Examining Student and Teacher Perspectives on Undisclosed Use of Generative AI in Academic Work

Research Background and Issues

• Identified Problems or Challenges:
  The authors point out that students may use generative artificial intelligence (Generative AI, GenAI) tools in academic work without disclosure, and their behaviors, motivations, and impacts remain unclear. Such behaviors could affect academic integrity, teacher-student relationships, and learning outcomes. Additionally, due to the immaturity of AI content detection technologies, misunderstandings between students and teachers may lead to conflicts.

• Importance of the Issue:
  With the widespread adoption of generative AI tools such as ChatGPT and Google Gemini, these tools are significantly transforming educational and academic work practices. If the use of these tools lacks transparency, it could undermine educational goals, impact student learning outcomes, and exacerbate issues of academic integrity.

• Research Motivation and Related Work:
  Current research primarily explores how generative AI tools support education (e.g., writing, topic exploration) and their negative impacts (e.g., fostering plagiarism and over-reliance on AI). However, there is limited research on students' and teachers' perspectives on non-disclosure behaviors related to GenAI and their coping strategies. The authors aim to fill this research gap.

Solution

• Methods or Solutions:
  The authors adopted a mixed-methods approach, using online surveys and semi-structured interviews to investigate students' and teachers' perspectives on non-disclosure behaviors related to GenAI.

  1. Conducted an online survey with 97 university students to understand their frequency of GenAI use, disclosure practices, and motivations.
  2. Conducted in-depth interviews with 15 students who engaged in non-disclosure behaviors and 9 teachers who suspected students of non-disclosure use of GenAI.
  3. Analyzed students' non-disclosure strategies and motivations, teachers' coping strategies, and how cognitive dissonance theory can explain students' behaviors.

• Innovativeness:
  This study is the first to apply cognitive dissonance theory to analyze students' concealment of GenAI usage, revealing their strategies to reduce internal conflict by adjusting behaviors or changing beliefs. Additionally, the study integrates a teacher-student interaction perspective, providing rich empirical data and proposing recommendations for improving transparency through teaching practices.

• Implementation Steps and Key Techniques:
  The study relies on statistical analysis methods for survey data (e.g., Mann-Whitney test and Friedman test) and reflexive thematic analysis for qualitative data. Responses from participants were coded and organized to summarize behavioral patterns and practical adjustments for both students and teachers.

Research Findings

• Specific Findings:

  • Student Behavior:

    • Students' decisions to disclose GenAI use are influenced by school policies, task types, and peer relationships. Their main strategies include rewriting AI-generated content to obscure its origin, limiting usage to low-stakes tasks, and selectively disclosing usage to specific peers.
    • Primary motivations for using GenAI include saving time, reducing academic stress, and completing tasks quickly. Students often justify their behavior by comparing GenAI to search engines or spell-check tools.
    • Emotional responses to non-disclosure behaviors are complex. Most students do not experience significant guilt but are concerned about potential academic penalties.

  • Teacher Strategies:

    • Teachers attempt to identify AI-generated content by detecting features such as generalized language and impersonal tone or by using AI detection tools, though they lack confidence in relying solely on these tools.
    • Teachers' coping strategies include adjusting course assignments to mitigate the risk of GenAI misuse, explicitly defining situations where disclosure is required, and redesigning assignments to be more challenging and meaningful.
    • Some teachers use classroom case studies to demonstrate responsible GenAI usage, fostering students' understanding of ethics and transparency.

• Advantages Over Existing Solutions:
  This study not only reveals the complex attitudes of students and teachers toward GenAI but also uncovers the psychological motivations behind students' non-disclosure behaviors through the lens of cognitive dissonance. It provides a deeper socio-technical framework for understanding these behaviors beyond mere technical detection.

• Experimental or Evaluation Results:
  The survey and interview data indicate that students primarily use GenAI to improve efficiency rather than out of laziness, while teachers face practical challenges in detecting GenAI usage. There is a notable discrepancy between the perceptions and behaviors of students and teachers.

• Limitations and Future Directions:

  • The teachers in this study were primarily from a single university, limiting the representativeness of the sample. Additionally, gender and racial distributions were uneven.
  • Social desirability bias may have influenced participants' self-reports.
  • Future research could explore how to encourage teachers to disclose their own use of GenAI and examine students' attitudes toward such disclosures, fostering two-way transparency in teacher-student communication.

Conclusion

This study explores the differing perspectives of students and teachers on the use of generative AI and analyzes the cognitive reasons behind students' concealment behaviors. It provides practical guidance for educators and calls for reducing opacity in AI tool usage in education through course design, ethics education, and reflective dialogue, thereby fostering a more honest and equitable educational environment.