How AI Can Enhance the Grading Process

After tucking my son into bed, I’m hit with the realization that since waking up at 6:00 a.m., I’ve not had a moment’s rest, nor have I managed more than brief exchanges with my wife. She’s deeply engrossed in grading sixth-grade math quizzes, a world away in her concentration. Shifting my focus, I dive into assessing a substantial stack of high school history essays, with a firm deadline to return them by tomorrow.

The punctuality I expect from my students is the same standard I set for myself in returning their assignments. Missing a deadline isn’t an option, unless unexpected events or an illness intervenes. In such cases, I may offer bonus points or postpone future deadlines. More than just a pledge, this is my commitment to their success, requiring their best effort and guaranteeing mine in return.

Enhancing Efficiency, Precision, and Fairness

Late that night, CoGrader—a new artificial intelligence (AI)–enhanced platform—piques my interest. A notification on social media directs me to their website, boasting a compelling promise: “Reduce grading time by 80% and provide instant feedback on student drafts.” The allure is heightened by the offer of a 30-day trial, free and without requiring a credit card.

Intrigued by this “AI copilot for teachers,” I sign up to see how its feedback stacks up against my own. I upload a student’s essay on Reconstruction, which I’ve already evaluated and annotated. The results astonish me with their accuracy and detail and are neatly presented in a customizable rubric.

“Your essay employs an organizational structure that shows the relationships between ideas, providing a cohesive analysis of the topic,” reads a portion of the written feedback. “You use transitions effectively to guide the reader through your argument.” This feedback mirrors my own observations, validating my assessment, and increasing my confidence in CoGrader.

CoGrader also deeply impresses me with its “Glow” section, which offers specific praise, and the guidance provided in its “Grow” counterpart, which provides similarly focused areas for improvement.

The inclusion of action items also fosters student inquiry, exemplified by one of the questions provided in my test upload: “How might you further enhance the connections between different sections of your essay to strengthen your argument?” This reflects my priority of encouraging critical thinking.

Even fatigued, I recognize that CoGrader’s value extends beyond saving time. By promising an “objective and fair grading system,” the platform provides a check against the unconscious biases that inadvertently influence grading, try as we might to curb them.

I dip into my caffeine stash and devote the rest of my waking hours to grading solo, yet I’m captivated by what CoGrader could offer me, my students, and the future of impactful feedback.

Annotated Feedback and Mitigating Teacher Guilt

The potential of CoGrader motivated me to contact its cofounder, Gabriel Adamante, to express my admiration and to get his thoughts about the rapidly advancing field of AI. I mentioned that I had expected something like CoGrader to emerge around the time of my son’s 10th or 11th birthday, not his fifth or sixth.

“I ask myself a lot of questions on what’s the responsible use of AI,” Adamante offered. “What is the line? I think we are living in the Wild West of AI right now. Everything is happening so fast, much faster than anyone would have expected. Like you said, you thought this would be five or six years away.”

I asked Adamante whether CoGrader, in this fast-paced environment, has plans to add annotated feedback on a student’s work.

“Honestly, I think we’re talking anywhere from three to five months until we do that,” Adamante said.

“That’s it?” I said in astonishment. “Really? I thought you were going to say three to five years.”

“No, no, no,” he replied. “It’s coming. It should come this year, in 2024. I won’t be happy if it comes in 2025.” 

Adamante acknowledged the dizzying effect of AI development, understanding the potential discomfort among educators.

“Of course teachers are mad when students use AI to do their work,” Adamante said. “That makes sense to me because the purpose of a kid doing the work is that they do the work. That’s the purpose. They should write to get practice. The purpose of a teacher grading is not that they grade. The purpose of a teacher grading is that they provide feedback to the student so that the student learns faster. The purpose of grading is not grading itself, whereas the purpose of writing an essay is writing an essay, because you’re practicing.”

Teachers who use CoGrader without reviewing the feedback contradict its intended purpose, which is to scaffold, but not replace, the human element, Adamante explained. He regularly communicates with educators, urging them to carefully read the comments instead of quickly clicking “approve” and moving on to the next submission. Taking time to read the results not only ensures that the teacher agrees, but also keeps the teacher informed about their students’ strengths and weaknesses.

Lessons Learned with ChatGPT 4 and Transparency

Following our conversation, I’ve remained mindful of applying Adamante’s insights to my current use of ChatGPT 4 to aid in providing written feedback. While I avoid asking it to generate feedback on my behalf, I do seek its assistance in clarifying my overarching comments and ensuring their coherence. With the paid subscription, I can even upload a Microsoft Word or PDF document for more precise and detailed assistance. I always meticulously review any output before sharing it with my students.

Adamante’s reminder highlights the importance of transparency in my use of AI to enhance feedback for my students. While I’m not particularly unsettled by my use of ChatGPT 4 as an assistance tool, I realize that I haven’t been as forthcoming about it as I should be regarding when and how I utilize it. I must delve into this matter within my stated classroom policies and through class discussions.

I want students to understand that I always thoroughly review their work and that employing AI to aid in providing feedback doesn’t diminish my dedication. Rather, it’s about finding the most effective way to leverage technology to support their growth as writers and learners. 

Reflecting on my discussion with Adamante, I’ve concluded that CoGrader’s precise feedback surpasses that of ChatGPT 4. Moreover, I foresee CoGrader further outshining it with the introduction of annotated feedback.

Now, if only I can get my wife on board with researching how AI can expedite and enhance her feedback on math work. We’d both get more rest and time together, before putting our son to sleep.