Balenet

Rate students by comparing them to AI

baleboy_a_human_and_a_robot_facing_each_other_both_taking_a_tes_d026755a-7244-4ea2-a4f7-1a668ccb298a-2.png

Image by Midjourney: "a human and a robot facing each other, both taking a test, and a judge between them evaluating the result, in a dystopian setting"

Like pretty much everyone else, I've been giving a lot of thought to AI lately. I regularly use Github Copilot for coding, ChatGPT for writing and comprehension, and Midjourney to create images for my posts, like the one above. The experience is both exhilarating and terrifying. It's undeniable that these tools are the biggest productivity boosters since the advent of the World Wide Web, and they're bound to have similar earth-shattering impacts on various occupations. Why would you want to visit a doctor or a psychologist who merely repeats information from a manual, when a chatbot can probably do the same, or even better? Humans need to outperform AI if they want to stay relevant.

I've thought of one potential solution: to pass their exams, university students must outperform AI. For instance, the final test in medical schools could involve a student making a remote diagnosis for a patient while an AI agent does the same; the patient then determines which diagnosis was superior. If the AI prevails, there seems to be little point in the student pursuing a career in medicine.

Even for math-based professions, where AI has an unfair advantage, it should still be possible to design a test that requires creativity and "humanity", for example by evaluating a full project rather than isolated questions.

P.S.: The tile for this post was generated by ChatGPT using the GPT-4 model. I asked a couple of follow-up questions:

something more tongue in cheek?

"Beat the Bot or Bust: The New Graduation Threshold"

how would al yankovic say it?

"Robots in the Classroom: Can You Outsmart Your Toaster?"