Summary

This article addresses the question, “Can I Use Generative AI in Ways That Avoid Academic Dishonesty & Plagiarism?”

The emergence of GAI (Generative Artificial Intelligence) tools has blurred the lines between academic integrity and plagiarism. These tools not only assist in traditional stages of writing—such as prewriting, drafting, editing, and proofreading—but also enhance the creative process itself, serving as a Thought Partner for developing and critiquing ideas, a Research Assistant for gathering and synthesizing sources, a Composition Assistant for refining ideas and structure, a Designer for creating visual elements, an Editor for polishing the prose, a Citation Assistant for managing references, and a Publishing Assistant for finalizing and distributing work across platforms. As AI continues to shape how we create, the boundaries between human authorship and AI-assisted collaboration grow increasingly complex, raising important questions about authorship, academic integrity, plagiarism, and copyright.

Traditionally, academic dishonesty encompasses actions like submitting someone else’s work as one’s own, copying without proper attribution, and receiving unacknowledged assistance. But with AI, things are not so clear-cut. If a writer prompts an AI to generate text and presents it as their own work without acknowledging the AI’s contribution, it may not be traditional plagiarism—after all, AI doesn’t “own” the text. However, it could still be viewed as academically dishonest if the writer doesn’t disclose the use of AI. Especially in school settings where teachers engage students in writing to learn deeply about a subject, a student’s superficial use of AI that is unacknowledged misrepresents the process and level of effort involved in creating the work.

According to U.S. copyright law, writers do not own AI-generated text, as it lacks the human authorship necessary for copyright protection (Federal Register, 2023). Only the original human contributions—such as creative decisions made during the writing process—are eligible for copyright. This legal reality complicates questions of ownership and academic honesty, leaving students and professionals vulnerable to allegations of plagiarism if they fail to clearly distinguish their own contributions from AI-generated content.

Most educators would likely consider it academically dishonest if students superficially engage with AI tools and present AI-generated text as their own. However, if a writer actively engages with AI-generated text—cutting words, adding others, revising, and refining the content to better meet their rhetorical goals—is this dishonest? Does the author own the copyright of substantively revised works?

The concepts of authorship, plagiarism, and copyright need to evolve to account for how writers are using AI to compose. In the future, AI may be perceived as just another tool a writer uses to express themselves, much like a word processor, grammar checker, or visualization tool. Authorship may be considered legitimate if writers remain in charge of their creative processes, interrogating every word and ensuring their inner speech and perception of the rhetorical situation drive the narrative. In this model, AI becomes a collaborative partner rather than a substitute for original thought and composition. The integrity of the work depends on the writer’s active role in crafting and controlling the content.

Presently, some educators view any AI-generated work as academically dishonest — and students certainly need to practice audience awareness and follow their teacher’s and institution’s plagiarism and academic integrity policies. Rightfully so, educators worry use of AI will diminish human agency, students’ cognitive development, and student learning.

Others, however, argue that AI is yet another technology in an ongoing evolution of communication technologies. They point to past disruptive technologies—such as the pen, the word processor, or the printing press. They note these tools have revolutionized how we communicate–that changes in technology do more than facilitate communication: they transform human agency and redefine what it means to be human. They argue the affordances and constraints of today’s hardware (e.g., Blackwell GPU) and software (e.g., Chato1-preview) remediate how we research, solve problems, interpret information, and think, compose, and create (Rainie and Aderson, 2023)

For example, the printing press not only democratized knowledge but also disrupted existing power structures by spreading ideas, empirical research methods, and textual criticism, thus contributing to the broader “conversation of humankind” (Bolter, 2001). As technorhetoricians like Dobrin (2002) argue, writing is always evolving, constantly shaped by new technologies. From this perspective, adopting AI tools is simply the next step in our quest to express ourselves and accomplish our goals more efficiently. Ultimately, as Lessig argues in Free Culture (2024), if a tool makes a process a human wants to do easier, it will be widely adopted.

The article below analyzes these questions by reviewing the APA and MLA conventions for citing AI-generated texts. It critiques these guidelines for being inconsistent and impractical in seven key areas: (1) they fail to account for how AI is integrated into all stages of writing, from prewriting to publishing; (2) they risk overwhelming texts with excessive citations, disrupting readability and making them difficult to follow; (3) they do not distinguish between minor AI assistance (e.g., grammar checks) and significant content generation; (4) they fail to acknowledge the iterative nature of AI use, where students engage with AI throughout the writing process; (5) they provide no guidance on how to specify the percentage of AI involvement in a text; (6) they lead to cluttered reference lists with multiple entries for the same tool used at different stages; and (7) they do not account for the labor involved in revising AI-generated content, thus failing to adequately reflect human contribution. This lack of clarity leaves students and professionals vulnerable to accusations of dishonesty—even if they believe they are following the rules.

Based on scholarship and research in writing studies, this article recommends that authors include a “metaprocess footnote” that explains how they used AI—whether as a (1) Thought Partner, (2) Research Assistant, (3) Composition Assistant, (4) Designer, (5) Editor, (6) Citation Assistant, or (7) Publishing Assistant. This initial model is broad, making it applicable across academic and professional fields, providing a flexible framework for documenting AI involvement in writing processes.

Additionally, the article presents three more detailed models specifically designed for educational settings, where teachers can use them as accountability measures and evaluation criteria:

1. Rhetorical Processes: This model focuses on how AI tools are integrated into rhetorical reasoning and analysis, helping writers develop an appropriate rhetorical stance based on audience, context, and purpose.
2. Composing Processes: This model maps AI usage across all stages of the writing process, including prewriting, inventing, drafting, collaborating, researching, planning, organizing, designing, rereading, revising, editing, proofreading, sharing or publishing. This framework is particularly useful in writing classrooms, where educators can assess how students use AI tools throughout the entire composing process, ensuring students remain engaged and responsible for their creative decisions.
3. Style: This model addresses how AI tools contribute to the refinement of style—including elements such as clarity, brevity, coherence, flow, inclusivity, simplicity, and unity. Instructors can use this model to evaluate whether students effectively used AI tools like Grammarly or ChatGPT to improve stylistic features of their writing while maintaining their voice and control over the content.

These four models provide a comprehensive approach, allowing writers and educators to integrate AI in ways that are transparent, accountable, and ethically sound, while ensuring that human agency remains central to the writing process.

AI Citation Guidelines: APA and MLA

Presently, style guidelines promulgated by the APA and MLA advise writers to cite content generated by or significantly influenced by AI.

APA 7th Edition (2023 update)

Format:

• Cite AI tools as software, not as authors
• Include: name of AI tool, version, organization, year, URL (if available)

Example:

In-text citations: (ChatGPT, 2023; Claude, 2024; Perplexity AI, 2024)

Sentence: Estimates of nuclear warhead counts for the top five nuclear powers vary between AI tools: Russia (5977-6257), United States (5550), China (350-410), France (290), and the United Kingdom (225-260) (ChatGPT, 2023; Claude, 2024; Perplexity AI, 2024).

Reference list entries:

• OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com
• Anthropic. (2024). Claude (3.5 Sonnet version) [Large language model]. https://www.anthropic.com
• Perplexity AI. (2024). Perplexity [Large language model]. https://www.perplexity.ai

MLA 9th Edition (2024 update)

Format:

• Cite AI-generated content as a source, not an author
• Include: name of AI tool, version, creator, access date, prompt information

Example:

In-text citations: (OpenAI, “Nuclear arsenals”; Anthropic, “Nuclear weapons count”; Perplexity AI, “Top 5 nuclear powers”)

Sentence: Estimates of nuclear warhead counts for the top five nuclear powers vary between AI tools: Russia (5977-6257), United States (5550), China (350-410), France (290), and the United Kingdom (225-260) (OpenAI, “Nuclear arsenals”; Anthropic, “Nuclear weapons count”; Perplexity AI, “Top 5 nuclear powers”).

Works Cited Entries:

• OpenAI. “Nuclear arsenals of top 5 countries.” ChatGPT, 2 Oct. 2024, chat.openai.com/chat. Accessed 2 Oct. 2024.
• Anthropic. “Nuclear weapons count for major powers.” Claude, 2 Oct. 2024, www.anthropic.com. Accessed 2 Oct. 2024.
• Perplexity AI. “Top 5 nuclear powers and their arsenals.” Perplexity, 2 Oct. 2024, www.perplexity.ai. Accessed 2 Oct. 2024.

Problems with Current APA and MLA AI Citation Guidelines

Misalignment with Real-World AI Usage

Current guidelines assume that AI contributions are discrete and easily identifiable. In reality, writers may use multiple AI tools throughout the writing process. Authors may engage with a dozen tools just to refine a single phrase, making it difficult to track individual AI inputs.

Impracticality for Extensive AI Use

Citation rules become impractical when writers frequently use AI. Using multiple AI tools within a single sentence or paragraph (e.g., for research, summarization, or phrasing) can lead to excessive in-text citations, overwhelming readability and disrupting the natural flow of the text.2. Impracticality for Extensive AI Use

Disruption of Text Flow

Frequent in-text citations for every AI tool used can disrupt readability. MLA’s guidelines, for example, could result in a paragraph filled with citations from different AI tools used across various stages (prewriting, drafting, revising), potentially requiring pages of cited sources for a single paragraph.

Cluttered Reference Lists

The heavy reliance on AI tools can clutter reference lists, leading to lengthy works cited sections that include multiple entries for the same tool used across different points in the writing process. This adds unnecessary complexity to bibliographies.

Lack of Nuance in Contribution

Current guidelines don’t distinguish between minor AI assistance (e.g., grammar checking) and significant content generation. Theses guidelines not acknowledge that writers use AI for various stages, including prewriting, inventing, drafting, collaborating, researching, planning, organizing, designing, rereading, revising, editing, proofreading, sharing or publishing. 

Difficulty in Tracking Iterative AI Use

Writers engage in iterative dialogues with AI tools over extended periods, refining content and revisiting ideas. Current guidelines fail to account for this ongoing interaction, treating AI as a static source instead of a collaborative process that shapes the entire writing journey.

No Mechanism to Specify AI Involvement

The guidelines lack clarity on how much AI involvement requires citation. For example, does an AI suggestion of a few sentences require the same level of citation as a full paragraph generated by AI? This absence of guidance leaves ambiguity around the percentage of AI contributions.

How Should Universities and Publishers Ask Authors to Cite Use of AI?

Below are three frameworks that writers could follow to report their use of AI throughout the composing process. These three models provide transparency about how writers used AI — whether superficially or substantively, and thereby clarify whether they hold copyright per U.S. copyright law. The first model is abbreviated and may work across the disciplines and professions. The second two models are a bit more nuanced and would work well in writing courses where the goal is to introduce students to a broad suite of composing strategies.

Sample MetaProcess Footnotes for Academic and Professional Disciplines

To protect their works from allegations of plagiarism and academic dishonesty, writers should write a “metaprocess footnote” that explicitly states whether they used GAI tools as a (1) Thought Partner; (2) Research Assistant; (3) Composing Assistant; (4) Designer; (5) Editor; (6) Citation Assistant; or (7) Publishing Assistant.

• Thought Partner: AI acts as a reflective partner during the prewriting and brainstorming stages, helping writers think critically about their ideas and the direction they want to take.
• Research Assistant: AI tools help writers identify key research, synthesize multiple sources, and explore new methods to enrich their writing.
• Composing Assistant: AI supports the drafting process by offering alternative approaches, helping writers structure their arguments, and assisting with rhetorical reasoning.
• Designer: AI enhances the visual appeal of documents, helping writers create scannable, visually engaging content.
• Editor: AI tools help writers refine their prose, ensuring stylistic consistency, grammatical correctness, and overall coherence.
• Citation Assistant: AI tools streamline citation management, ensuring proper formatting and organization of sources.
• Publishing Assistant: AI supports the final steps of publication, helping writers prepare content for different platforms and formats.

Examples of MetaProcess Footnotes for Academic and Professional Disciplines

1. I used ChatGPT as a Thought Partner to help generate initial ideas for this project. Throughout the drafting process, ChatGPT prompted me with questions about my thesis and argument structure, which helped refine the focus of the final draft.
2. During the research phase, I employed Semantic Scholar and ResearchRabbit as Research Assistants. These tools helped me identify relevant research studies, track the evolution of scholarly conversations in the field, and map out potential gaps in the literature.
3. In preparing this manuscript, I relied on ChatGPT as a Composing Assistant to suggest ways to organize my argument and improve transitions between sections. I also used Claude to explore alternative perspectives on the topic, which helped diversify the rhetorical strategies used in the discussion section.
4. For the design of this report, I utilized SciSpace as a Designer to create a set of visually engaging charts and infographics that summarize the data. I also used HeyGen to generate an AI-powered video summary, providing a multimedia option for readers.
5. To ensure clarity and style consistency, I used Grammarly as an Editor. It was particularly helpful in identifying sentences that required rephrasing for brevity and clarity, and it highlighted areas where the tone did not align with the formal nature of the document.
6. I used Scite AI and Zotero as Citation Assistants to organize and manage all references, ensuring adherence to the latest APA citation guidelines. These tools also helped me discover relevant articles that were cited in recent studies in my field.
7. For publication, I used SciSpace as a Publishing Assistant to format the document according to the submission guidelines of the targeted journal. It also helped generate an accessible version of the article for online readers by creating a video abstract.

Sample MetaProcess Footnotes for Writing Classrooms

Depending on the assignment, teachers may find it useful to ask their students how they used AI to facilitate

1. Rhetorical Processes
2. Composing Processes
3. Style

AI Support for Rhetorical Processes

• Rhetorical Analysis: This process involves breaking down the elements of a rhetorical situation—such as exigence, audience, and constraints—that influence communication. Writers use rhetorical analysis to understand the dynamics shaping a particular discourse and to critique the effectiveness of various rhetorical strategies. AI tools like ChatGPT can assist in analyzing these components by offering feedback on tone, audience expectations, and potential rhetorical challenges. Additionally, tools like Consensus can help writers summarize and critique texts, providing alternative rhetorical frameworks.

• Rhetorical Reasoning: This analytical process allows writers to decide on the most appropriate rhetorical stance by sorting through different rhetorical moves (e.g., appeals, devices, and genres). Writers engage in rhetorical reasoning to determine how best to respond to an exigency or call for discourse. AI tools like ChatGPT and Perplexity assist by generating multiple argument structures, helping writers evaluate various rhetorical strategies and approaches to a topic. AI can also help writers investigate how scholarly conversations around a topic have evolved and identify gaps in the discourse that need to be addressed.

Sample Metaprocess Footnotes for Rhetorical Analysis

1. In conducting the rhetorical analysis for this essay, I used ChatGPT to help identify the rhetorical elements of the chosen texts, such as their appeals to ethos and pathos, and the constraints shaping the discourse. The AI assisted in generating a comparative analysis of audience expectations between two texts, helping to clarify the persuasive strategies each author employed.
2. To critique the effectiveness of the rhetorical strategies used in this speech, I relied on Consensus to summarize key rhetorical moves and critique the argument’s reliance on logical appeals (logos). ChatGPT provided feedback on how well the speaker adapted their rhetorical stance based on audience constraints.
3. For this project, I used Perplexity to explore how various rhetorical appeals (ethos, pathos, and logos) were employed across different genres, particularly comparing academic articles with media reports. This tool helped summarize how each genre’s rhetorical strategies catered to different audience expectations.
4. During my analysis of the rhetorical situation in this case study, I utilized ChatGPT to provide insights into how exigence and audience influenced the structure of the argument. ChatGPT helped me identify potential areas where the rhetorical devices were either underused or overly reliant on certain appeals, guiding my critique of the argument’s balance.

AI Support Across Composing Processes

Sample Metaprocess Footnotes for Composing Strategies

1. In this essay, I used ChatGPT during prewriting to brainstorm ideas and generate initial topic outlines. I then used ResearchRabbit to explore interdisciplinary connections and track research gaps within related fields. To refine the clarity and flow, I employed Grammarly during editing, and used Scite AI to organize and format my citations.
2. For the revision process, I used ChatGPT to simulate feedback from multiple stakeholder perspectives, which helped me improve the argument’s clarity and coherence. Additionally, I leveraged Rocky.ai to stay accountable for my writing deadlines and reflect on the inductive reasoning present in the draft, transitioning it toward a deductive structure.
3. Throughout my research process, I used Semantic Scholar to identify key studies and applied Voyant Tools for corpus analysis, helping to analyze trends within the data. To prepare the final document for submission, I utilized SciSpace to generate a video summary, making the document accessible in a different format.

Style Categories: Using AI to Enhance Style

This table focuses on how AI tools can help writers refine their style, including clarity, brevity, coherence, and more. It shows how AI can be used to improve specific stylistic elements and make the writing more effective.

Related Concepts

• Plagiarism – Reimagining Authorship & Citation Practices in The Age of AI
• Key Benefits of Writing Without Artificial Intelligence
• The Secret, Hidden Writing Process: How to Tap Your Creative Potential
• The Ultimate Blueprint: A Research-Driven Deep Dive into The 13 Steps of the Writing Process
• The Writing Process – Research on Composing