Dec 26, 2022 - Technology

Professor warns about chatbot cheating: "Expect a flood"

This was the answer when I asked ChatGPT about nefarious activity.

Darren Hick, a philosophy professor at Furman University in Greenville, S.C., warns in an eye-opening Facebook post that cheating using the new OpenAI chatbot, ChatGPT, can be tough to catch:

  • "ChatGPT responds in seconds with a response that looks like it was written by a human — moreover, a human with a good sense of grammar and an understanding of how essays should be structured."

Why it matters: The chatbot generates new words based on "the soup of data in its neural network" — rather than searching the web. So there's no specific passage you can point to as proof a student cheated.

  • And plagiarism software probably won't catch it. "It's simply not designed to deal with this sort of thing," Hick told me by email.

I found Hick's post via a link in a "Tech Fix" column by the N.Y. Times' Brian X. Chen ("Using A.I.? Please Do It Responsibly"), who pointed out:

  • "People experimenting with ChatGPT were quick to realize that they could use the tool to win coding contests."

State of play: The Furman professor busted one student, who admitted she had used the chatbot.

  • Hick was tipped off because her essay "confidently and thoroughly" discussed her topic in ways "that were thoroughly wrong."
  • The assignment was to write a short essay on the paradox of horror. The student focused on the philosopher David Hume.

"It did say some true things about Hume, and it knew what the paradox of horror was, but it was just bull----ting after that," Hick writes.

  • "ChatGPT also sucks at citing, another flag."
  • "For freshman-level classes ... [e.g.] the dominant themes of Moby Dick, or the causes of the war in Ukraine ... this is a game-changer."

Hick told Axios the "big issue isn’t catching cheaters who use A.I.; it’s proving they did it, to whatever standard a school requires of proof."

What's next: "Administrations are going to have to develop standards for dealing with these kinds of cases ... FAST," Hick wrote on Facebook.

  • "Expect a flood, people, not a trickle. ... I expect I'm going to institute a policy stating that if I believe material submitted by a student was produced by A.I., I will throw it out and give the student an impromptu oral exam on the same material."

Want more stories like this? Sign up for Axios AM

Go deeper