Should Schoolwork Produced in Collaboration with ChatGPT Ever be Graded?

And if so, how?

Since the launch of ChatGPT in November 2022, I have frequently asserted that I would not grade work produced by a student in collaboration with this AI tool. I would only grade a student’s ability to answer logically anticipatable questions related to their work after it’s been submitted, I kept repeating.

The reasoning behind this position was that I considered it crucial, as most teachers undoubtedly do as well, to assess a student’s own intellectual capabilities rather than the polished outcome of an AI-assisted task.

But now, influenced by ongoing conversations with respected education leaders I know, I am gradually reconsidering this standpoint, though I am inclined to assign the grading of the collaborative far less weight, contemplating a 90–10% split between the follow-up responses and the quality of the AI-assisted work.

With this in mind, I have produced the below scoring rubric for work produced in collaboration with ChatGPT, with this scoring rubric not to be used to help score the student’s responses to logically anticipatable follow-up questions, only the work produced in collaboration with ChatGPT.

Category 1: Collaboration with ChatGPT (5 points)

• 1 point: The work reflects little to no collaboration with ChatGPT. It seems like the student submitted a single, poorly constructed prompt and didn’t refine it or interact further with the AI.
• 2 points: The work shows some level of interaction with ChatGPT, but the prompts are not critically thought out or there is no evidence of substantial back-and-forth exchanges.
• 3 points: The work indicates that the student made a good effort to engage with ChatGPT, using well-worded prompts and some exchanges. However, the depth or effectiveness of the collaboration is inconsistent or limited.
• 4 points: The work demonstrates strong collaboration with ChatGPT. There’s clear evidence of multiple well-worded, critical thinking prompts, as well as active exchange. Yet, the process could have been pushed further for an even richer interaction.
• 5 points: The work excellently demonstrates a high level of collaboration with ChatGPT, suggesting multiple, carefully constructed and critically thought-out prompts. The back-and-forth exchange was substantive and meaningful, enhancing the overall quality of the work.

Category 2: Original Thinking (5 points)

• 1 point: The work lacks original thought or ideas from the student. There are no instances where the student has added value beyond the information provided by ChatGPT.
• 2 points: The work contains slight evidence of original thought or ideas, but these are either weak, superficial, or not meaningful to the overall work.
• 3 points: The work displays some original thinking from the student, including their own arguments or viewpoints, but the depth or integration of these thoughts could be improved. There is a limited effort to add value beyond what the bot has provided.
• 4 points: The work exhibits a high level of original thinking with well-articulated arguments, viewpoints, and additional research (like expert interviews, student polls, book quotations etc.). However, the original contributions could be more comprehensive or deeply integrated into the work.
• 5 points: The work brilliantly demonstrates extensive original thinking. The student’s arguments, viewpoints, and additional research are integral to the work, providing significant value beyond the bot’s contributions. The original thoughts are not only serious and meaningful but also showcase the student’s critical thinking skills to a great extent.

Thoughts?

• Do you think that work produced by a student in collaboration with ChatGPT should never be graded? Or, conversely, should we recognize and evaluate such collaborations as a part of our evolving educational landscape?
• My current proposal assigns only 10% of the final grade to the quality of the collaboration with ChatGPT, with the remaining 90% being determined by the student’s ability to answer logically anticipatable follow-up questions. What are your thoughts on this distribution?
• In the suggested scoring rubric, I have considered only two criteria — the quality of the collaboration itself and the extent to which the work provides evidence of the student’s original thinking. Do you think additional criteria should be included?