The Impact of AI Detection Tools
on BNBU Students' Writing Anxiety
and Coping Strategies

EAP II Research Presentation

You spent a weekend writing an essay.
You didn't use any AI.
You submitted it.

And then — the system says
"This is AI-generated."

This is happening at BNBU.

Background

The Problem with AI Detection

What are AIDTs?

Software (e.g., Turnitin) used by universities to identify AI-generated content in student work

The Bias

AIDTs produce high false-positive rates against non-native English speakers

(Liang et al., 2023)

Research Questions

RQ1 — To what extent do BNBU students experience or fear false-positive misjudgments?
RQ2 — How does awareness of AIDT biases impact writing anxiety?
RQ3 — What coping strategies do students adopt, and what motivates them?

Methodology

How We Studied This

39 BNBU undergraduates

14-item questionnaire

5-point Likert scale (SLWAI)

Convenience sampling

Descriptive + Correlation analysis

Anxiety scale adapted from Cheng's (2004) SLWAI framework

RQ1 — False Positives

59%

reported being falsely flagged

97.4%

worried about future false positives

0

selected "Not worried at all"

RQ2 — Writing Anxiety

4.30 / 5.0

Overall anxiety score (SD = 0.74)

Avoidance behavior scored highest.
Students are shifting from
"writing well" → "avoiding being flagged"

RQ2 — Group Comparison

False-positive experience
amplifies anxiety

4.54

Experienced false positives
(n = 23)

3.39

No false-positive experience
(n = 7)

t = 3.08 — a statistically significant difference

RQ3 — Coping Strategies

82.1%

used humanizer tools

97.4%

changed writing habits

82.1% say their writing became less natural

Discussion

The Paradox

AI detection tools are designed to protect writing quality.

But the data shows students are
deliberately degrading their own writing
to avoid being flagged.

The tool designed to protect writing...
is damaging writing.

(Balalle & Pannilage, 2025)

Implications & Limitations

Shift approach

From rigid detection to constructive AI integration

Transparent systems

Appeal mechanisms & process-based assessment

Future research

Larger samples, qualitative methods, longitudinal designs

Limitation: N=39, 94.9% freshmen, cross-sectional design

The question is no longer
"Are students using AI?"

Are our detection tools
doing more harm than good?

References

Balalle, H. & Pannilage, U. (2025). AI technologies and academic integrity: A systematic review. Journal of Academic Ethics.
Cheng, Y.-S. (2004). A measure of second language writing anxiety. Journal of Second Language Writing, 13(4), 313–335.
Lazarus, R. S. & Folkman, S. (1984). Stress, Appraisal, and Coping. Springer.
Liang, W. et al. (2023). GPT detectors are biased against non-native English writers. Patterns, 4(7).
Perkins, M. et al. (2024). GenAI detection tools and academic integrity. Intl. J. of Educational Integrity, 20(9).

Thank you. Questions?

The Impact of AI Detection Toolson BNBU Students' Writing Anxietyand Coping Strategies