From the Chronicle of Higher Education: ...The ed-tech marketplace is filled with tools that promise to either help students cheat or to detect AI cheating. It’s an accelerating arms race, pitting the newest AI technologies that mimic human behaviors against those that discern what is authentically human. The depth and layers of surveillance are “really sort of shocking,” said Marc Watkins, assistant director of academic innovation at the University of Mississippi. He spends many hours staying on top of these tools and how they are advertised to instructors and to students.
“Imagine setting up your system where half your time is going to be watching a sped-up process-tracking report to see if a student is copy and pasting something here, or going into an AI-detection report and wondering if 38 percent of this is AI-generated versus 15, and what level becomes a point where you bring a student up,” he said. “It’s too much, and people check out.”
The latest AI-detection tool, Pangram, has won attention for its accuracy in spotting AI writing. Yet surveillance seems to be a losing proposition. Watkins noted that a student could purchase Pangram alongside a generative-AI program such as Claude Code, then have it repeatedly rework an AI-written essay 100 times over until it passes cleanly through the detector.
The students who win the cheating game “would be the ones who had the most resources and also had the most knowledge about how these systems work,” he said. “Our assessment strategy would basically only really detect and punish those students who didn’t have that access.” ...
Full story at https://www.chronicle.com/article/the-new-nanny-state.
No comments:
Post a Comment