Information Systems Education Journal (ISEDJ) 22 (3)

ISSN: 1545-679X July 2024

Ethics-Driven Education: Integrating AI

Responsibly for Academic Excellence

Chukwuemeka Ihekweazu
emekaihs@[Link]

Bing Zhou
bxz003@[Link]

Computer Science Department

Sam Houston State University
Huntsville, TX 77340, USA

Elizabeth Adepeju Adelowo

eaadelowo@[Link]
Educational Leadership and Higher Education Department
St. Cloud State University
St Cloud, MN 56301, USA

Abstract
This study delves into the opportunities and challenges associated with the deployment of AI tools in
the education sector. It systematically explores the potential benefits and risks inherent in utilizing these
tools while specifically addressing the complexities of identifying and preventing academic dishonesty.
Recognizing the ethical dimensions, the paper further outlines strategies that educational institutions
can adopt to ensure the ethical and responsible use of AI tools. Emphasizing a proactive stance, the
paper suggests that by implementing these strategies, schools can harness the benefits of AI tools while
mitigating the risks associated with potential misuse. As the adoption of AI tools in education continues
to expand, all stakeholders must stay abreast of the latest developments in the field. This knowledge
equips educators to navigate the opportunities and challenges posed by AI tools, fostering a learning
environment that is both secure and conducive to empowering students to realize their full potential.

Keywords: Artificial Intelligence, Academia, Detectors, Machine Learning, Natural Language Processing

Recommended Citation: Ihekweazu, C., Zhou, B., Adelowo, E.A., (2024).Ethics-Driven Education:
Integrating AI Responsibly for Academic Excellence. Information Systems Education Journal, 22(3),
pp.36-46. [Link]

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 36

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

Ethics-Driven Education:
Integrating AI Responsibly for Academic Excellence
Chukwuemeka Ihekweazu, Bing Zhou and Elizabeth Adepeju Adelowo

1. INTRODUCTION
O'neil (2003) asserts that there is undeniable
Artificial intelligence (AI) holds the promise of evidence that modern technology has enhanced
reshaping the landscape of education across students' ability to engage in academic
multiple dimensions, including personalized dishonesty. While there may not be a definitive
learning and automated evaluation. Murphy solution to addressing breaches of academic
(2019) underscores AI's potential in aiding integrity, especially in an online environment,
educators to pinpoint students grappling with there are strategies that must be implemented to
specific concepts, delivering customized instill a sense of responsibility in students. Similar
assistance tailored to their needs. Additionally, to how PowerPoint transformed classrooms more
AI's ability to scrutinize extensive datasets than two decades ago and how real-time media is
facilitates the identification of patterns and currently reshaping them, AI is poised to
forecasting outcomes, empowering educators to revolutionize the learning experience in the
base decisions on well-founded, data-driven coming years (Bain, 2015).
insights. The integration of AI in education
extends to automating administrative tasks like Schiff (2022) suggests that ethical considerations
grading and lesson planning, allowing teachers to must be at the forefront of AI implementation in
allocate more time to the instructional aspect. education. This includes ensuring that AI systems
Nevertheless, the infusion of AI into education are transparent, explainable, and unbiased.
introduces complexities, particularly in combating Furthermore, Striepe et al. (2023) argue that
academic dishonesty. Cotton et al. (2023) draw educators should prioritize teaching students
attention to the escalating challenges posed by about the ethical implications of AI and how to
online education, making it more convenient for use it responsibly. By promoting ethical AI
students to engage in various forms of cheating. practices, educators can help ensure that the use
While AI can play a role in detecting plagiarism, it of AI in education is consistent with academic
concurrently opens doors to novel possibilities for integrity. It is worth noting that academic
academic misconduct. For instance, students may dishonesty is not solely the result of AI.
exploit AI-powered language generators for essay Personality traits such as low conscientiousness
creation or resort to chatbots to manipulate and high neuroticism are associated with
online exams. Despite these hurdles, certain academic dishonesty. Therefore, promoting
researchers assert that AI can contribute academic integrity requires a multifaceted
positively to upholding academic integrity within approach that considers individual characteristics
educational environments. as well as the technological tools and systems
used in education.
Haye and Kyobe (2020) found that using Turnitin,
an AI-powered plagiarism detection tool, can help As the use of AI in academia becomes more
students improve their writing skills and reduce widespread, there is a need to verify that
instances of plagiarism. By providing students academic work is of human origin. There are
with feedback on their writing and identifying several AI detection tools that report a confidence
areas for improvement, Turnitin can encourage level that a given textual input is of human or AI
students to engage in ethical writing practices. To origin. In this paper, we evaluate the accuracy of
effectively use AI in education while promoting these tools and comment on their suitability for
academic integrity, educators and administrators detecting academic dishonesty. We use scenario-
need to be aware of the potential risks and based testing to design prompts for the two
benefits of AI, as well as best practices for leading AI chatbots, ChatGPT and Google Bard.
integrating AI into their teaching practices. Chan When their generated output is fed directly into
and Tsi (2023) suggest that educators should the AI detection tools, it is reliably identified as
focus on using AI to augment, rather than being of AI origin, suggesting that the tools are
replace, traditional teaching methods. They also indeed suitable for detecting academic
emphasize the importance of transparency and dishonesty. However, when the output is further
fairness in AI-based assessment systems. processed using AI paraphrasers, all of the

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 37

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

detection tools examined fail to identify the origin of ensuring that the use of AI in digital forensics
as AI and instead attribute it to humans. Thus, it is done ethically and responsibly, with
is impossible to accurately detect AI-generated appropriate safeguards and oversight.
content when it has also been paraphrased by AI.
3. LITERATURE REVIEW
2. ETHICAL CONSIDERATIONS IN THE USE
OF AI IN ACADEMIA According to Rutner & Scott (2022), technology
offers numerous advantages to society,
The integration of artificial intelligence (AI) in particularly college students, by granting
scientific practices introduces a myriad of ethical convenient access to diverse resources. However,
considerations that demand careful examination. with this accessibility comes the allure of utilizing
These concerns primarily come from the information in ways that bypass the traditional
utilization of AI detection tools intertwined with learning process.
elements of digital forensics involving the scrutiny
of digital traces. The ethical considerations Educational institutions have been fighting
encompass the following dimensions: against dishonest behavior since it became a
persistent problem in academia decades ago
Bias and Discrimination: The deployment of AI in (Schiff, 2022). With the proliferation of
digital forensics poses a risk of bias and technology, students have access to a wide range
discrimination, as the algorithms utilized for data of tools and resources that facilitate the dishonest
analysis may inadvertently mirror the biases and behavior of cheating, such as plagiarism detection
prejudices embedded by their developers. This software, online forums for sharing answers, and
gives rise to the potential for unfair treatment of even devices like smartphones that can be used
specific individuals or groups (Narayanan & to access information during exams. Largely,
Reddy, 2019). academic dishonest behavior is exacerbated by
AI-based tools that can generate complete
Privacy and Surveillance: The application of AI in research papers and texts that resemble human
digital forensics raises apprehensions regarding speech, making it more difficult for educators to
privacy and surveillance. While AI enhances the identify and stop such behavior (Sharma &
efficiency of data analysis, it simultaneously Baliyan, 2020). ChatGPT and Bard are AI chatbots
sparks concerns regarding the extent to which based on large language models that are
personal data is subjected to monitoring and expected to have a variety of applications in a
analysis (Narayanan & Reddy, 2019; Koops et al., number of fields.
2021).
Over the recent years, AI technologies have been
Transparency and Explainability: The opacity and implemented in education for use in assessment
lack of explainability in AI systems employed for and instruction. A recently created artificial
digital forensics present challenges. intelligence (AI) model called ChatGPT, which
Understanding the decision-making processes of performs complex cognitive tasks, has gained
these systems becomes intricate, making it popularity among academics in recent months.
challenging to hold individuals or organizations Due to this technology, the potential for academic
accountable for their actions (Goodman & dishonesty among students who utilize ChatGPT
Flaxman, 2017). to generate essays and assignments has been a
subject of concern (Murphy, 2019). According to
Intellectual Property and Copyright: The Goodman and Flaxman (2017), ChatGPT has
incorporation of AI in digital forensics raises been found to pose a risk to the integrity of
intellectual property and copyright infringement submitted essays, particularly in higher education
issues. While AI systems facilitate the analysis of settings where such requirements are common.
extensive data, there exists a risk of unintentional
usage of copyrighted material without proper Google Bard AI is a newly introduced text-based
attribution or permission (Azab et al., 2020; artificial intelligence chatbot similar to ChatGPT.
Sharma & Baliyan, 2020). It uses machine learning and natural language
processing to generate responses in real-time. It
Cybersecurity: The utilization of AI in digital can be useful for completing creative projects,
forensics introduces cybersecurity concerns. AI explaining complicated topics, and gathering
systems may be susceptible to cyberattacks and knowledge from many online sources.
data breaches, posing a threat to the integrity of Additionally, Google Bard AI can provide complex
the analyzed data (Sharma & Baliyan, 2020). answers, such as finding recipes that use your
These considerations underscore the importance current items in your fridge. It transforms Google

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 38

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

from a simple search engine into a powerful is safe, equitable, and considerate of students,
virtual assistant (Azab et al., 2020). The teachers, and all other stakeholders.
effectiveness of any language model, including
ChatGPT and Bard AI, depends on factors such as Koops et al. (2021) note that the use of AI in
the size of the model, the quality of training data, education raises important questions, such as
and the fine-tuning process. Differences in these what should be taught and how, the changing
factors could lead to variations in performance nature of the teacher, and the social and ethical
and accuracy across different tasks and domains. implications of the technology. There are also
Users can interact with the chatbot just as they many difficulties, including issues of access and
would with ChatGPT. It is worth noting that with equity in education. There is also a growing
the ongoing research and development efforts, understanding that the use of AI in education may
it's likely that both ChatGPT and Bard AI will change the fundamental basis of teaching and
continue to evolve, offering even more learning (Almeida & Apar'ıcio, 2020). An alarming
sophisticated capabilities and applications in the trend of ChatGPT being cited as co-authors on
future. scientific articles began to emerge (OpenAI,
2023). A number of journals quickly responded by
Due to the impact of the COVID-19 pandemic on enacting policies prohibiting ChatGPT authorship
society and higher education, there is a renewed on the grounds of plagiarism, errors,
interest in academic dishonesty (Cotton et al., misinformation, and false information, which led
2023). The reasons why students intentionally to preprint papers in which ChatGPT was later
violate standards of academic integrity remain removed as an author.
unknown despite a substantial body of research
on why and how they do so. Social philosophy can In contrast, several journals promote the use of
be used to fully address the broad question of why ChatGPT to improve writing, especially in cases
students engage in unethical academic practices. where English is a second language. According to
However, according to Hayes and Kyobe (2020), Google (2023), ethical writing is an important
their research shows that online academic issue in both education and research.
dishonesty is indeed pervasive, and the factors Unfortunately, cheating in class happens to
that contribute to this behavior include undergraduate and graduate students alike. As a
personality, cognition, and instruction. Chan and result, written essays and articles are subject to
Tsi (2023) assert that in order to address the specific detection procedures, and the majority of
problem of students plagiarizing assignments academic institutions use a variety of
using artificial intelligence tools such as ChatGPT, technologies to combat plagiarism. However,
one must first understand what ChatGPT actually cutting-edge artificial intelligence (AI) offers a
is and the concept of AI. Striepe et al. (2023) new platform for new types of serious academic
suggest strategies that schools can use to ensure misconduct that are difficult to detect and even
the moral and ethical use of these ChatGPT. harder to prove.
These approaches include creating policies and
procedures, offering assistance and support, and 4. RESEARCH QUESTIONS
using a variety of tools to identify and stop
cheating. They conclude that while the use of AI Using two AI chatbots, ChatGPT, and Google
in higher education presents both opportunities Bard, and seven AI detection tools, we seek to
and challenges, universities can successfully answer the following questions:
address these issues by using these tools in a
proactive and moral manner. 1. How accurate are the AI detection tools?

The ethical and responsible use of ChatGPT in Based on our research design, we were aware in
educational contexts is a complex, multifaceted advance that all input was generated by artificial
issue that requires a multidisciplinary, nuanced intelligence.
approach. Recent studies have highlighted the
need for responsible and ethical use of artificial The AI detection tools report a confidence level as
intelligence in education. The studies that have to whether the input is AI-generated or human-
been conducted on this particular topic have generated, which leads to the second research
focused on issues such as privacy, bias, and the question:
potential for AI to widen the digital divide
(Narayanan & Reddy, 2019). When using 2. Can AI tools aid in academic dishonesty?
ChatGPT in educational settings, it is important to
follow responsible and ethical procedures to
ensure that the technology is used in a way that

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 39

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

5. METHODOLOGY AI Chat-Box
AI chatbots are computer programs that use
We employ a scenario-based testing strategy, artificial intelligence and natural language
shown schematically in Figure 1, to assess the processing to simulate human conversations with
accuracy and bias of AI detection tools. We users. They can be used for customer service,
designed appropriate prompts for the two information retrieval, and personal assistance.
prominent AI chatbots, ChatGPT and Google There are two types of chatbots: rule-based and
Bard, to simulate common academic applications machine learning-based. Rule-based chatbots use
of generative AI tools in four distinct scenarios. rules and scripts to respond to user input, while
The four scenarios include: (1) Ask the AI tool to machine-learning chatbots use algorithms to
summarize a longer text and then verify the learn from user interactions. Chatbots positively
accuracy of the output by comparing it to the impact the customer experience, increasing
original text and confirming that it was written satisfaction and loyalty (Almeida & Apar'ıcio,
completely by the AI tool; (2) Ask the AI tool to 2020).
paraphrase a sentence or paragraph from a
specific source and then check the output for ChatGPT: ChatGPT is a large language model
plagiarism by comparing it to the original text and developed by OpenAI based on the Generative
confirming that it was completely written by the Pre-trained Transformers (GPT) architecture. It is
AI tool; (3) Ask the AI tool to provide a designed to generate human-like responses to
personalized answer about its opinion on text-based prompts and can be accessed via the
plagiarism. Sample question; and (4) Ask the AI OpenAI API or online chat platforms (OpenAI,
to confirm that it actually recognizes its own 2023).
work, we will refer to its individual responses to
the scenario. Bard: Google Bard is a Large Language Model
(LLM) chatbot developed by Google AI and
Their responses were either fed directly into powered by the Language Model for Dialogue
seven different AI detection tools or post- Applications (LaMDA) language model. It can
processed by one of three AI paraphrasing tools generate text, translate languages, write creative
before being sent to the detection tool. The AI content, and answer questions in an informative
detection tools report confidence, expressed as a manner. It has the potential to be a powerful tool
percentage, that the input was generated by for communication, creativity, and learning
either a human or an AI. Based on this reported (Google, 2023).
confidence, we provide an assessment of whether
or not the specific tool is suitable for detecting The main difference between ChatGPT and Google
generative AI output. Bard is the LLMs used to build them. The better
the data LLMs are trained on, the better their
ability to generate accurate and readable content.
Bard and ChatGPT are run on different LLMs, each
trained on different datasets: Bard uses PaLM 2
while ChatGPT uses GPT-3.5 and GPT-4, a larger
and more updated version of GPT-3.5. Bard is
free to use, while ChatGPT is free to access GPT-
3.5, but access to GPT-4 costs $20 per month.

While Bard and ChatGPT may be used

interchangeably in most instances, Bard excels at
responding to user concerns and providing more
decisive answers. ChatGPT, on the other hand, is
better at creating long-form material and
responding to user inputs more extensively and
nuancedly. Bard also gives more up-to-date
information due to its up-to-date LLM and access
to the internet. ChatGPT, on the other hand,
struggles to provide accurate information on
events after 2021. However, it is worth noting
Figure 1: Proposed Methodology that ChatGPT and Bard are still under
development, and their capabilities are
continuously expanding.

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 40

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

AI Paraphraser different machine learning (ML) algorithms such

AI paraphrasers are software programs that use as GPT, Robustly Optimized Bidirectional Encoder
AI and NLP techniques to automatically rewrite Representations from Transformers Pretraining
the text in a different but equivalent way. Using Approach (RoBERTa), and Large Language Model
AI-powered paraphrase tools, authors may Meta AI (LLaMa).
quickly generate alternate versions of their text
in order to avoid plagiarism or improve AI Machine Platform
readability. However, it is important to review and Detector Learning
edit the output to ensure that it is suitable for the Algorithm
Zero GPT GPT-4 Web-Based
intended purpose.
GPT Radar GPT-3 Web-Based
Content at
Over the recent years, different AI paraphrasers Scale AI
Semantic analysis Web-Based
have been developed for academic, business, and Content
personal purposes. Some of the popular AI Detection
paraphrasers include: GPT-Zero GPT-3, GPT-2, Web-Based
LLaMA
Writers AI GPT-3 Web-Based
QuillBot: QuillBot is an AI-powered paraphrasing Content
Detector
and writing tool that uses NLP and machine
OpenAI GPT2
learning algorithms to automatically rewrite text. Output
RoBERTa Web-Based
It offers different modes and allows users to Detector
adjust the level of paraphrasing to suit their Writefull GPT-3, GPT-4 Web-Based
needs.
Table 1: AI Detectors Comparison

Grammarly: Grammarly is an AI-powered writing 6. IMPLEMENTATION

tool that uses NLP technology to provide real-time
grammar and spelling checks, punctuation This study employs an inductive research
suggestions, and style recommendations to help methodology to generate the findings from the
users create clear, error-free writing. scenarios. In inductive research methodology,
the researcher starts with data and observations
WordTune: WordTune is an AI-powered writing and then uses data patterns to develop theories
tool that uses NLP and machine learning or generalizations (Goyal & Bengio, 2022). In this
algorithms to suggest alternative phrases for case, the researcher prompts the AI tools
written text to improve clarity, conciseness, and ChatGPT and Google Bard and analyzes the
effectiveness. WordTune is developed by the AI output.
writing technology company, AI21 Labs.
The four scenarios in this study include:
The three paraphrasing tools have a free version
with limited features and a premium version with Scenario 1: Ask the AI tool to summarize a longer
unlocked features. QuillBot and Wordtune can text and then validate the accuracy of the output
both generate high-quality rewrites. However, by comparing it to the original text and confirming
they may not always be correct. However, that it was written completely by the AI tool.
QuillBot offers a full set of writing tools for editing Example question: Can you summarize the main
and improving paraphrased material. For points of this article in a few sentences?
example, you may utilize QuillBot's grammatical "Teaching in the Online Classroom: Surviving and
check to receive rewrite ideas. Additionally, the Thriving in the New Normal" (Education Next,
ability to freeze words and limit synonym usage 2023).
contributes to more desirable outcomes. These
additional capabilities make QuillBot a more Scenario 2: Ask the AI tool to paraphrase a
powerful paraphrasing tool. sentence or paragraph from a specific source and
then check the output for plagiarism by
AI Detection Tool comparing it to the original text and confirming
AI detection tools use artificial intelligence that it was completely written by the AI tool.
techniques to identify patterns, objects, or Example question: Can you give me an article
behaviors in data. Examples include image about cybersecurity and paraphrase it?
recognition, speech recognition, fraud detection,
malware detection, sentiment analysis, and
Scenario 3: Ask the AI tool to provide a
object detection. Here, we will use seven different
personalized answer about its opinion on
AI detectors, listed in Table 1, some of which use
plagiarism. Sample question: Can AI plagiarize?

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 41

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

Scenario 4: Ask the AI to confirm that it actually

recognizes its own work, we will refer to its
individual responses to scenario 1. Example
question: Did you write this?

7. TESTING

Scenario 1
ChatGPT: A Survival Guide for Distance Teaching
is a book that focuses on how a teacher can make
the world of distance learning as much like the
real world of teaching and learning as possible. It
was written by a team of people who have a deep
respect for the relationships that are built
between teachers and students and how
Figure 3: Output From Forbes
important those relationships are to student
learning and development. The book does not try
to sugarcoat the problems inherent in distance
learning, but it does offer some techniques that
teachers can use to make the experience as
positive as possible for both themselves and their
students.

Google Bard: The article is a review of the book

"Teach Like a Champion: A Survival Guide for
Distance Teaching" by Doug Lemov. The book
provides practical strategies and techniques for
instructors who teach at a distance. The author
emphasizes the importance of maintaining high
expectations, building relationships with
students, and engaging them in meaningful Figure 4: Output From CNN
learning activities. The book also includes
examples of successful distance learning Scenario 2
practices and guidance on how to adapt teaching ChatGPT: Paraphrased Summary: This CNN
techniques to an online environment. Overall, the article provides tips and strategies for small
article concludes that the book is a valuable businesses to protect themselves from
resource for teachers facing the challenges of cybersecurity threats. The article notes that small
distance education. businesses are particularly vulnerable to
cyberattacks due to their limited resources and
lack of dedicated cybersecurity teams. The article
recommends steps such as implementing strong
passwords, using multi-factor authentication, and
backing up data regularly to protect against data
breaches and ransomware attacks. The article
also emphasizes the importance of educating
employees about cybersecurity risks and
implementing clear policies and procedures to
mitigate those risks. Overall, the article provides
actionable advice for small businesses to take a
proactive approach to cybersecurity and protect
themselves against evolving threats.

Google Bard: Cybersecurity is the practice of

protecting systems, networks, and data from
unauthorized access, use, disclosure, disruption,
modification, or destruction. A cybersecurity
professional is someone who is responsible for
Figure 2: Output From WSJ protecting an organization's information assets

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 42

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

from cyber threats. unequivocal in affirming AI's potential for

plagiarism. Several avenues through which AI can
Cybersecurity is a critical issue for organizations engage in plagiarism can be delineated, including
of all sizes. In recent years, there has been a automated research, essay generation,
significant increase in the number and translation and paraphrasing, coding
sophistication of cyber-attacks. These attacks can assignments, and cheating on online
have a devastating impact on businesses, costing examinations. It is imperative to recognize that
them millions of dollars in lost revenue, damaged AI tools possess the potential to facilitate
reputations, and legal fees. There are a number academic dishonesty. Nevertheless, the ethical
of things companies can do to protect themselves dimension of AI deployment hinges upon the
from cyber-attacks. These include Implementing motivations and actions of the user.
strong security measures such as firewalls,
intrusion detection systems, and data encryption.
Training employees on cybersecurity best
practices, such as how to recognize phishing
emails and avoid clicking on malicious links.
Conduct regular security audits to identify and
remediate vulnerabilities. Have a plan in place to
respond to a cyberattack. Cybersecurity is an
ongoing challenge, but one that organizations
must take seriously. By implementing strong
security measures and educating employees on
cybersecurity best practices, businesses can help
protect themselves from cyber-attacks. Here are Figure 5: Output From ChatGPT
some additional tips for businesses to improve
their cybersecurity: Use strong passwords and
change them regularly. Keep software up to date.
Be careful about the information you share online.
Use a firewall and anti-virus software. Back up
your data regularly. Stay informed about the
latest cyber threats. By following these tips,
businesses can help protect themselves from
cyber-attacks and keep their data safe.
For Google Bard, they had no links or citations to
suggest further research on their answer.

Scenario 3
In this scenario, we want to evaluate how Figure 6: Output From Google Bard
ChatGPT and Google Bard evaluate the ability of
AI tools to plagiarize, i.e., copy verbatim the work Based on the scenario, it is arguable that the
of others. This question is particularly interesting ethical and judicious utilization of AI tools
for generative AI tools trained exclusively on regarding plagiarism falls squarely upon the
external sources such as book texts, news shoulders of those who employ them. Educators,
articles, and academic papers. The responses of students, and academic institutions are
ChatGPT and Google Bard to the question "Can AI collectively responsible for upholding academic
plagiarize?" are shown in Figures 5 and 6, integrity and ensuring that AI serves legitimate
respectively. In the initial image, the assertion and constructive educational purposes.
regarding AI's capacity for plagiarism lacks direct
confirmation. Scenario 4
In our final scenario tests, we want to determine
However, it appears to suggest that such a whether AI tools have the ability to recognize
perception should be acknowledged if one
perceives AI as capable of plagiarism. The
statement underscores the distinction between
AI, a neutral tool, and the human user, thereby
implying that the moral responsibility for
plagiarism resides with the latter.

Conversely, the second image is notably

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 43

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

their own words or previous responses.

Figure 10: Output From Google Bard Non-

Recognition

Figure 7: Output From ChatGPT Recognition AI DETECTOR ACCURACY

(BEFORE
PARAPHRASING)
CHAT GOOGLE
GPT BARD
Zero GPT 62% AI 100% AI
GPT Radar 77% 76% Human
Human
Content at Scale 77% for 37% AI
AI Content Detection AI
or Human
GPT-Zero 87% AI 100% AI
Writers AI 6% 14% Human
Content Detector Human
OpenAI GPT2
99.98% 99.96% AI
Output Detector AI
Writefull 92% AI 71% AI
Table 2: Before Paraphraser Results
Figure 8: Output From ChatGPT Non-
Recognition However, there are very powerful and
sophisticated AI summarizers and paraphrasers
that can completely transform AI-written texts
into human-like texts with human emotions that
would generate false positive rates, as seen in
Table 3.

Inaccuracy of AI tools can produce false

knowledge, and in scenario two, we discovered
that although AI does a great job of generating
concise and accurate text, it can also be guilty of
falsifying information. As we can see from the
Figure 9: Output From Google Bard results pages in the second scenario, the links
Recognition provided were not accurate and did not produce
results that justified the accuracy and
8. RESULTS authenticity of the AI tools.

Certain AI detectors have demonstrated their In scenario three, we can see that while ChatGPT
purported accuracy and dependability, while does not explicitly say that it plagiarizes, it does
others have not. In our first scenario, we see that admit that the tool can be modified to plagiarize.
some of the popular AI detectors failed to fully Subsequently, Google's Bard admits that AI as an
distinguish between AI-written text and human- entity can indeed plagiarize, and it can do so by
written text, as shown in Table 2. This could be a producing a body of information without
problem in academia if we cannot fully rely on AI attribution or by paraphrasing a source in a way
detectors to justify the accuracy of their results
effectively.

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 44

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

that it can still be recognized. reveal that absent additional paraphrasing, the
detection process performs admirably across the
AI seven tools under consideration. Applying a 50%
DETECTOR
ACCURACY accuracy threshold, indicating a likelihood that
(AFTER the input is AI-generated rather than human, all
seven AI detection tools successfully identify the
PARAPHRAS AI origin of the input. However, a noteworthy
divergence emerges when ChatGPT and Google
ING)
Bard outputs undergo post-processing by an AI
QUILLB WORD- GRAM
paraphrasing tool. In this scenario, none of the
OT TUNE MARL
seven AI detection tools can ascertain the AI-
Y
generated nature of the input. Consequently, the
Zero GPT 100% 100% 100% amalgamation of generative AI with AI
Human Human Huma paraphrasers renders existing detection tools
n virtually incapable of discerning non-human
GPT Radar 100% 100% 100% origin.
Human Human Huma
n The pervasive issue of academic dishonesty poses
Content at a significant challenge, eroding the integrity of
Scale 100% 100% 100%
educational pursuits, undermining authentic
AI Human Human Huma
scholarly endeavors, and casting a shadow on the
Conte n
reputation of academic institutions. Educational
nt establishments have implemented robust
Detect measures to combat such misconduct,
ion
incorporating plagiarism detection software,
GPT-Zero 100% 100% 100% online examination monitoring, and establishing
Human Human Huma honor codes, and penalties ranging from failing
n grades to potential expulsion. While the
Writers AI 100% 100% 100% enforcement of rules and penalties is crucial, an
Content
Detector Human Human Huma equally vital aspect involves fostering a culture
n that champions learning and ethical conduct. As
OpenAI 100% 100% 100% key influencers, educators play a pivotal role in
GPT2 Human Human Huma instilling principles of honesty, proper citation,
Output and genuine commitment to the learning process
Detector n
in their students.
Writefull 100% 100% 100%
Human Human Huma In conclusion, our study unveils the potential of
n AI in detecting and preventing academic
Table 3: After Paraphraser Results dishonesty while underscoring pertinent ethical
concerns that necessitate attention. The paper
Another interesting discovery we made in our scrutinizes the current state of AI technology
research in our fourth scenario is the ability of AI applied to address academic dishonesty,
to recognize its own work. Using the output from encompassing plagiarism, cheating, and
our first scenario, we ran the two tests on misinformation. Despite highlighting the
ChatGPT and Google Bard in Figures 7-10 and advantages of AI, such as its potential to address
asked if they recognized the texts, ChatGPT academic dishonesty, the study accentuates the
confirmed its own text but did not recognize imperative of a balanced approach, considering
Google's Bard. Alternatively, Google's Bard did potential drawbacks such as privacy
not recognize its own words and did not recognize infringement, false positives, and the risks
ChatGPT's words. associated with excessive reliance on technology.
The overarching message is that while AI holds
9. CONCLUSION immense promise in preserving academic
integrity, its application demands responsible and
Utilizing a scenario-based approach, we employ a ethical usage to ensure fairness and integrity.
methodology to assess the efficacy of AI detection Acknowledging our endeavors, we advocate for
tools. An AI systematically generates all inputs to further research, particularly in areas like AI bias
these tools, and in specific test configurations, the and transparency.
resulting output undergoes paraphrasing using
additional AI paraphrasing tools. Our findings

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 45

[Link] [Link]
Information Systems Education Journal (ISEDJ) 22 (3)
ISSN: 1545-679X July 2024

10. REFERENCES (ICTAS) (pp. 1-6). IEEE.

[Link]
33977
Almeida, T., & Apar´ıcio, M. (2020). The impact
of chatbots on customer experience: A study Koops, B.-J., Newell, B. C., Timan, T., &
in the retail industry. Journal of Retailing and Skorva´nek, I. (2021). Ethical considerations
Consumer Services, 55, 102103. of artificial intelligence in surveillance. AI and
Ethics, 1-19.
Azab, M. M., Elshazly, H. M., & Shaban, H.
[Link]
(2020). Intellectual property protection and
ss2/4
cybersecurity challenges in AI-based
applications. IEEE Access, 8, 224524- Murphy, R. F. (2019). Artificial Intelligence
224540. Applications to Support K-12 Teachers and
Teaching: A Review of Promising
Bain, L. Z. (2015). How students use technology
Applications, Opportunities, and Challenges.
to cheat and what faculty can do about it.
Perspective. RAND Corporation.
Information Systems Education Journal,
[Link]
13(5), 92. [Link]
Narayanan, A., & Reddy, R. (2019). The ethics of
Chan, C. K. Y., & Tsi, L. H. (2023). The AI
artificial intelligence. Communications of the
Revolution in Education: Will AI Replace or
ACM, 62(11), 54-59.
Assist Teachers in Higher Education? ArXiv
[Link]
preprint ArXiv:
[Link] O'neil, T. D. (2003). Technology and Academic
Integrity, Cheating Goes Cyber. Information
Cotton, D. R., Cotton, P. A., & Shipway, J. R.
Systems Education Journal, 1(3), 3-9.
(2023). Chatting and cheating: Ensuring
[Link]
academic integrity in the era of
ChatGPT. Innovations in Education and OpenAI. (2023). About OpenAI. Retrieved May 1,
Teaching International, 1-12. 2023, from [Link]
[Link]
Rutner, S. M., & Scott, R. A. (2022). Use of
90148
Artificial Intelligence to Grade Student
Education Next. (2023). A survival guide for Discussion Boards: An Exploratory Study.
distance teaching - education next. Retrieved Information Systems Education Journal,
May 5, 2023, from 20(4), 4-18.
[Link] [Link]
guide-for-distance-teaching-book-review- pdf
lemov/
Schiff, D. (2022). Education for AI, not AI for
Goodman, B., & Flaxman, S. (2017). European Education: the role of education and ethics in
Union regulations on algorithmic decision- national AI policy strategies. International
making and a "right to explanation". AI Journal of Artificial Intelligence in
magazine, 38(3), 50-57. Education, 32(3), 527-563.
[Link] [Link]
2
Google. (2023). Bard. Retrieved May 1, 2023,
from [Link] Sharma, V., & Baliyan, S. (2020). The
cybersecurity challenge of artificial
Goyal, A., & Bengio, Y. (2022). Inductive biases
intelligence: A review. Journal of
for deep learning of higher-level cognition.
Cybersecurity, 6(1), tyaa003.
Proceedings of the Royal Society A,
[Link]
478(2266), 20210068.
[Link] Striepe, M., Thomson, S., & Sefcik, L. (2023).
Understanding academic integrity education:
Hayes, D., & Kyobe, M. (2020, March). The
case studies from two Australian
adoption of automation in cyber forensics.
universities. Journal of Academic
In 2020 Conference on Information
Ethics, 21(1)
Communications Technology and Society
[Link]

©2024 ISCAP (Information Systems and Computing Academic Professionals) Page 46

[Link] [Link]