In this episode of Deep Learning Dialogues, hosts Whitney McKinlay and Katrina Gouett sit down with Dr. Adam Pacton to unpack the controversial practice of \"prompt injection\" or hiding invisible text in assignments to catch students using generative AI. Moving past the initial \"ick\" factor or ethical intuition that such tactics turn classrooms into policing environments rather than spaces of mutual trust, the conversation explores the concrete legal, technical, and pedagogical failures of these traps. Dr. Pacton explains how hidden instructions act as an accessibility nightmare under regulations like the ADA Title II and Ontario's AODA, because screen readers accidentally surface the invisible text, confusing students who rely on assistive technology. Rather than relying on flawed detection mechanisms, the episode advocates for a systemic paradigm shift toward \"process over product,\" urging institutions to provide teachers with the cover and resources needed to meaningfully redesign assessments and foster relational intelligence. You can read the episode summary here.
Dr. Adam Pacton is an associate teaching professor and Dean's fellow for AI literacy and integration at the College of Integrative Sciences and Arts at Arizona State University. Working at the intersection of pedagogy, ethics, and emerging technology, he focuses on institutional AI governance while helping educators move away from fear-based compliance toward thoughtful, human-centered classroom integration. With a PhD in rhetoric and composition, Dr. Pacton brings a deeply grounded perspective to language practices, change management, and student agency in the age of artificial intelligence. Outside of his academic and administrative leadership roles, he rounds out his human perspective as a father, an avid video game player, and a passionate fan of fantasy and sci-fi literature.
Episode Link:
-ASU Guidance on Hidden-Text/Prompt Injection Techniques in Assignments
","author_name":"Katrina Gouett and Whitney McKinley"}