Efficient But Not Effective? Student Perceptions Of AI feedback

I have a confession to make. In my mathematics classroom, technology takes a backseat to connection. While I admire its efficiency, my students have taught me a valuable lesson: sometimes, the most powerful tools are the ones that feel human, tangible, and relatable. Technology offers accuracy and speed, qualities undeniably valuable in mathematics. Yet, what it often lacks is the warmth of connection, the reassuring stroke of a pen that says, “I see you. I understand what you’re struggling with, and I’m here to help.”

What AI Feedback Offers

Let me be clear, I do use AI tools. When I’m writing emails home, summarising student progress, or even creating personalised comments, AI lends me a hand. It’s like having a very polished assistant who’s good at getting to the point. For example, I can quickly use AI to reframe technical assessment jargon into something more accessible for students and parents. When combined with rubrics and outcome statements, it makes the administrative side of teaching far more efficient.

But even in these moments, AI feels like a tool, useful, but ultimately secondary to my voice. And that’s where the problem arises. No matter how much I personalise a comment or inject a bit of “teacherly” warmth, students can still work out what’s not entirely me.

What AI Feedback Misses

A few days ago, I had a discussion with my Grade 11 students. We were going over feedback from a recent test as we often do, and the topic turned to how feedback is delivered. One student casually mentioned, “Yeah, but Mr X generates all his feedback using AI.” Curious, I asked how they knew. “Too clean,” “It’s in ChatGPT format,” and “Doesn’t sound like him” were their responses.

What stood out wasn’t their critique of the content but their perception of authenticity, or lack thereof. They also pointed out, with more than a little irony, that they could generate similar feedback themselves. Their trust in the feedback wasn’t about the accuracy or detail; it was about the voice behind it. They wanted to feel my presence in the words, not an algorithm’s polish.

Why Handwritten Feedback Matters

When I asked my students which feedback they valued most, their answers were unanimous: handwritten comments on their papers. “If you could pick just one piece of feedback, what would it be?” I asked. “Handwritten on papers,” they chorused. Even if I carefully crafted an AI summary tailored to each student, it still wouldn’t replace the power of a scribbled note in the margin or a star sticker next to a correct answer.

This preference reminded me of the little joys I’ve seen in my classroom. Students light up at a handwritten “Well done!” or a unicorn sticker on their work. I’ve even had students tell me they’ve kept those notes for years. I still treasure a card my history teacher gave me for good luck before my GCSEs. There’s something inherently human about these tangible connections, they’re imperfect, personal, and irreplaceable.

The Implications for Educators

Later, I shared this with a colleague. We joked about our own reluctance to accept AI-generated essays. “If we don’t want AI essays,” he quipped, “why would our students want AI assessing theirs?” It was a lighthearted moment, but it underscored a deeper truth: students devalue comments they perceive to be AI-generated, no matter how personalised or polished they are.

So, what does this mean for educators and edtech developers? Maybe there’s something inherently undervalued in AI’s polish, not because it’s wrong or unhelpful, but because it’s missing the imperfections that signal a human mind at work. Perhaps we’re all wired to crave that human connection, even in something as procedural as feedback.

A Call for Reflection

My takeaway is simple but profound: feedback isn’t just about delivering information. It’s about building relationships and fostering trust. The gold star or unicorn sticker might seem small, but it’s a tangible sign of my personal engagement. The students see it, feel it, and value it in a way that no AI can replicate... yet.

What about you? Have you noticed similar preferences in your classroom or workplace? Do you think there’s a way to bridge the gap between AI efficiency and human connection? Let’s keep the conversation going.