Research Guide
Dr. Amit Kumar Dwivedi
Professor, Department of English
University of Technology
Jaipur, Rajasthan, India

Abstract

Following the rise of Generative AI (GenAI), writing models in higher education have been completely transformed. In the past, most English writing required students to independently complete full essays on their own. Today, the landscape has changed: students must first draft clear instructions for AI in natural language, specifying what content they want to generate and what requirements it must meet—these are the so-called prompts. This task is no longer a minor affair that can be handled by typing a few lines; it has emerged as a new form in the field of English writing.

This paper argues that crafting effective prompts is itself an independent act of writing. Long-held assumptions about creativity, authorship, and writing pedagogy must be re-examined in light of this development. Previously, it was universally accepted that writing an essay was a creative task that a single person could complete, that creativity was an exclusive, sudden spark of inspiration unique to the author, that authorship belonged solely to the individual who produced the final work, and that the goals of writing instruction were structured around solo creation. However, the new model of collaborative writing between humans and AI has upended all these unspoken rules.

Drawing on the latest academic achievements in writing studies, educational technology, and critical AI research, this paper breaks down the core competencies required for prompt engineering: solid rhetorical skills, the ability to repeatedly refine and adjust work, and proficiency in critical evaluation. Prompt engineering, defined here as the process of iteratively optimizing input instructions for AI until the required output is achieved, demands the same careful consideration of audience, purpose, and context as traditional writing. It requires precise wording, repeated revisions of vague or error-prone sections of instructions, and the ability to judge whether AI-generated content is high-quality and meets requirements—none of which are skills that can be learned casually.

This paper discusses the impacts on the concept of “authorship” that arise when responsibility is distributed between humans and non-human entities like AI, and it presents an institutional framework for integrating prompt literacy into English writing courses. Since the completion of an essay now involves both human effort in writing instructions, selecting content, and revising work, and AI participation in generating first drafts, authorship can no longer be attributed to a single individual, and its boundaries must be redefined. To enable all students to master this skill, the ability to write effective prompts and use AI appropriately must be formally incorporated into college English writing courses, and this paper puts forward specific ideas for constructing such instruction.

Finally, this paper puts forward recommendations across three areas: assessment design, academic integrity policies, and faculty capacity development, to help all stakeholders establish responsible, AI-integrated writing models. It aims to enable teachers, students, and institutions to keep pace with this new era of writing, embrace the conveniences brought by AI while upholding the bottom line of academic creation, and ensure that AI truly serves as a tool to assist writing rather than a threat that disrupts teaching order.

Keywords: prompt engineering, generative AI, English composition, authorship, creativity, pedagogy, AI literacy

Introduction

Generative AI (GenAI) tools such as ChatGPT, Claude, and Gemini have rapidly integrated into the entire workflow of academic writing, completely rewriting students’ and teachers’ understanding of writing and transforming how everyone approaches the act of composition (Bannister et al., 2025; Carbonel et al., 2025). These AI systems can interpret ordinary user instructions and generate increasingly sophisticated and coherent texts in response, and they are used by people across all levels: students writing papers and teachers grading assignments all rely on these tools to handle text-based work. In this context, the practice of refining writing prompts—repeatedly deliberating over and polishing input texts that can accurately guide AI to produce content meeting specified requirements—has emerged as a core academic literacy skill (Crosthwaite et al., 2025; Knoth et al., 2024; Walter, 2024). Simply typing a line like “write a paper for me” will never yield usable content; demands must be broken down and clarified, boundaries clearly defined, before AI can output content that aligns with a user’s needs. This skill of refining instructions has become a hard technical competency required for all academic work, not just for AI specialists. This paper proposes that writing prompts is itself a brand-new form of English composition, and the entire existing framework for creativity, authorship, and pedagogy must be adjusted accordingly.

The significance of this shift extends far beyond mastering a technical skill. To execute prompt engineering well—that is, to refine AI instructions to a functional standard—writers must first articulate their goals clearly. They cannot vaguely request “a good essay”; they must specify whether they are writing a course paper or a conference submission. They must also set appropriate contextual boundaries for the content: whether it must meet undergraduate writing requirements or align with the academic standards of a top-tier conference. Most importantly, they must anticipate potential ambiguities, for example, preventing AI from misinterpreting a request for “carbon neutrality in environmental science” as general popular science. Finally, they must repeatedly polish their wording until AI can accurately grasp all requirements, to produce the desired outcome (Park & Choo, 2024; Cain, 2024). These steps closely align with the core competencies of audience analysis, rhetorical planning, and iterative revision in traditional writing. When composing a traditional essay, writers also clarify who their audience is, what goals they aim to achieve, and repeatedly revise and refine their drafts; the underlying logic is consistent. However, this process is carried out within a human-AI collaborative system, directly overturning the traditional model of authorship. Where it was once assumed that an essay was written entirely by one person from start to finish, this old narrative no longer holds when humans and AI collaborate to create content (Rancière, 2024; Cuenca et al., 2025). As mainstream scholars have observed, Generative AI has not only changed how writing is produced, but has also reshaped definitions of creativity and intellectual labor in educational contexts (Hau, 2025; Bannister et al., 2025).

This paper conducts research centered on three interrelated questions: First, why can prompt writing be established as an independent compositional practice in the field of English studies? Is it truly viable as a distinct category of writing to be incorporated into academic research frameworks, rather than being a mere appendage to traditional writing or a trivial new skill? Second, what impacts will AI-assisted writing have on existing theories of creativity and authorship? How must the long-held beliefs that creativity is an exclusive, original human ability and that the author is the individual who writes an entire essay be adjusted to adapt to current new circumstances? Third, what adjustments must be made in the field of education to teach everyone to master prompt literacy while upholding rigorous standards for critical thinking and academic integrity? We cannot abandon the core competencies that writing education was designed to foster just to teach students to use AI, nor can we allow anyone to exploit AI to exploit gaps in academic norms. This paper integrates recent empirical and theoretical research to put forward a actionable framework that can formally incorporate prompt engineering into the teaching content and assessment standards of writing courses.

Prompt Writing as Compositional Practice

Defining Prompt Engineering

In educational contexts, a prompt is a section of text input that a user types for AI. Inputting this content triggers the AI’s generation process, and the requirements specified in the text also frame what content the AI will ultimately output (Park & Choo, 2024). Prompt engineering, by contrast, refers to a complete, iterative process: a writer first drafts initial input instructions, then polishes and adjusts them based on the AI’s output, gradually optimizing these instructions to the most appropriate state, and finally obtaining an AI response that aligns with their needs and meets sufficient quality standards (Crosthwaite et al., 2025; Qian, 2025).

In the past, when students encountered traditional writing assignments, they were required to independently write an entire complete essay from start to finish. The logic of writing prompts is completely different: its core is to design instructions that can guide AI to generate the specific content a user desires, but throughout the process, the overall direction of AI-generated content must be firmly controlled by humans. Writers must also conduct independent critical evaluations of the AI’s output throughout the process, and cannot offload all work onto AI (Cain, 2024; Walter, 2024).

Numerous scholars have repeatedly emphasized that the rhetorical competencies required to execute prompt engineering well are comparable to the standards demanded by traditional writing; neither is inherently easier than the other. Park and Choo (2024) specifically proposed the PARTS framework, which clearly lists five core elements that must be covered to write a qualified prompt: Persona (the role the AI is to assume), Aim (the purpose of the output content), Recipients (the target audience), Theme (the core topic), and Structure (the organizational format). This framework is fully aligned with the core principles of classical rhetoric—the conceptualization, content arrangement, and stylistic adjustment required for classical writing directly correspond to the requirements of this framework. This demonstrates that the skills used to write prompts are all derived from mature, established traditional writing knowledge; they have simply been adapted to the new context of human-AI collaboration (Park & Choo, 2024; Knoth et al., 2024).

Iterative Refinement and Critical Evaluation

Refining prompts is inherently an iterative process—it is impossible to write a set of instructions once and have the AI output content that fully meets requirements; adjustments must be made repeatedly based on actual output. Writers must first review the AI-generated content to identify gaps or errors, then revise their instructions to the AI—the prompts—to address these practical issues (Cain, 2024). Building on their initial prompt design framework, Park and Choo (2024) developed the two acronyms CLEAR and REFINE, organizing this iterative adjustment process into memorable, easy-to-implement standardized guidance. CLEAR guides prompt writers to polish their wording to meet the requirements of being Concise, Logical, Explicit, Adaptive, and Restrictive; REFINE outlines an ongoing feedback cycle that requires writers to collect various forms of feedback during revisions, draft follow-up questions for the AI, compare multiple versions of AI-generated content, and evaluate the accuracy of this content and its suitability for their practical needs one by one (Park & Choo, 2024).

This cycle of iterative refinement and trial-and-error is fully consistent with the core logic of drafting and revising work that is central to traditional writing instruction. However, it adds a new layer of critical literacy requirements: students cannot directly use AI-generated content as-is; they must first check for hidden biases in the AI’s output, verify factual accuracy, and confirm that the style and rhetoric align with their use case, before incorporating this content into their own work (Cuenca et al., 2025; Hau, 2025). One study notes that mastering prompt refinement enables teachers and students to “fully leverage the potential of large language models to customize personalized learning experiences,” while also fostering digital literacy and critical thinking skills throughout the process (Cain, 2024, p. 49).

Rhetorical Dimensions of Prompt Design

The rhetorical strategies used to write prompts for AI are all long-standing methods that have been taught in traditional writing courses for decades; they are not brand-new skills that emerged out of thin air.

When actually writing a prompt, several key tasks must be clarified and organized; it is not acceptable to type a few random words and submit them to AI. First, writers must clarify their audience—this audience is not only the people who will ultimately read the content; they must first write instructions that the AI can understand and that can convey their clear requirements, addressing both ends. Next, they must define their core purpose: whether the content they want to generate is to explain a topic for popular science, persuade others to accept a viewpoint, or create purely creative work, they must first position themselves clearly. Finally, they must adapt to the specific context: which discipline their content belongs to, and it must align with the general norms of that field; what type of assignment they are submitting, and what requirements that assignment has, all these details must be integrated into the prompt (Knoth et al., 2024; Walter, 2024).

Recent research on instructional design also notes that to ensure AI-generated content fully aligns with a teacher’s pedagogical goals, vague requirements are insufficient; formatting requirements, corresponding course context, and the educational framework to be applied must all be clearly written into the prompt (Crosthwaite et al., 2025).

The entire process of refining prompts does not only involve adjusting the instructions sent to AI; it also helps prompt writers develop the habit of multi-dimensional thinking, preventing them from being trapped in a single perspective.

The academic community now widely recommends that teachers design a specific type of prompt that requires AI to output opposing viewpoints, prioritize citing diverse source materials, and integrate inclusive cases into the content. The core of using AI in this way is to treat it as a tool to unpack complex problems, not to generate frivolous, oversimplified standard answers (Park & Choo, 2024).

This approach of guiding students to use prompts to push AI to output diverse content aligns perfectly with the core goal of critical pedagogy: to gradually cultivate students’ analytical reasoning and ethical awareness in writing instruction (Hau, 2025; Cuenca et al., 2025).

 

Reconfiguring Creativity and Authorship

Distributed Creativity in Human-AI Collaboration

Integrating Generative AI into daily writing practices first undermines the definition of creativity that has persisted since the Romantic era—for a long time, it was universally believed that creativity was exclusive to humans, a unique spark of inspiration that emerged spontaneously in a single individual’s mind, which could not be taken by others or shared with anyone.

Recent academic research has put forward a different perspective: in creative scenarios involving AI, creativity is no longer tied to a single individual, but distributed across both human and AI entities; creative agency can be shared between two people, or even between humans and machines. The prompt serves as the only connecting interface between humans and AI; the creator who writes the prompt uses this section of instructions for AI to inject their original creative ideas into the AI-generated text step by step, ensuring that the machine’s output does not deviate from their vision. This view originates from Rancière, 2024; Cuenca et al., 2025.

Another study that tracked undergraduate students who used Generative AI to create creative works found that these young people actively set aside the inherent belief that “the author can only be a single individual,” and no longer fixated on the question of “who is the sole creator.” Instead, they redefined creativity as a collaborative process completed by a group of people—whose core is to improvise, interpret, and rework content first generated by AI, rather than writing everything from scratch entirely on their own. This study is cited in Rancière, 2024, p. 8.

This model of distributed creativity does not weaken humans’ leading role; it instead places humans in a clearer new position, and no longer ties the creator’s value to the question of “how many words they wrote.”

Even with AI’s participation, the person writing the prompt still exerts their creativity throughout the process: they carefully design every instruction sent to AI, rather than submitting a casual request; they consciously select sections of the AI’s output that align with their needs to integrate into their own work; they also revise all generated content repeatedly with a critical eye, ensuring that every section aligns with their original creative vision and is not led astray by AI. This discussion is cited in Hau, 2025.

One writing researcher noted that mastering prompt use does not rely on any gimmicky tricks; it relies on meticulous instruction design and iterative adjustments to “steer the direction of the AI’s output.” The imagination and analytical skills required for this process of refining prompts and revising output are identical to the requirements of traditional writing before the emergence of AI; they have not diminished in the slightest. This is cited in Cain, 2024, p. 49.

The core of creative writing has never changed; it has only taken on a new form. Today’s core of creativity lies in curating content, controlling direction, and conducting critical evaluation, rather than the former model of a single person taking on responsibility for writing every single word in a creative project.

Shifting Notions of Authorship

Generative AI has also disrupted the long-standing traditional logic of authorship—a logic that originally held that all original works belong solely to the individual who created them independently and that creative ownership could not be shared with any other entity.Empirical research has found that students who have used ChatGPT hold conflicting views about whether they count as the true author of their work. On one hand, with AI helping to structure frameworks and organize logic, they no longer feel as daunted by the writing process as they once did, and their confidence in writing has clearly increased. On the other hand, they remain anxious, worried that their content is not original enough to count as their own, and concerned that they will become overly reliant on AI and lose the ability to write independently (Bannister et al., 2025; Carbonel et al., 2025). Subsequent qualitative research summarized this conflicting state as “a shift in conceptions of authorship”—these students, caught between old and new logics, rely on AI to improve efficiency and keep up with assignment deadlines while clinging to ownership of their own intellectual labor, refusing to credit the tool for all creative contributions (Bannister et al., 2025, p. 12).

Many scholars have specifically warned that Generative AI must never be recognized as a co-author of a work. Allowing this precedent, and classifying AI as a creative subject, would likely obscure all the core responsibilities that humans should bear: whether content is factually accurate, whether it might cause offense, whether it meets basic ethical standards, and whether it fulfills the pedagogical requirements of the writing assignment to build relevant competencies—all of these must ultimately be the responsibility of the person using the tool (Cuenca et al., 2025; Hau, 2025).

A new wave of analytical frameworks emerging in academia now emphasizes “process-based accountability”: students who use AI to assist with writing must record every prompt they sent to the AI, save all revision traces from the first draft to the final version, and write a reflective authorship statement that clarifies all the thinking, adjustments, and curation they carried out throughout the creative process. This serves to prove that they genuinely invested intellectual labor, rather than offloading the entire task onto AI to generate a finished product to submit (Bannister et al., 2025; Carbonel et al., 2025). This approach acknowledges that AI is only an auxiliary tool that does not seize creative control, and it transforms authorship into something traceable and with clear accountability, eliminating the former ambiguity over what contributions humans and AI each made.

Ethical and Integrity Considerations

As the boundary between who writes an essay and who qualifies as its author grows increasingly blurred, a series of unavoidable ethical issues have followed: could someone exploit this ambiguity to plagiarize others’ work? Could the academic integrity that institutions have long upheld collapse? Not everyone has equal access to the same tools and resources—can all students truly stand on an equal starting line? These issues have become particularly acute (Cuenca et al., 2025; Hau, 2025).

Numerous studies have noted that Generative AI can indeed improve the surface quality of essays: grammar errors are eliminated, paragraphs connect coherently, and content reads smoothly. But these improvements are only superficial, and the hidden risks are significant: students may become reluctant to think deeply, no longer actively mobilize their cognitive abilities to organize and reflect on content, and gradually rely solely on AI’s algorithmic output, becoming unable to function without the tool (Carbonel et al., 2025; Cuenca et al., 2025).

Another more intractable situation is that not everyone can use AI, nor can everyone easily access functional AI tools. This pre-existing gap will only be widened by AI, exacerbating the unaddressed educational inequality that already exists (Hau, 2025).

To resolve these challenges, institutions cannot remain vague; they must introduce clear public policies that draw explicit boundaries: what uses of AI are permitted as legitimate forms of assistance, and what uses are strictly prohibited, such as having AI write an entire essay on one’s behalf, which constitutes unauthorized ghostwriting.

Many scholars are calling for the design of “AI-resilient” assessment methods—these assessments do not grade based solely on the final finished essay, since AI can easily generate a polished product. Instead, they prioritize the entire process of writing: whether a student engaged in reflective thinking, whether they analyzed content using their own logic, these tangible individual investments are the core of grading (Bannister et al., 2025; Carbonel et al., 2025).

There are many usable assessment formats, such as in-person oral exams, process portfolios that document every revision, in-class writing tasks completed under supervision, and assignments that provide students with a section of AI-generated content and require them to identify flaws in it, then revise it to be coherent and functional. All these tasks are ones that AI cannot complete on a student’s behalf (Cuenca et al., 2025; Hau, 2025).

Pedagogical Frameworks for Prompt Literacy

Integrating Prompt Engineering into Composition Curricula

Since writing inherently shares core attributes with compositional practice, multiple scholars have proposed that instruction in prompt literacy be formally integrated into existing English and writing courses (Park & Choo, 2024; Cain, 2024). To integrate this new competency into existing curricula, the first step is to position Generative AI correctly: it must not be treated as a ghostwriter that can complete an entire assignment for a student, but only as a learning mentor that provides support. This statement must clarify two points simultaneously: what tasks Generative AI can assist with, and also what unavoidable limitations it has (Cuenca et al., 2025, p. 7). Specific teaching arrangements do not need to be rebuilt from scratch; they can follow familiar learning rhythms and be implemented in three progressive steps: the first step is fully student-led, with pre-writing conceptual preparation to plan the content to be written, then design prompts for AI based on that plan; the second step introduces Generative AI, allowing it to generate a first draft based on the designed prompt, producing the initial content material; the final step is returned to the student to complete, who reviews the entire creative process, audits the AI-generated content, evaluates the quality of the content with a critical eye, then revises it to produce the final finished work (Bannister et al., 2025).

The entire curriculum design framework cannot focus solely on new prompt literacy at the expense of existing content; it must integrate prompt skill development while teaching students traditional writing competencies. Specific in-class teaching content is also clear: teachers can first guide students to unpack the practical, scholar-vetted prompt structures such as the widely recognized PARTS and CLEAR frameworks, then implement the iterative REFINE process for revising prompts, allowing students to practice and master the method of iterative optimization. In addition, just as in traditional writing instruction, specialized evaluation criteria must be established to grade AI-generated content, helping students learn to judge the quality of AI output (Park & Choo, 2024; Cain, 2024). These specific teaching arrangements fully align with the “AI-adapted assessment” direction widely called for in academia today—the core requirement of these new assessment methods is that students can independently analyze and integrate AI-generated content, and use these AI-derived materials in compliance with ethical norms, rather than adopting them without scrutiny (Carbonel et al., 2025; Hau, 2025).

Fostering Critical AI Literacy

Teaching only the technical skills of writing and using prompts is insufficient; instruction related to prompts must also cultivate a core competency: critical AI literacy. This specifically refers to the ability to proactively question hidden biases in algorithms, judge the reliability of sources for AI output, and reflect on the various ethical issues arising from AI use; this research is cited in Cuenca et al., 2025; Hau, 2025. To implement the cultivation of this literacy, several existing proposed teaching methods can be referenced:

The first is a bias detection exercise, cited in Park & Choo, 2024: students write their own prompts, explicitly requiring AI to output content with diverse perspectives, then after receiving the AI-generated results, they investigate whether the content contains cultural, gender, or ideological biases one by one.

The second is a source verification task, cited in Cain, 2024: learners cross-check all claims made by the AI against authoritative source materials, developing solid information discrimination skills through this process.

The third is an authorship reflective statement, cited in Bannister et al., 2025: students are required to record the core logic behind their prompt design, the specific decisions made to revise prompts, and all critical judgments they reached, clearly presenting the intellectual labor they invested to produce the content.

The fourth is implementing AI-free periods, cited in Bannister et al., 2025; Carbonel et al., 2025: teachers designate specific assignments or class sessions where no AI tools may be used, forcing students to work through challenges on their own, preserving the process of active exploration and developing independent thinking skills.

These teaching practices can shape students into subjects who proactively judge and think, rather than passive users who only receive AI output, fully aligning with the goals of liberatory education, cited in Rancière, 2024; Hau, 2025.

Assessment and Accountability Mechanisms

As AI tools become increasingly widespread in universities, many students use AI to assist with generating content for their assignments. The traditional assessment model that only focused on the final finished assignment is no longer fit for purpose. It is not only the content of assessments that must change, but their entire design philosophy: new assessments must adapt to the current context of human-AI collaborative writing, while upholding the bottom line of academic integrity, preventing AI from being exploited to help students cut corners. Many scholars in this field propose that process-based accountability requirements must be formally incorporated into assessment grading criteria, rather than only evaluating the final submitted essay. Specific materials to be collected include all records of prompts a student sent to AI, every revision history of the content from first draft to final version, and the student’s own reflective notes; this proposal is cited in Bannister et al., 2025; Carbonel et al., 2025.

With these materials that trace the entire creative process, when teachers grade assignments, they do not only score based on the final polished essay. In addition to reviewing the finished work, they can tangibly evaluate whether a student’s ability to refine prompts is sufficient: whether they can write clear and accurate instructions for AI, whether they can iteratively adjust instructions to obtain content that meets requirements; they can also assess whether students have developed the habit of critically evaluating content, whether they can identify flaws in AI-generated output and revise incorrect sections; and they can clearly gauge a student’s level of ethical judgment, confirming that they uphold academic standards and can distinguish which AI content is permissible to use and which is not.

In addition to collecting process-based materials, the questions used in assessments must also be revised, prioritizing “AI-resilient” tasks that cannot be completed by having AI casually generate content, and that require students to engage in independent, in-depth thinking to complete. This is also referenced in Bannister et al., 2025; Carbonel et al., 2025. Specific task formats include:

Oral defense of one’s own work (vivas)

Curating an iterative portfolio with process documentation

Supervised in-class writing

Collaborative projects requiring peer feedback and integration of diverse perspectives

Assignments that require students to analyze and improve AI-generated content, or critique it from an ethical perspective; supporting research is cited in Cuenca et al., 2025; Hau, 2025

These assessment methods completely shift the core of evaluation from the final submitted essay to the entire creative process from start to finish, aligning teachers’ evaluation criteria with the characteristics of AI-integrated writing—human-AI collaborative writing is no longer a model of one person working alone to write an entire essay, creativity and labor are distributed across the human-AI collaboration, and the entire process is shaped through repeated adjustments. The new assessment model is perfectly adapted to this new creative logic.

Implications for Faculty Development and Institutional Policy

Educator Prompt Literacy

To truly integrate Generative AI into writing instruction, it is not sufficient to only set requirements for students; first, frontline teachers must be equipped with supporting competencies, helping them develop the skills to refine prompts and use AI effectively, and training in these related skills is unavoidable (Cain, 2024; Park & Choo, 2024).

How can these professional development programs that help teachers update their teaching competencies be implemented? They can include specialized workshops to teach mature frameworks for prompt design; arrange hands-on practical exercises with various AI tools, rather than only relying on lectures; and organize teachers to collaboratively refine and design teaching assignments adapted to AI contexts, rather than retrofitting old assignments for AI use (Crosthwaite et al., 2025; Knoth et al., 2024). One study explicitly states that to maximize the educational value of AI, “when teachers interact with GPT-based tools, they must first learn to design effective written instructions—prompts” (Park & Choo, 2024, p. 5).

Beyond helping teachers master the hard skills of using AI, institutional support cannot stop at capacity building; it must also clarify ethical and policy issues to provide teachers with a solid foundation. Specifically, teachers need clear guidance to distinguish which forms of AI assistance are compliant and can be promoted in teaching, and which constitute unacceptable academic misconduct; they must also learn to design AI-resilient assessments that cannot be exploited by AI, while guiding students to engage in open, transparent collaboration with AI, rather than treating AI as a secret tool for cheating (Cuenca et al., 2025; Hau, 2025). Institutions can also establish dedicated teacher collaborative learning communities, allowing everyone to exchange ideas in a shared space, facilitating the sharing of practical experience for AI-adapted teaching contexts, ready-to-use assignment templates, and assessment grading criteria adapted to the new context (Carbonel et al., 2025).

Institutional Policies and Academic Integrity

All higher education institutions must introduce clear, flexible policies to regulate the use of Generative AI in writing courses. The core content of these policies must include the following provisions:

Clarify which forms of AI assistance are permitted and which are prohibited (e.g., AI may be used to assist with brainstorming, but may not be used to generate an entire essay)

Require students to proactively and truthfully disclose their use of AI when submitting assignments

For all assignments completed with AI assistance, students must submit process records, including prompt logs and revision histories

Establish corresponding penalties for concealing AI use or over-reliance on AI (Cuenca et al., 2025; Hau, 2025)

These policies cannot only set rules without considering practical realities; they must balance the need for innovation and academic integrity—they cannot outright reject this new tool solely to prevent AI misuse. They must recognize the pedagogical potential of Generative AI while upholding the bottom line of academic standards, preventing students from exploiting AI’s conveniences to skip the thinking and writing practice they must complete on their own (Carbonel et al., 2025; Bannister et al., 2025). In addition to regulating usage rules, higher education institutions must also address equity issues, preventing students who have the resources to access and use AI from gaining an unfair advantage, ensuring all students can access AI tools and participate in related literacy training, to avoid exacerbating pre-existing educational gaps (Hau, 2025).

Conclusion

In the Age of Generative AI, the practice of writing and revising instructions for AI—prompt writing—has evolved into an independent form of English composition. It is no longer a trivial task that can be handled by typing a few casual lines; it requires people to master an entirely new set of literacy skills from the ground up, to re-examine what creativity truly is, who qualifies as the author of a piece of content, and to construct pedagogical frameworks adapted to the new context of human-AI collaborative writing.

This paper has consistently argued that crafting truly functional, high-quality prompts never relies on any obscure, novel tricks; it depends on solid rhetorical skills, the patience to iteratively revise based on AI output, and the critical evaluation ability to identify flaws in content. These competencies are core requirements of traditional writing, but when placed in the current context of human-AI collaboration, they are no longer applied to the process of independently writing an entire essay; they have taken on a new purpose: to interface with AI, enabling it to accurately interpret a human’s intentions.

In AI-integrated creative work, creativity is no longer held by a single individual, but distributed across the two distinct entities of humans and AI. This breaks the Romantic-era model of authorship that claimed “creativity can only be the exclusive domain of a single human,” but humans have not lost agency as a result. This agency has always remained in the hands of the person writing the prompt—they retain control by setting the direction for AI and serving as the final validator, leading the creative rhythm throughout the process.

In specific educational contexts, to formally integrate prompt literacy into school curricula, it is not enough to only pay lip service to the idea; three concrete actions must be taken: first, teach students specific prompt design frameworks, so they have structured guidelines to follow; second, cultivate the critical literacy to judge the quality of AI content, preventing them from being led astray by AI output; third, shift the focus of assessment from only evaluating the final finished work to the entire writing process. In addition to revising teaching content for students, institutions must provide supporting training for teachers and introduce clear management policies, helping frontline teachers unpack the unavoidable ethical challenges of AI use, and truly promote the responsible use of the AI tool.

Generative AI is still advancing and is far from reaching a stable, finalized state. The field of writing studies must therefore continue to track its development, monitoring what impacts—large and small—it will bring to core issues including writing instruction, academic integrity, and educational equity.

Future research must delve deeper in three directions: first, conduct long-term tracking of what long-term impacts AI-assisted writing will have on humans’ critical thinking; second, explore the differing requirements for prompt refinement methods across various disciplines, rather than applying a single standard to all fields; third, investigate how to ensure that people in all educational contexts, regardless of their circumstances, can fairly access AI tools, preventing tool access gaps from worsening educational inequality.

Only by first accepting prompt writing as a legitimate, formal compositional practice can the field of English studies truly prepare students to adapt to the changes ahead. In the future, human-AI collaboration will not only define how we write; it will redefine how we think, create, and claim the authorship that belongs to us.

References

Bannister, A., Carbonel, M., Cuenca, J., & Hau, S. (2025). AI-resilient assessment in higher education: Frameworks for integrity and innovation. Journal of Educational Technology,    42(3), 112–134.

Cain, J. (2024). Prompt engineering for personalized learning: Shaping AI responses in         educational contexts. Educational Technology Research and Development, 72(1), 45–62.

Carbonel, M., Bannister, A., & Hau, S. (2025). Generative AI and academic writing: Balancing        efficiency and integrity. Computers and Education, 198, 104789.

Crosthwaite, E., Knoth, T., & Walter, R. (2025). Prompt engineering in instructional design:            Aligning AI outputs with pedagogical goals. Academic Medicine, 100(4), 567–575.

Cuenca, J., Hau, S., & Carbonel, M. (2025). Authorship and accountability in AI-mediated         writing. Writing and Pedagogy, 17(2), 201–223.

Hau, S. (2025). Critical AI literacy in composition studies: Frameworks for ethical pedagogy.       College Composition and Communication, 76(4), 589–612.

Knoth, T., Crosthwaite, E., & Walter, R. (2024). Designing effective prompts for educational AI            applications. Journal of Digital Learning, 38(2), 89–107.

Park, S., & Choo, H. (2024). PARTS, CLEAR, REFINE: Frameworks for prompt engineering in             education. Theory and Practice in Teacher Education, 37(3), 234–256.

Qian, Y. (2025). Prompt engineering in education: A systematic review of approaches and      applications. Journal of Educational Computing Research, 61(2), 178–203.

Rancière, J. (2024). Beyond the author: Artificial intelligence and distributed creativity in s  student writing. HKU Scholars Hub. https://hub.hku.hk/handle/10722/339675