⁹ It also prompts students to reflect on the quality of writing produced by GenAI, what they retained from GenAI use and whether or not they understand GenAI produced writing to use it in future assignments. Asking students to reflect on their GenAI use can also be informative for instructors as it allows them to see how students are using GenAI and ensure adequate learning. Examples of assignments prompting student reflection after GenAI use can be seen in A.2.1.-A.2.4.

2.5.7 Acknowledging and citing GenAI use in academic work

Students should cite GenAI tools when using them in their academic work. This practice allows readers to differentiate content created by the student versus information that was referenced or pulled from other sources. There is some debate about whether GenAI tools can be considered an author and the resulting implications on citation practices; however, for the purposes of student learning, it is important for students to transparently distinguish between their work and others, as well as learn how to properly attribute that work. Additionally, asking students to cite GenAI tools in their work will provide insight to where and how GenAI tools were used and may prompt additional student reflection. Resources for appropriate GenAI citations can be seen in Resources (section R.2.g.).

2.5.8. Evaluation and grading: Assess the human-dependent components of writing assignments

Assessing student assignments where GenAI use is allowed or encouraged can be challenging as instructors may have to distinguish between AI-generated and student-generated work. GenAI can review large bodies of information, and summarize and compare findings using text prediction algorithms. Consequently, it is limited in its ability to transfer information to new contexts, generate text with a unique voice, or apply reasoned judgement. Acknowledging these capabilities and limitations can help when considering how to assess student assignments. The August & Anderson framework presented in section 2.3.4. can help by clearly outlining each assignment’s learning objectives and corresponding AI capabilities and limitations relative to those learning objectives. By separating out which components are capable of being produced by GenAI, instructors can focus their evaluation or assessment on assignment components where GenAI is limited.^57,71 Lastly, for assignments or assignment components with limited or open GenAI use, instructors can ask students to reflect on their experience using GenAI. Student reflections are human-dependent components of the writing process and therefore can be assessed. Furthermore, student reflections may help illustrate for students the benefits and pitfalls of GenAI use as well as prompt critical examination of generated output.