A history professor at Alcorn State University, Dr. Jason Gibson, devised an "invisible prompt trap" to identify students using AI for their assignments. He embedded instructions in white text within the assignment prompt, directing the AI to include nonsensical references to Madagascar. Out of 35 students across two classes, 32 copied and pasted the prompt into an AI, and subsequently submitted AI-generated responses that included the out-of-place Madagascar references, leading to their failure on that portion of the midterm. This method highlights the ongoing challenge educators face with AI-assisted cheating, as demonstrated by similar incidents and widespread student use of AI tools. AI
IMPACT Educators are developing creative methods to detect AI-generated content, highlighting the ongoing arms race between AI-assisted cheating and detection.
RANK_REASON Article describes a method used by an educator to detect AI misuse, rather than a new AI product or research.
Read on Mastodon — fosstodon.org →
- Alcorn State University
- Dr. Jason Gibson
- Madagascar
- TikTok
- TechSpot
- Brown University
- ChatGPT
- Industrial Revolution
AI-generated summary · Google Gemini · from 8 sources. How we write summaries →