Lindenwood University Lindenwood University
Digital Commons@Lindenwood University Digital Commons@Lindenwood University
Faculty Scholarship Research and Scholarship
2-2023
Working With (Not Against) the Technology: GPT3 and Arti8cial Working With (Not Against) the Technology: GPT3 and Arti8cial
Intelligence (AI) in College Composition Intelligence (AI) in College Composition
James Hutson
Lindenwood University
Daniel Plate
Lindenwood University
Follow this and additional works at: https://digitalcommons.lindenwood.edu/faculty-research-papers
Part of the Arti8cial Intelligence and Robotics Commons
Recommended Citation Recommended Citation
Hutson, James and Plate, Daniel, "Working With (Not Against) the Technology: GPT3 and Arti8cial
Intelligence (AI) in College Composition" (2023).
Faculty Scholarship
. 490.
https://digitalcommons.lindenwood.edu/faculty-research-papers/490
This Article is brought to you for free and open access by the Research and Scholarship at Digital
Commons@Lindenwood University. It has been accepted for inclusion in Faculty Scholarship by an authorized
administrator of Digital Commons@Lindenwood University. For more information, please contact
Abstract
The use of articial intelligence (AI) for improvement of writing is commonplace with word-processing software and cloud-
based writing assistants such as Grammarly and Microsoft Word. However, more and more options are cropping up that move
beyond assistance with grammar, spelling, and punctuation to complete essay generation. The free availability of AI essay
generators has led to lamenting the coming death of college writing. But AI has been used in the previously noted examples
for decades without such a reaction. In fact, the idea that the use of essay generating software is synonymous with academic
dishonesty is as passé as worries about allowing students to use calculators or chalkboards. Both are tools that emerged by
aording students a dierent type of learning which was not rote memorization.
The questions now become how AI tools can and should be used to teach English composition and to what extent. In the con-
ceptual age where AI is used to augment all other facets of human creativity, providing students with the tools they will need
for eective communication becomes inevitable. These new AI tools may allow students to master grammar and syntax more
quickly in order to move on to important research questions that will contribute to knowledge in their given elds. This study
investigates the current and potential uses of AWE, AAG and AI essay generators in a rst-semester English composition
classroom.
Students in the study were provided with the same assignments and learning outcomes as are standard in English, composition
courses but were encouraged to use AI applications when prompted to discover the usefulness and limitations of such technol-
ogy. Results from the study conrm that use of such tools does not automatically lead to plagiarism or academic dishonesty. On
the contrary, higher-order thinking skills and metacognition are required to use AI tools appropriately to learn writing skills.
Furthermore, the tools themselves became the topic covered in the class for the study and led to further social and ethical
implications.
Citation: Hutson, j. (2023). Working With (Not Against) the Technology: GPT3 and Articial Intelligence (AI) in College Composition.
J Robot Auto Res, 4(1), 330-337.
Volume 4 | Issue 1 | 330
Working With (Not Against) the Technology: GPT3 and Articial Intelligence (AI) in College
Composition
Research Article
James Hutson*, Daniel Plate
Professor, Department of Art History and Visual
Culture, College of Arts and Humanities, Lindenwood
University, St. Charles, United States
Associate Professor, English, College of Arts and
Humanities, Lindenwood University, St. Charles,
United States
*
Corresponding author
James Hutson, Professor, Department of Art History and Visual Culture,
College of Arts and Humanities, Lindenwood University, St. Charles, United
States McCluer Hall 15, 209 S. Kingshighway St. Charles, MO 63301.
Submitted: 21 Feb 2023; Accepted: 24 Feb 2023: Published: 27 Feb 2023
Journal of Robotics and Automation Research
ISSN: 2834-7706
J Robot Auto Res , 2023
Introduction
Adopting emerging technologies in higher education has histori-
cally been met with suspicion and even outright rebellion. When
the chalkboard was introduced to classrooms in 1801, students re-
volted against something perceived to challenge the centuries-old
tradition of memorization and oral exams. As a technology, the
chalkboard allowed a new type of learning where students could
write out information to be learned and shorten time-on-task
(Krause, 2001). Later in the century, the magic lantern was in-
troduced in college classrooms but was slow in adoption due to
the perception of the device as valuable for entertainment and
not learning (Shepard, 1987). These and other technologies have
shaped and reshaped higher education today with another revolu-
tion upon us. With open access to a seemingly unlimited amount
of information, students have immediate access to knowledge once
only gained in the classroom.
The shift in role of faculty is being felt as educators transition from
imparting information to facilitating learning in active learning en-
vironments supported by technology (Brownridge, 2020). As Gen-
erations Z and Alpha enter and move through the college experi-
ence, there is a disjuncture between their previous experiences and
use of technologies and what they encounter. These digital natives
have never experienced a time in their lives without computers,
nor do they remember a time without smartphones (Flynn & Frost,
2021). On the other hand, most faculty teaching today are digital
immigrants. They were not trained to incorporate new technolo-
gies into their classrooms and instead adapted to them later in life
Volume 4 | Issue 1 | 331
(Prensky, 2001). As such, emerging technologies are often slow to
be adopted, even if proven eective, and the blame often leveled
at teachers assuming unwillingness to adopt (Ertmer & Ottenbre-
it-Leftwich, 2010; Howard & Mozejko, 2015).
Educators are, however, already adopting emerging technologies
like AI in their classrooms in various ways, from delivery of con-
tent in an LMS to providing feedback to students. For instance, AI
is closer to being accepted for essay assessment and grading and
is currently seen in applications such as Intellimetric, Packback,
MyAccess! and other automated writing evaluation (AWE) soft-
ware [1,2]. Algorithms are used to provide real-time diagnostic
feedback, assessment, and grading, especially in shortform student
writing assignments. But while AI has been accepted as a tool to
assist instructors with grading and assessment, the same has yet
to extend to acceptance in the process of creating student artifacts
themselves. In fact, studies continue to warn of the abilities AI has
to assist in cheating for students [3].
While so-called “authentic writing” has been held as a standard
for education and research, many scholars are concerned about the
growing diculty in identifying AI-generated essays [3, 4]. While
plagiarism detection software such as Turnitin is standard across
higher education, emerging Automatic Article Generator (AAG)
writing that is powered by AI provides new ways to bypass such
safeguards. Highlights this ability with Generative Pre-trained
Transformer 3 (GPT-3) and OpenAI, warning of the inevitable
inltration of the university. Other researchers, such as, have pre-
sented studies to help academics identify AAG writing and con-
tinue to raise awareness of the perceived danger such technology
represents for academic integrity [4, 5].
Even when attempting to integrate into the classroom in a purpose-
ful fashion, studies frame the activity as encouraging “cheating”
among students on their nal papers instead of how AI may be
used as a useful tool [6].The use of articial intelligence (AI) in
writing has had two recent areas of study. The former is in the
ability to improve proper use of grammar and clear syntax with
cloud-based typing assistants that review spelling, grammar, punc-
tuation, clarity, engagement, and delivery mistakes. Such software
uses AI to identify and search for an appropriate replacement for
the located error. The latter has applications in creative writing to
move beyond grammar and usage. As these tools become more
widely used pedagogically, they have become useful as a means
of generating suggested alternative content for the given goal of
a writing exercise (e.g. marketing, poetry, short stories). Unfor-
tunately, there is often a false dichotomy in discussions of these
tools between the well-crafted essay (supported near the end of the
writing process by grammar/usage AI functions) and the “human”
domain of idea creation and development. In many ways, English
professors would and should expect a better crafted essay when
using AI tools to generate the early drafts or “skeleton ideas” of
an essay.
This support would allow the student to get to the crafting of the
proceeding draft stages of writing more quickly and eciently, al-
lowing the focus to be placed on the content instead of being con-
cerned with the rules of grammar. Many students struggle with the
“blank page” stage of writing and end up with “cookie-cutter” es-
say beginnings anyway. Therefore, using a tool that "jump starts"
the essay-generation process would allow more focus on craft, not
less. As such, this study seeks to determine the most eective ped-
agogical use of AI essay generators for college composition class-
es, and how assignments may be better crafted to accommodate
changes in technology. Students from a rst-year English compo-
sition course were instructed to complete a rhetorical analysis and
AI essay writing collaborative project using AI tools, such as Es-
sayAiLab, OpenAI, and AI Article Writer 3.0. Importantly, the AI
tools selected for the case study were all freely available and were
not designed to complete full-length essays and research papers.
Results from the study indicate that instead of students adopting
these tools to write on their behalf, the process was collaborative
and required additional eort. The collaborative process to write
with an AI (instead of the AI writing for students) did not result in
rampant use for plagiarism or cheating and instead assisted with
the development of higher-order thinking skills to fully leverage
NLP resources. The future of college writing then is not the ban-
ishment of technology from the classroom, but an understanding
of how to craft assignments and adopt new pedagogical approach-
es to prepare students to work with (and not against) technology
that will be required for the future of work.
Literature Review
AI Essay Generators
Software and AI applications dedicated to essay generation are
more varied and accessible to students than ever. The dierent
types of essay generators can be located on a sliding scale from
those that can assist with generating topics to outlines to those that
write full papers and Automatic Article Generator (AAG) writing
that is powered by AI. The most common writing assistants today
for most students are Microsoft Word, included in the Oce Suite,
and Grammarly. In fact, the technology has become so common-
place that one would be hard pressed to nd a writing or research
assignment in higher education that did not require the use of word
processing software to check for spelling and grammar, format re-
search, and package for electronic sharing.
At the same time, advances in NLP and machine learning (ML)
have moved so rapidly that academia is continually challenged
to keep pace and integrate these newer technologies into existing
policies and procedures of teaching and learning. For instance, as
noted the web-based GPT-3 software program, developed by Ope-
nAI (https://openai.com/api/ , is able to generate prose from any
prompt that cannot be detected from anti-plagiarism software as
the output cannot be found elsewhere. The ML algorithm scours
the entire internet each time a new query is submitted and produc-
es a unique output that is always dierent. The outputs from GPT-3
J Robot Auto Res , 2023
Volume 4 | Issue 1 | 332
can be tailored and specic to any form of writing, including op-
eds, jokes, advertisements, and so on.
As noted, given that each prompt costs less than one cent and the
cost to hire a writer to produce a college-level essay is around $15
to $35 per page, the value proposition is quite enticing for students
[5]. And GPT-3 is just one of a growing number of inexpensive
or free options available including Moonbeam (https://www.go-
moonbeam.com/), The Good AI (https://www.the-good-ai.com/),
EssayAiLab (https://www.essayailab.com/), Paper Typer (https://
papertyper.net/), My Assignment Help (https://myassignmenthelp.
com/essay-typer.html), and EssaySoft (https://www.essaysoft.net/
essay-generator.html) to name a few. The initial reaction from ed-
ucators, especially those who teach English composition, has been
to design assignments that thwart a student’s ability to use these
generative tools.
AI in the Composition Classroom
In order to push back the adoption of AI essay generators by stu-
dents, relays several strategies currently employed by college fac-
ulty [5]. For instance, students can be required to draw on mate-
rials covered in class in the essays and to revise work in response
to instructor feedback. As these tools are unable to cite sources
and readily edit content, scaolding assignments is one strategy
to discourage their use. Writing prompts can also be designed to
specically address localized issues not available online. Another
method is to have students produce artifacts that AI is unable to
create, such as PowerPoints, podcasts, or verbal presentations. Fi-
nally, students could be required to complete written assignments
using proctoring software, such as ProctorU or Honor Lock, or
oine in a live proctored computer lab.The preceding examples
all represent eorts that might be adopted to discourage the use of
AI writing tools and focus instead on traditional teaching methods
of college composition, which have a healthy corresponding body
of scholarship [7-12].
Research into the use of AI essay generators in college compo-
sition classes, however, remains in a nascent phase. The limited
studies that have been conducted point to the limitations of us-
ing such tools and the need for instructional interventions on the
part of peers and instructors. for instance, conducted a class-
room-based approach to determine the viability of using automatic
writing evaluation (AWE) software for pedagogical purposes in
the teaching of writing. As a tool, AWE is designed to provide
instant scoring for submitted essays along with diagnostic feed-
back. The study specically looked at the implementation of the
AWE software MY Access! in three EFL college writing classes
in Taiwan to determine how students perceived its eectiveness in
improving their writing. Results indicated a negative perception
of the software by students overall but more positive in the early
stages of the drafting and revising process, which also included
feedback from the teacher and peers later in the process [13].
Using the tool as a surrogate writing coach for students led to
frustration and was found to limit learning of proper writing. also
studied the use of AI to teach English as a Second Language in
Indonesian classrooms [14]. As with Chen and Cheng, students
were surveyed on their perceptions of using an AI application in
the processing of learning to write in English. Overall, students
responded positively to the use of AI in the process, reporting that
not only could AI be used to assist during the writing process and
help with grammar and vocabulary, but also assist in understand-
ing theoretical concepts. However, in reviewing how the assign-
ments were structured, a trend emerges again. The use of the AI
essay generating tools was accompanied by pre-planned interven-
tions on the part of the instructor, as well as communication and
collaboration skills reinforced in groupwork. In other words, stu-
dents found the tool helpful in the learning process as long as it
supplemented traditional feedback instead of supplanting it. The
same results were more recently conrmed in a study by in a study
of the use of AI to support teaching and learning in college creative
writing courses [15].
The study included an NLP application that provided students
with the ability to check their grammar against the principles and
techniques covered in class. The algorithm was also intended to
improve creativity in students; however, results indicated that par-
ticipants were willing to admit the part played by the tool in im-
proving grammar, but not in improving their creativity as writers.
Students did nd the use of the algorithm useful but only as part
of the larger context of peer and instructor interventions in their
writing process. As demonstrated in the studies above, the relative
usefulness of AI in the college composition classroom is large-
ly dependent upon the design of the assignment and the teacher’s
pedagogical and technological competences.
Students benet from being exposed to the tools, especially at the
formative stages of writing, but need support and interventions
from peers and instructors in the later stages. What is absent from
the previous literature is how prevalent the technology will be-
come and the expectation of basic use of ML and AI commonplace
in the workforce will be as common as the Microsoft Oce Suite
today. relates that some faculty have already begun embracing AI
and reimagining how to teach using the new tool in order to ensure
students have what will be required for future workplaces [16].
On the other hand, adoption will be slow; for as an Inside Higher
Ed poll (2022) recently demonstrated, all higher education respon-
dents stated that students submitting essays completely composed
by AI are behaving unethically.
The same respondents, however, agreed that there is a “gray area”
and that a level of use of AI tools is acceptable. As such, D’Agosti-
no argues that while most faculty will continue to defend tradition-
al methods of teaching English Composition and attempt to iden-
tify the dierence between cheating and assisting student writing
with AI, this ignores the realities of education. These new tools can
assist students who do not consider themselves writers, as well as
those underrepresented populations that struggle to nd their place
in literature. Overcoming writer’s block is one such area of forma-
tive assistance that AI has already proven eective. Other students
J Robot Auto Res , 2023
Volume 4 | Issue 1 | 333J Robot Auto Res , 2023
who are more procient will be able to use AI to further hone their
abilities, and studies speak to this [15].
Methodology
The mixed-methods study included data from surveys collected
from students. The sample was collected from Lindenwood Uni-
versity, a private, four-year, liberal arts institution in the suburban
ring of St. Louis, Missouri. Participants included 21 undergraduate
students from all colleges across the University enrolled in English
Composition I in Fall of 2022. The purpose of the study was to
investigate the usefulness and limitations of AI essay generators
for teaching and learning in college composition classes. Students
were instructed to complete a rhetorical analysis and AI essay
writing collaborative project using any of six AI tools, including
EssayAiLab, Essay Soft, Good ai, My Assignment Help, OpenAI,
Paper Typer, Study crumb, Article Generator, and AI Article Writ-
er 3.0. Students were also encouraged to nd other examples. Ex-
ercises to introduce students to the functionality of these tools took
place for the rst three weeks, three days a week.
Early prompts were notably open-ended and intended to show-
case potential future uses in the class. Students tested the AI es-
say generators for two weeks and then completed a reective
essay on their perceptions of the experience. This project uti-
lized a mixed-methods study design which included qualitative
(open-ended comments) and thematic (quantitative) results from
two online surveys. Students were contacted either through the
University course management system or were emailed with links
to online surveys. The rst survey was administered in the second
week of the term, prior to assignments using AI. The pre-survey
collected data on student demographics, comfort with technology,
dedication to self-improvement, experience using AI in general
and essay generators in particular. The second survey was admin-
istered in the eighth week, after the two AI essay assignments had
been completed.
The post-survey collected the same student demographic data, as
well as the experiences using the AI tools and asked students to
rank the usefulness for learning writing. Participants were asked
to indicate via a 1-10 Likert scale their perceptions of AI tech-
nology in general and the new tool in particular. Students were
asked an open-ended question regarding their experience. All data
was collected using Qualtrics to ensure privacy and anonymity of
responses. These results were sorted based on demographics, and
data were exported for the survey system. Descriptive statistics
were calculated and used for comparisons between groups.
Results
Of the 21 student respondents, all were between 18-24 years of
age; 95.24% were First Year students; 61.9% identied as female
and 38.10% male; 73.08% identied as White, 15.38% Black or
African American, 7.68% Asian, and 3.85% American Indian or
Alaskan Native; 95.24% were non-international students; 71.43%
identied as student athletes on the University campus; 9.52%
identied as rst-generation college students; 90.48% stated they
lived on campus as a residential student; and 85.71% claimed to
primarily take classes in a face-to-face modality. All students were
enrolled in the course as a General Education requirement to fulll
their rst-year college writing requirement.
The pre-assignment survey asked students to rate their openness
to self-improvement as an indicator of proper use of an AI tool.
Of respondents, 90.48% selected moderately or very open to im-
provement of their writing. Students were then asked to estimate
how many hours in a given week they spend trying to improve
their performance on skills that matter for them personally- 4.76%
stated more than 10 hours, 28.57% 7-9 hours, 14.29% 4-6 hours,
28.57% 2-3 hours, and 23.81% more than 1 hour. Next, students
were asked a series of questions about their level of comfort and
experience with technology in general and AI in particular. The
majority of students (57.14%) claimed to be somewhat comfort-
able with technology in general, but 33.33% selected somewhat to
extremely uncomfortable. 61.9% also claimed to have never used
an AI application to help improve their writing, 23.81% had, and
14.29% were unsure. Finally, participants were asked to rank in or-
der the ways AI essay applications may improve writing followed
by a free response to clarify. Students then ranked the following in
order of importance from most to least (Figure 1):
1. Help organize existing ideas
2. Assist in creating new ideas
3. Ensure proper syntax is used
4. Check for correct grammar
5. Suggest creative solutions
6. Provide a scientic approach to writing
Figure 1: Student ranking of AI usefulness in improving writing
before use
The following free responses claried the selections and highlight-
ed many of the suggested uses outlined in the Literature Review.
For instance, one student noted how they did not identify as a writ-
er and the tools could help with creativity: “I am not super creative
so hopefully it could help with that.” Another reiterated the senti-
ment and stated that: “these tools can be helpful because it could
help in areas i struggle with, it can give me ideas.” Nearly half
of the responses highlighted the ability of AI tools to assist with
the mechanical aspects of writing including grammar and syntax:
“They would be helpful because we have a tool that will always
Volume 4 | Issue 1 | 334J Robot Auto Res , 2023
be handy to help us write and caught mistakes maybe you or any-
one else couldn't see.” Likewise, another student wrote: “They can
help x mistakes that you don't see yourself like grammer or spell-
ing issues.”
Many students also touched on the ethical use of AI and highlight-
ed the potential benets: “If used properly the tools can ensure
more creativity as the user would become less focused on details
of proper writing and more so on the information they are putting
down.” Directly addressing what the last student alluded to were
a number of others who questioned, “If we rely on a computer to
write our essays, how do we improve their writing?” And another
noted the potential for academic dishonesty and abuse, claiming,
“It could also be a negative thing if people use it as a way of get-
ting an essay wrote for them.”
The post-assignment survey asked students to reect on their ex-
perience with AI and to reassess their previous assumptions. Stu-
dents were rst asked if they preferred having the AI essay gener-
ator exercises as part of the class. 50% stated that they did, 20%
did not and 30% were unsure. In keeping with the feedback from
the pre-survey, none of the students claimed that the tools helped
improve their writing. In fact, 60% claimed resoundingly that it
did not, and 40% responded as being unsure. The ambivalence
over the usefulness of such AI tools surfaced again when students
were asked if they could see themselves using something like them
again in the future for writing. 60% stated “maybe” with 30% stat-
ing “yes” and only 10% selecting “no.” As suspected, as with the
studies in the Literature Review, the researchers had assumed that
students would start the class with the belief that AI could, in fact,
write complete and nished essays and would end the class being
more dubious.
The results were borne out with 30% believing beforehand that
AI could eectively produce essays, 20% unsure, and 50% stating
that it could not. After the assignments using AI in class, 70% stat-
ed that AI could not produce quality full-length essays with 20%
unsure and only 10% believing that it could. Students were then
asked to re-rank the ways in which AI may be useful in the writing
process again with the same options (Figure 2).
The results were as follows:
1. Assist in creating new ideas
2. Ensure proper syntax is used
3. Suggest creative solutions
4. Check for correct grammar
5. Provide a scientic approach to writing
Figure 2: Student ranking of AI usefulness in improving writing after use
Notably, no one selected “Help organize existing ideas,” which
was the rst selection in the pre-survey of the ranking. In com-
paring the responses from the pre-and post-survey results, before
using the tools, students ranked creativity and organizing ideas the
highest. After using the tools, students more emphatically in their
ranking (the last three in the post-survey were tied and one was not
even selected) relayed the belief that the tools could really only
check for mechanical aspects of writing and help start the writing
process if writer’s block occurred. In the written artifacts produced
and reective essays, students reiterated their ndings of using AI
to assist in the writing process and not replace or usurp that pro-
cess.
Roughly half of the students noted how the tools could be used to
assist in starting the writing process. As one student noted, “To be
completely honest, I have been fairly impressed with some of the
pieces of writing that have been produced from the AI tools that
I have ddled with. These tools could for sure be something that
people can look to, when running into a kind of mental block with
writing. It happens often to many where it seems impossible to nd
topics to write about especially when given a specic prompt and
not a whole lot of creativity.” Another student provided a concrete
example of this during their writing process: “The positive thing
about using AI to help write papers is getting started on what top-
ics I can do. For my other paper, I used an AI to help with writing
prompts for an essay. It helped by showing me what I can write
on the topic, and I can expand on what the AI gives me. So, if the
topic were about why coral reefs are dying, it would help give me
ideas on what to write on.”
Volume 4 | Issue 1 | 335J Robot Auto Res , 2023
While many students were quick to note how writing with ma-
chine-in-loop processes worked well in formative stages, ultimate-
ly, most noted the limitations of AI in writing complete essays or
even ideas. One student noted that there is not a substantial amount
of material generated for a full essay: “The rst issue is how it does
not produce enough material to dene its answer as an essay. This
limits the AI’s usability as an essay writing tool as it gives small
and concise answers. This would make it more like a brainstorm-
ing tool or a chat bot, but it can’t be dened as an essay writer. Of
course as a brainstorming tool the AI’s human-like answers give
a wonderful source to assist writers in getting feedback.” Another
noted the limitations on following through with reasoning on giv-
en topics: “One of aws many of the AIs presented was the lack
of specics.
For example, they would be able to provide a statement saying
bullying has a negative eect on one’s mental health, maybe be-
cause that is accepted by society, but wasn’t able to explain in what
ways or provide examples.” The same frustrations were expressed
by another student who noted the disjointed nature of generative
sections: “These tools do not even truly write most of the work
but nd the work of authors and hand them to the user. Paragraphs
felt as though they were just ripped out of an article rather than
incorporated into a proper opening to an essay. In a way, these
websites are gloried google searches, giving its user the ability
to browse the work of others rather than witnessing the intellect of
an AI.” In all, students in the study liked working with the AI tools
and claimed that their use would be preferable in the future, but
only in specic aspects of the writing process and not in writing a
coherent, and well-written or crafted essay.
AI as Subject and Object
Instructor reection on the use of AI tools for the class includ-
ed recommendations for a measured approach in implementing
these tools into assignments. While faculty may consider students
to be highly procient in technology, students in this study did
not demonstrate self-suciency in the use of the tools provided.
For instance, even though students were allowed to nd their own
AI tools, few actually did. The instructor had expected students to
immediately take to the use of the NLP with prompts but after the
rst day decided on a more targeted approach in guiding students
through demonstrations of how to use the technology properly.
Testing the potential of the writing applications for several weeks
led to the realization that the assignment was too open-ended and
was, therefore, rened and structured with more meta-writing in
the next phase of the class in order to have students write about
writing itself. One of the limitations of composition classes is that
there is no subject matter inherent to the course as the focus is
on writing. Instead, composition readers assigned to classes of-
ten focus on political issues of the day, supplemented with some
instruction in technical topics such as improving usage. As such,
rst-year students are not exposed to primary source research and
read secondary sources as context for developing arguments. What
developed in the next phase of this class is a potential model for
others rethinking the composition class. When students began us-
ing the AI tools, the instructor predicted that the grammatical and
technical functionality would be most useful for students, but that
was not the case. Students did not focus on the act/craft of writing
or the granular level of processes but instead gravitated towards
the macro-level of the tool to be used and began using prompts on
how AI would impact society and related ethical considerations.
Instead of accepting articles on given topics as fact, students would
use the AI to test whether what they were reading about AI was ac-
curate; the process became self-reective. For instance, after nd-
ing a statistic that AI would automate jobs and displace workers,
students began asking AI what the future of work would entail and
how AI would impact it. Next, students asked the AI if AI tools are
a “good idea,” and attempted to discover if the algorithms had any
inherent moral or ethical biases by attempting to prompt the AI
into the position that “racism was a good thing.” Students discov-
ered that AI reected contemporary values familiar to the students
as drawn from their own communities. The process of how ML
and NLP actually functioned became a primary area of interest
as students began to notice how AI think about things. As they
were intrigued by how AI processed information, many students
attempted to see if they could make the algorithm malfunction or
do or say “silly” things.
Writing as Iterative and Collaborative
As demonstrated in the data collected from this case study, the
use of AI certainly falls into a current “gray area” between en-
tirely student-produced and augmented content. The greatest mis-
conception about the use of such tools thus far seems to be in an
“all or nothing” mentality: either students write in isolation from
emerging technologies, or they are guilty of academic dishones-
ty. But as the study from North Carolina State demonstrates, the
situation is not as clear cut. 87% of students who were accused
of “cheating” by integrating AI-generated content into a nal es-
say reported that actually doing so was much more complicated
than writing the paper themselves [16]. The claims of students are
borne out in other studies as well, which note that for successful
use of AI essay generating tools, pre-planned interventions on the
part of the instructor must be included, as well as communication
and collaboration skills reinforced in groupwork [13-15].
In other words, students found the tool helpful in the learning pro-
cess as long as this use supplemented traditional feedback instead
of supplanting it. As noted by the instructor in this study, the most
common response from students was that in order to use the tools
successfully as a writer, close contact between the student and
tool needed to be maintained during the process. As one student
wrote of the collaborative exchange between human and AI: The
AI-written paper wasn’t completely smooth, but that’s probably
because a large portion of it was generated. The issue could prob-
ably be solved with a better ratio between AI and human written
parts. With a ratio leaning more towards the human written side,
it brings not only a more coherent paper, but a more ethical one.
Using an AI to type some of a paper should not be seen as com-
Volume 4 | Issue 1 | 336J Robot Auto Res , 2023
pletely immoral. After all, it’s not so dierent from asking a peer
for writing advice, and using some of their suggestions. It does be-
come an issue if one uses AI to type an entire paper, and then tries
to pass it o as if it was their own. The AI generation was used
as the majority of the mock paper to see how it would turn out,
not to claim the entire paper as an original work... In conclusion,
with their many strengths yet apparent weaknesses, AI tools like
OpenAI are good tools for strengthening papers or bridging the
gaps. It would be more trouble to constantly nd transition words
than just writing a paper on your own. AI should be used as a sort
of guide to follow on paper typing, or using generated parts as a
springboard for adding text when needed. Given that the process
is iterative, the AI could generate content, but students still need-
ed to check the suggested material against the overall essay and
decide whether to use it or not. The results suggest that writing
with computational assistance currently must be collaborative to
be successful and demands active engagement on the part of the
human participant [16].
Conclusion
While technical instruction using AI for college composition is
further out, these tools can be adopted today to prompt interesting
conversations or generate ideas on topics. This case study demon-
strated the self-reective nature of student engagement with tech-
nology, which led to discussions about social ethics, bias and tech-
nology in society. One of the issues with these services is that AI
essay generating applications are marketed as ways to replace, and
not improve, the writing process, compounding misunderstand-
ings about their actual use and capacities at present. The fears and
anxieties over such technologies making college composition out-
moded are unfounded.
This is not to say that there are not paid services that are able to
eectively generate college-level papers that can receive high
marks and are undetectable by plagiarism software. However, the
best paper generating tools are over $100 a month, which places
them out of reach for most college students. At the same time,
paid services such as PapersOwl have existed for decades over
the internet, allowing students to pay others to write their papers
and pass them o as their own for around $10 a page. Therefore,
the concern over the latest method to avoid writing your own term
papers is just one in a long line.
The eects on college students in the short term will be two sides
of the same coin. Students will be able to focus on studies and
spend less time writing essays and focus on the material to be
covered. On the other hand, aspects of the writing process will be
replaced with automated processes [5]. In the end, the teaching of
college composition will still follow some traditional best prac-
tices, including scaolding assignments, instructor and peer feed-
back during the drafting process, and balancing new approaches
with other forms of knowledge gathering and transference. But
AI will be benecial in other ways and support pedagogical areas
composition instructors have found challenging. For instance, the
social context of writing determines writing conventions, and yet
most rst-year writing classes are still focused on the development
of transferable writing skills across disciplines.
As writing curricula need to teach strategies of how to respond to
contextual elements that may inhibit situations of writing [17]. also
notes how important successful academic collaborations are in the
writing process [18]. point out the importance of writing in con-
text and how important prior knowledge is in developing students’
writing abilities. In three models, the authors present how students
respond to and use new knowledge, including assemblage, remix,
and critical incident [19]. With the debate over a general approach
to college composition and one grounded in a discipline-specic
framework, AI has the ability to address and augment many of
the recommended writing pedagogies out there today and move
students beyond their initial perceived self-limitations and open up
new possibilities for writers in the future. On behalf of all authors,
the corresponding author states that there is no conict of interest.
Data is available upon request for the study.
References
1. Warschauer, M., & Ware, P. (2006). Automated writing eval-
uation: Dening the classroom research agenda. Language
teaching research, 10(2), 157-180.
2. Machicao, J. C. (2019). Higher education challenge charac-
terization to implement automated essay scoring model for
universities with a current traditional learning evaluation sys-
tem. In Information Technology and Systems: Proceedings of
ICITS 2019 (pp. 835-844). Springer International Publishing.
3. Abd-Elaal, E. S., Gamage, S. H., & Mills, J. E. (2019, Jan-
uary). Articial intelligence is a tool for cheating academic
integrity. In 30th Annual Conference for the Australasian As-
sociation for Engineering Education (AAEE 2019): Educators
becoming agents of change: Innovate, integrate, motivate (pp.
397-403). Brisbane, Queensland: Engineers Australia.
4. Abd-Elaal, E. S., Gamage, S. H., & Mills, J. E. (2022). Assist-
ing academics to identify computer generated writing. Euro-
pean Journal of Engineering Education, 47(5), 725-745.
5. Schatten, J. (2022). Will Articial Intelligence Kill College
Writing? The Chronicle of Higher Education. September 14,
2022. Accessed.
6. Fyfe, P. (2022). How to cheat on your nal paper: Assigning
AI for student writing. AI & SOCIETY, 1-11.
7. Corbett, E. P. (1987). Teaching composition: Where we've
been and where we're going. College Composition and Com-
munication, 444-452.
8. Harris, J. (2012). A teaching subject: Composition since 1966.
University Press of Colorado.
9. Hassel, H., & Giordano, J. B. (2013). Occupy writing studies:
Rethinking college composition for the needs of the teaching
majority. College Composition and Communication, 117-139.
10. Hesse, D. (2017). We know what works in teaching composi-
tion. The Chronicle of Higher Education, 3.
11. Baker, S., Formo, D. M., Headley, C., & Mecucci Springer, L.
M. (2021). Transforming the feedback paradigm: A qualita-
tive study examining a student-centered, question-based ped-
Copyright: ©2023: James Hutson. is is an open-access article distributed
under the terms of the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium,
provided the original author and source are credited.
Volume 4 | Issue 1 | 337J Robot Auto Res , 2023
agogy in college composition and literature courses. Teaching
English in the Two-Year College, 48(4), 387-412.
12. Shafer, G. (2022). Listening to Our Students while Teaching
Composition in the Community College. Community College
Journal of Research and Practice, 46(8), 602-608.
13. Chen, C. F. E., & Cheng, W. Y. E. C. (2008). Beyond the de-
sign of automated writing evaluation: Pedagogical practices
and perceived learning eectiveness in EFL writing classes.
14. Sumakul, D. T. Y., Hamied, F. A., & Sukyadi, D. (2022, Feb-
ruary). Students’ Perceptions of the Use of AI in a Writing
Class. In 67th TEFLIN International Virtual Conference &
the 9th ICOELT 2021 (TEFLIN ICOELT 2021) (pp. 52-57).
Atlantis Press.
15. Plate, D., & Hutson, J. (2022). Augmented Creativity: Lever-
aging Natural Language Processing for Creative Writing. Art
and Design Review.
16. D’Agostino, S. (2022). Machines Can Craft Essays. How
Should Writing Be Taught Now? Inside Higher Ed. October
26, 2022. Accessed
17. Beaufort, A. (2008). College writing and beyond: A new
framework for university writing instruction. University Press
of Colorado.
18. Myatt, A. J., & Gaillet, L. L. (2017). Writing program and
writing center collaborations. Palgrave Macmillan.
19. Yancey, K., Robertson, L., & Taczak, K. (2014). Writing
across contexts: Transfer, composition, and sites of writing.
University Press of Colorado.
20. Buamanti, S., David, D., & Morris, R. (2006). Imposed
upon: Using authentic assessment of critical thinking at a
community college. Teaching English in the Two Year Col-
lege, 34(1), 70.
21. Carter, T. M. (2013). Use what you have: authentic assess-
ment of in‐class activities. Reference Services Review, 41(1),
49-61.
22. Carter, S., & Nielsen, M. (2017). Using articial intelligence
to augment human intelligence. Distill, 2(12), e9.
23. Feng, X., Liu, M., Liu, J., Qin, B., Sun, Y., & Liu, T. (2018,
July). Topic-to-essay generation with neural networks. In IJ-
CAI (pp. 4078-4084).
24. Francke, E., & Bennett, A. (2019, October). The Potential In-
uence of Articial Intelligence on Plagiarism: A Higher Ed-
ucation Perspective. In European Conference on the Impact
of Articial Intelligence and Robotics (ECIAIR 2019) (pp.
131-140).
25. Gaur, M., Arora, M., Prakash, V., Kumar, Y., Gupta, K., &
Nagrath, P. (2022). Analyzing Natural Language Essay Gen-
erator Models Using Long Short-Term Memory Neural Net-
works. In International Conference on Innovative Computing
and Communications: Proceedings of ICICC 2021, Volume 3
(pp. 233-248). Springer Singapore.
26. Sharples, M. (2022). Automated essay writing: an AIED opin-
ion. International Journal of Articial Intelligence in Educa-
tion, 32(4), 1119-1126.
27. Whitlock, B., & Nanavati, J. (2013). A systematic approach
to performative and authentic assessment. Reference Services
Review, 41(1), 32-48.
