Part 3: What Students Thought About AI-Assisted Grading and Where This Experiment Goes Next
This is part 3 of a three-part series. Catch up by reading Part 1 and Part 2.
When I began experimenting with AI-assisted grading, I had several concerns.
Would the feedback feel impersonal?
Would students think the responses were artificial or generic?
Would the quantity of feedback become overwhelming?
Would students actually read and apply it?
After a full semester of testing, the answer to most of those concerns was surprisingly clear: students overwhelmingly valued the experience.
This third and final article in my series focuses on what students actually said about the grading approach, what worked, what I learned, and where I plan to take the experiment next.
For context, the AI was never making grading decisions. I remained fully responsible for evaluation, scoring, interpretation, and judgment. The AI functioned primarily as an organizational and drafting partner. I dictated raw observations and thoughts while reviewing student work, then used AI tools to help structure and refine that feedback into clearer, more detailed, and more actionable responses.
What I did not anticipate was how strongly students would respond to the depth and specificity of the feedback itself.
Across final reflection papers, unsolicited Canvas comments, course evaluations, and graduation surveys, students consistently described the feedback as one of the most valuable parts of the course experience.
Several themes appeared repeatedly:
• Students appreciated the depth and quantity of the feedback
• They felt the feedback clearly explained both strengths and weaknesses
• They said it helped them improve on future assignments
• They felt more confident about assignment expectations
• They viewed the feedback as constructive rather than punitive
• They felt supported while still being challenged academically
This became especially important in courses built around formative assignments leading toward a large summative deliverable. In those environments, detailed iterative feedback matters because students are building toward a final project step-by-step. If the formative feedback is weak, delayed, or unclear, the final outcome suffers.
Instead, many students described the feedback as a running guide throughout the semester.
One student wrote:
“I LOVED everything about the specific feedback given on assignments. It was so fast and so helpful. I would often skim through it the second I saw it hit my inbox, and then I would go back to it later when I had time to really dig into it. The balance of positive vs critical was great, and the tone felt constructive and supportive, I think. Overall, it made it seem like you actually knew what was going on in my assignments and were invested on a level that most professors don’t go for.”
–Student Capstone Reflection
Another student explained how the feedback improved future performance:
“I appreciated how clearly you identified both strengths and areas for improvement, as it made it much easier to apply those learnings to future assignments and, in the end, strengthen my final campaign.”
– Student Capstone Submission Comment
Additional student comments reinforced similar themes:
“Professor Kropp was always extremely detailed with his feedback on our work which was super helpful.”
– Student Course Evaluation
“Dr. Evan Kropp was incredible. He provided thorough feedback that helped me grow significantly over the semester. He encouraged freedom of expression, which made me feel that my voice and perspective were valued. Rather than imposing a strict right or wrong, he was very open, approachable, and undeniably helpful.”
–Student Graduation Survey
“Evan Kropp challenged me more than any other instructor during my time at UF, particularly through the capstone course. While the experience was demanding, his detailed and thorough feedback pushed me to refine my thinking and produce stronger, more strategic work. He created an environment that was both rigorous and supportive, and I can confidently say I learned the most from his class throughout my experience as a UF student.”
–Student Graduation Survey
“Evan’s lectures and his very detailed and helpful feedback upon grading assignments was a step above anything I received during the program. I felt totally supported and empowered to do my work and know how I can improve.”
–Student Graduation Survey
One surprising outcome was that students did not describe the feedback as excessive. I initially worried that the volume of comments might become overwhelming. Instead, students often described it as substantial but useful. Many indicated they returned to the feedback multiple times while working on later assignments.
Equally noteworthy: I did not intentionally omit negative student feedback from this article. There simply was not meaningful criticism of the grading approach itself. That does not mean the system is perfect, and it certainly does not mean the experimentation should stop. It simply suggests that students recognized value in receiving detailed, actionable, growth-oriented feedback.
That distinction matters.
Too often, discussions around AI in education focus exclusively on efficiency. Efficiency matters, particularly in large online courses, but this experiment reinforced something more important for me personally: AI may help educators scale individualized support more effectively than many of us initially assumed.
The next phase of this experiment will focus less on individual assignments and more on longitudinal student development.
Next semester, I plan to create individualized feedback folders for each student. Rather than viewing assignments in isolation, I want to explore whether AI can help identify broader developmental patterns across an entire semester.
For example:
• Which weaknesses consistently improve over time?
• Which challenges persist despite repeated feedback?
• Which types of feedback appear most actionable for students?
• Are students implementing suggestions between assignments?
• Are there recurring communication, strategy, research, or writing issues that become visible only across multiple submissions?
I also plan to conduct post-semester analysis by uploading formative assignment feedback alongside final summative evaluations to determine whether there were missed themes, overlooked patterns, or opportunities for earlier intervention.
Importantly, this remains an evolving process. I still review, edit, confirm, and refine all feedback personally. The goal is not automation of teaching. The goal is augmentation of teaching.
At its best, this approach allows instructors to spend less time wrestling with formatting and organization and more time focusing on what actually matters: helping students learn, improve, and grow.
For me, that has been the biggest takeaway from this experiment. AI did not reduce the human side of teaching. In many ways, it amplified it.
One of the most interesting takeaways here is that AI-assisted feedback appears to have made students feel more supported, not less. The distinction between automating teaching and augmenting teaching is important. Used thoughtfully, AI may help instructors provide the kind of detailed, developmental feedback that students often want but faculty don't always have time to deliver at scale.
A great example of using AI to enhance, not replace, teaching. The focus on better feedback and student growth, rather than just efficiency, is what makes this approach compelling.