Peer review is a popular strategy in writing classrooms, but its results vary widely. Students often hesitate to critique classmates or default to comments like "good job" and "add more details." When the draft is an analysis of All the Light We Cannot See, weak peer feedback leaves writers without direction while the teacher has little time to fill the gap.

AI feedback has its own limits. It can apply a rubric consistently and flag structural issues, but it does not replace the experience of having a real reader respond to an argument. A student who hears that their claim confused a peer learns something different from a student who reads an automated comment about thesis clarity.

Used together, the two forms of feedback strengthen each other. AI provides a reliable baseline aligned with the rubric, and peers contribute the human reaction of a fellow reader. The teacher then supervises the process and steps in where deeper guidance is needed.

A Sequence That Works

A strong sequence begins with the AI feedback, so that students revise the most obvious issues before a classmate reads the draft. This prevents peer reviewers from wasting time on missing topic sentences or unsupported claims that a tool could have flagged. Peers can then focus on higher-level questions such as whether the argument is convincing and whether the analysis feels fresh.

• Students submit a complete draft and receive rubric-based AI comments within a day
• Each writer revises based on the automated feedback before meeting with a partner
• Peers read the revised draft and answer three guided questions about the argument
• Writers summarize what they will change and why in a short revision plan
• The teacher reviews the plans and conferences with students who need more support

Guiding Peers Toward Useful Comments

Peer reviewers need structure to give substantive feedback. Instead of asking them to comment freely, provide questions tied to the rubric, such as "Which quotation best supports the thesis, and which one feels least connected?" Specific questions produce specific answers and reduce the awkwardness of critiquing a friend.

Modeling is equally important. Before the peer session, walk through a sample draft on the board and show what a strong comment looks like compared with a weak one. When students see the difference between "this is confusing" and "I could not tell whether you were arguing about Werner's guilt or his ignorance," their feedback improves noticeably.

What the Teacher Still Needs to Do

Neither peers nor AI can fully replace teacher judgment. A teacher can recognize when a student's interpretation is original but underdeveloped, or when a peer suggestion would steer an essay in a weaker direction. Reviewing the revision plans gives the teacher a window into how students are thinking about their own writing.

The teacher also handles the social and emotional side of feedback. Some students are sensitive to criticism, and an automated comment that lands too harshly might need a personal word of encouragement. Human attention ensures that the process supports growth rather than discouragement.

Measuring Whether It Works

To find out whether the layered approach is helping, compare rubric scores between the first draft and the final essay. If most students improve on thesis and evidence, the feedback is working. If certain skills remain weak, the teacher can adjust the peer prompts or add instruction on those topics.

Student reflections are another useful measure. Asking which feedback was most helpful, and why, reveals whether students value the automated comments, the peer comments, or both. That information helps refine the process in future units.