Adversarial AI

The "Adversarial AI" (Integration)

Rather than banning AI, force students to expose its limitations. This treats AI as a faulty source that must be fact-checked and critiqued.

• The Assignment: "The Hallucination Audit"
  Step 1: The student prompts an AI to write a literature review on their topic.
  Step 2: The actual assignment is to critique that output. The student must verify every citation (finding real DOIs), identify which key scholars were omitted, and highlight where the AI’s synthesis is superficial or biased.

• The Output: A "Correction Report" where the student acts as the professor grading the AI.

• The Assignment: "The Counter-Argument Defense"
  - Students write their thesis, then feed it to an AI and ask it to "act as a skeptical peer reviewer and find holes in this argument."
  - The Requirement: The student’s final paper must include a section explicitly addressing the AI's critique and explaining why the human researcher is right and the AI reviewer is wrong (or how they adjusted their work based on it).