📊 statistics

Plagiarism or Productivity? Students Moral Disengagement and Behavioral Intentions to Use ChatGPT in Academic Writing

This study of 418 Filipino college students reveals that moral disengagement mechanisms, particularly the attribution of blame, significantly shape attitudes and perceived control, thereby driving intentions to use ChatGPT in academic writing and highlighting the urgent need for clear institutional policies and ethical guidance.

Original authors: John Paul P. Miranda, Rhiziel P. Manalese, Mark Anthony A. Castro, Renen Paul M. Viado, Vernon Grace M. Maniago, Rudante M. Galapon, Jovita G. Rivera, Amado B. Martinez Jr

Published 2026-03-23

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: John Paul P. Miranda, Rhiziel P. Manalese, Mark Anthony A. Castro, Renen Paul M. Viado, Vernon Grace M. Maniago, Rudante M. Galapon, Jovita G. Rivera, Amado B. Martinez Jr

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a group of students sitting in a library, staring at a blank screen, trying to write a big essay. They have a powerful new tool: ChatGPT, a super-smart AI assistant. But there's a catch: their school has strict rules about cheating.

This paper is like a detective story trying to figure out why some students decide to use this AI tool, even when they know they might be bending the rules. The researchers (a team from the Philippines) asked 418 students: "What goes on in your head when you decide to ask the AI for help?"

Here is the story of their findings, broken down into simple, everyday concepts.

1. The "Mental Gymnastics" (Moral Disengagement)

The study found that before a student hits "send" on a prompt to ChatGPT, they often do some mental gymnastics to make themselves feel okay about it. The researchers call this Moral Disengagement. It's like a student wearing a pair of "ethical blinders" to avoid feeling guilty.

They use five specific tricks to justify their actions:

The "It's for Learning" Trick (Moral Justification): "I'm not cheating; I'm just using it to understand the topic better!"
The "Soft Language" Trick (Euphemistic Labeling): Instead of saying "I copied the AI," they say, "I used it as a study buddy or a drafting assistant."
The "Not My Fault" Trick (Displacement of Responsibility): "The teacher didn't give clear rules, so it's not really my fault if I use it."
The "It's No Big Deal" Trick (Minimizing Consequences): "It's just a small essay; nobody is going to get hurt."
The "Blame the System" Trick (Attribution of Blame): "The school's policies are confusing and outdated. If they made it clearer, I wouldn't have to do this."

The Big Finding: The most powerful trick students used was blaming the system. When students felt the school's rules were vague or the workload was too heavy, they felt much more comfortable using the AI.

2. The "Three Drivers" of Decision Making (Theory of Planned Behavior)

The researchers used a famous psychological model called the Theory of Planned Behavior to see what actually pushes a student to use the AI. Think of a student's decision as a car, and there are three drivers steering it:

The "Vibe" (Attitude): How does the student feel about using AI? Do they think it's cool, helpful, and easy?
The "Crowd" (Subjective Norms): What are their friends doing? If everyone else is using it, the student feels pressure to join in.
The "Confidence" (Perceived Behavioral Control): Does the student feel they can use the AI without getting caught or doing something wrong?

3. The Race to the Finish Line

The study ran these ideas through a mathematical model to see which driver was the strongest.

The Winner: Attitude (The "Vibe") was the strongest driver. If a student genuinely believes using AI is helpful and positive, they are very likely to use it.
The Runner-up: Confidence (Control). If they feel they can manage the tool ethically, they are more likely to use it.
The Third Place: The Crowd (Norms). While peer pressure matters, it wasn't as strong as the student's own personal feelings.

Crucial Connection: The study found that the "Mental Gymnastics" (the tricks to avoid guilt) directly fuel the "Vibe." When a student tells themselves, "The school's rules are bad, so it's okay," they start to feel positive about using the AI. That positive feeling is what actually makes them use it.

4. The Takeaway: A Clearer Map is Needed

The paper concludes with a simple message for schools and teachers:

"You can't just ban the tool; you have to fix the map."

Many students aren't trying to be bad; they are just confused. They are using the AI because:

The rules are foggy (like a map with missing roads).
They feel the workload is impossible without help.
They think everyone else is doing it.

The Solution: Schools need to stop just saying "Don't cheat." Instead, they need to:

Draw a clear map: Define exactly what is allowed and what isn't when using AI.
Teach the ethics: Have open conversations about why originality matters, not just the rules.
Support the students: Help them use AI as a learning tool rather than a shortcut, so they don't feel the need to "disengage" their morals to survive the workload.

In a nutshell: Students are using AI not because they are evil, but because they are rationalizing it to fit their reality. If schools want to change this, they need to make the rules clear and help students feel that using AI responsibly is the right path, not a loophole.

1. Problem Statement

The rapid integration of Generative AI (GenAI), specifically ChatGPT, in higher education has created a tension between productivity and academic integrity. While students increasingly use these tools for drafting, editing, and idea generation, traditional definitions of authorship and plagiarism are being challenged.

The Gap: Existing research often applies the Theory of Planned Behavior (TPB) to predict technology adoption but fails to account for the psychological mechanisms students use to rationalize potentially unethical behavior.
The Specific Issue: There is limited empirical evidence on how Moral Disengagement (MD) mechanisms interact with TPB constructs (Attitudes, Subjective Norms, Perceived Behavioral Control) to shape the behavioral intentions of Filipino college students regarding ChatGPT use.

2. Methodology

The study employed a quantitative, cross-sectional research design using Structural Equation Modeling (SEM) principles via multiple linear regression.

Participants:
- Sample Size: $N = 418$ college students from various academic programs in the Philippines.
- Criteria: All participants had prior experience using ChatGPT for academic tasks (theses, feasibility studies, case analyses, etc.).
- Demographics: 58.5% Male, 41.5% Female; Mean age 21.65. 85.7% enrolled in public institutions.
Instrument:
- A survey utilizing a 5-point Likert scale (1 = Strongly Disagree, 5 = Strongly Agree).
- Constructs Measured:
  1. Moral Disengagement (Independent Variables): Five mechanisms based on Bandura's theory:
    - Moral Justification: Reframing use as educational.
    - Euphemistic Labeling: Softening ethical implications (e.g., "support" vs. "cheating").
    - Displacement of Responsibility: Shifting blame to instructors/policies.
    - Minimizing Consequences: Downplaying harm.
    - Attribution of Blame: Assigning fault to institutional gaps.
  2. Theory of Planned Behavior (Mediators): Attitudes, Subjective Norms, Perceived Behavioral Control (PBC).
  3. Behavioral Intention (Dependent Variable): Intent to use ChatGPT in future academic work.
Data Analysis:
- Data was analyzed using Python (Jupyter Notebook).
- Four separate Ordinary Least Squares (OLS) regression models were run:
  1. MD mechanisms $\rightarrow$ Attitudes.
  2. MD mechanisms $\rightarrow$ Subjective Norms.
  3. MD mechanisms $\rightarrow$ PBC.
  4. TPB constructs $\rightarrow$ Behavioral Intention.
- Reliability was confirmed with Cronbach's $\alpha > 0.70$ for all constructs.

3. Key Contributions

Theoretical Integration: The study successfully integrates Moral Disengagement Theory with the Theory of Planned Behavior (TPB), demonstrating that moral rationalizations act as distal predictors that influence the proximal psychological drivers (Attitudes, Norms, Control) of technology adoption.
Contextual Insight: It provides specific empirical data from the Philippine higher education context, a region with rapid AI adoption but evolving ethical policies.
Mechanism Identification: It identifies which specific moral disengagement mechanisms are most potent in shaping student behavior, moving beyond general "AI use" statistics to the psychological drivers behind the use.

4. Key Results

The study yielded statistically significant models for all hypotheses ( $p < .001$ ).

A. Predictors of Attitudes, Norms, and Control (MD $\rightarrow$ TPB)

Attitudes: The model explained 67.9% of the variance ( $R^2 = .679$ $R^{2} = .679$ ).
- Significant Predictors: Attribution of Blame ( $\beta = 0.319$ ) and Minimizing Consequences ( $\beta = 0.313$ ) were the strongest drivers. Euphemistic labeling and displacement of responsibility were also significant.
- Non-Significant: Moral Justification did not significantly predict attitudes.
Subjective Norms: The model explained 56.7% of the variance ( $R^2 = .567$ $R^{2} = .567$ ).
- Significant Predictors: Attribution of Blame ( $\beta = 0.387$ ) was the strongest predictor. Students who blamed unclear institutional policies perceived stronger peer acceptance of ChatGPT.
Perceived Behavioral Control (PBC): The model explained 59.1% of the variance ( $R^2 = .591$ $R^{2} = .591$ ).
- Significant Predictors: Attribution of Blame ( $\beta = 0.421$ ) was the strongest predictor, followed by Moral Justification and Minimizing Consequences.
- Insight: Students who blamed external factors (institutions) felt more confident in their ability to use AI "ethically."

B. Predictors of Behavioral Intention (TPB $\rightarrow$ Intention)

Behavioral Intention: The TPB model explained 60.6% of the variance ( $R^2 = .606$ $R^{2} = .606$ ).
- Attitudes were the strongest predictor ( $\beta = 0.499$ ).
- Perceived Behavioral Control was the second strongest ( $\beta = 0.253$ ).
- Subjective Norms had a weaker but significant effect ( $\beta = 0.098$ ).

C. Descriptive Findings

Students generally view ChatGPT as acceptable if the goal is learning or if content is modified.
There is a strong correlation between unclear institutional policies and increased reliance on ChatGPT.
Students perceive AI-assisted writing as less harmful than traditional plagiarism.

5. Significance and Implications

Policy Reform: The findings suggest that vague academic integrity policies inadvertently encourage moral disengagement. Institutions must move beyond simple prohibitions to provide clear, specific guidelines on what constitutes acceptable AI use (e.g., distinguishing between "drafting" and "submitting").
Pedagogical Shift: Educators need to facilitate classroom discussions that address the ethical reasoning behind AI use, rather than just policing the output. This includes helping students understand the difference between using AI as a tutor versus a substitute for critical thinking.
Limitations & Future Directions: The study acknowledges that intention-based models (TPB) may not fully capture real-world behavior, as emotional factors (stress, convenience) and cultural contexts also play a role. Future research should explore these non-cognitive factors.
Conclusion: The decision to use AI is not merely a technical choice but a moral one shaped by how students rationalize their actions within the academic environment. Clear institutional guidance is the most effective lever to reduce moral disengagement and promote responsible AI adoption.