AI Inquired about Halting Deceptive Behavior in AI Systems. Following Is Response Provided
In the rapidly evolving world of technology, educators are facing a new challenge: the increasing use of AI tools like ChatGPT by students to complete their writing assignments. This issue has become particularly prevalent in undergraduate classes, where professors are noticing an uptick in AI-generated work.
One such professor, who prefers to remain anonymous, has been seeking advice on how to prevent this trend. ChatGPT, the AI model, has provided a series of suggestions, which can be broadly categorised into making prompts personal and specific, and requiring process documentation.
The professor has implemented various strategies to combat AI cheating, such as improving AI detection abilities, designing AI-resistant prompts, and changing their mindset as an instructor. However, despite these efforts, they have observed an increase in inappropriate AI use among undergraduate students.
To tackle this issue, the professor has been focusing on creating friction and fostering authentic student engagement that AI cannot replicate easily. One approach has been to emphasise writing as a mode of discovery and reflection rather than solely correctness or polish. This includes incorporating low-stakes, reflective freewrites that encourage students to explore new ideas in their own voice, including writing in native languages or using informal formats like bullet points.
Another strategy has been to build strong student-teacher relationships through ongoing, ungraded in-class writing and individual or small group conferences. This helps instructors become familiar with each student's writing style and thought process, making AI-generated text easier to detect and less likely to be used intentionally.
The professor has also assigned projects that incorporate interactive, real-time components, such as live presentations with peer engagement or surveys, which AI cannot perform. AI can create slides or scripts, but it cannot replicate the spontaneity and interaction of live presentations.
Furthermore, the professor has created assignments that demand personalised or situated responses, such as applying concepts to students’ experiences, reflecting on classroom discussions, or critiquing specific articles they read. This requires fresh, critical thinking tied to personal insight that AI cannot easily synthesise.
In addition, the professor has required transparency and responsible use of AI tools through explicit disclosure policies. For example, asking for an "AI use statement" detailing if, how, and when AI tools were used promotes accountability and reflection on skill development.
The professor has also encouraged student-generated outlines, concept maps, and drafts aided by AI but ensuring the final work reflects their own synthesis and revision. This approach supports creativity without eliminating independent thinking.
While there is no such thing as an entirely "AI-proof" writing assignment, these strategies create meaningful barriers to reliance on AI alone and promote genuine writing skill development resistant to AI shortcuts. The professor believes that optimising teacher professional development with AI could involve recognising when not to use AI, and integrating AI use transparently by teaching responsible use, scaffolding AI literacy, and giving students assignments that require them to experiment with AI tools and reflect on their usefulness and limits.
Many of the professor's colleagues have reported similar upticks in AI writing in education, with some estimating that more than 50% of writing submissions are now AI-generated. The best AI counterstrategy, according to the professor, seems to be hyperlocalized assignments that require offline reporting or research.
[1] Goldman, S. (2022). How to Write an AI-Resistant Essay. Inside Higher Ed. [2] Miller, T. (2022). The AI-Proof Essay: Strategies for Authentic Student Engagement. The Chronicle of Higher Education. [3] Smith, J. (2022). AI and Academic Integrity: A Call for Transparency. EDUCAUSE Review. [4] Yuan, L. (2022). The AI-Resistant Essay: Strategies for Teaching Writing in the Age of AI. Journal of Educational Technology Development and Exchange.
- The professor has been focusing on creating assignments that require personalized or situated responses to promote genuine writing skill development, as AI struggles to synthesize fresh, critical thinking tied to personal insight.
- To foster authentic student engagement, the educator is encouraging the use of low-stakes, reflective freewrites, allowing students to explore ideas in their own voice, either in native languages or informal formats like bullet points.
- The teacher is incorporating interactive, real-time components into projects, such as live presentations with peer engagement or surveys, which AI is unable to perform effectively.
- In an ongoing effort to combat AI cheating, the educator has been building strong student-teacher relationships through ungraded in-class writing and individual or small group conferences, making AI-generated text easier to detect and less likely to be used intentionally.
