I Used to Teach Students. Now I Catch ChatGPT Cheats

AI Writing vs Actual Understanding

• Multiple comments stress that AI-generated essays don’t equate to knowledge: being able to organize, argue, and write yourself is part of what’s being learned.
• Some argue that disciplines overly focused on “stringing words together” rather than ideas are now exposed, since AI can mimic that surface-level discourse.
• There’s concern that LLMs short-circuit the formative struggle of clear thinking, especially in philosophy/ethics, where the point is to form one’s own principles, not just produce text.

Impact on Students and Hiring

• Interviewers report that graduates from AI-permissive programs often can’t explain basic concepts or their own code; however, others note this was true of many students even pre-ChatGPT.
• Hiring managers say it’s relatively easy to spot candidates relying on LLMs during coding interviews by probing “why” decisions; online assessments are seen as more easily gamed.
• Some worry that widespread AI use will commoditize junior workers, rewarding those who can “operate AI” rather than deeply understand domains.

Cheating, Ethics, and Value of Degrees

• There’s disagreement over “cheaters only cheat themselves”: critics point out downstream harms—unsafe professionals, incompetent bureaucrats, and devalued degrees.
• AI makes old patterns of outsourcing work (e.g., buying papers) cheaper and more invisible, amplifying existing problems rather than creating them.

Assessment and Pedagogical Responses

• Proposed or existing countermeasures:
  • Oral exams (notably in Italy), viva-style defenses of projects, in-person code demos, and oral questioning on specific commits.
  • More controlled, invigilated exams; some professors say institutional policies actively obstruct this.
  • Explicit policies emphasizing student responsibility: the professor will teach, but not police every instance of cheating.
• Others suggest embracing AI: either ban it only for final outputs while allowing as a “library,” or design courses where demonstrated value beyond what an LLM can do is required.

Purpose of Higher Education

• Ongoing tension between education as:
  • Learning “how to learn” and to think independently, versus
  • A credential needed to access jobs.
• Several note that many students treat university as a hoop to jump through, making heavy AI use rational from their perspective.
• The thread repeatedly returns to signaling theory: if degrees become easy to fake intellectually, their signaling value and public support for the system may erode.