Research and classroom experience agree that feedback on drafts improves writing more than comments on finished work. The trouble is that reading two drafts of every essay doubles the grading load, so many teachers skip the draft stage or give only a quick check. Poetry essays on Keats are especially good candidates for draft feedback, since students often misread an early claim and build an entire essay on a shaky foundation. AI-assisted feedback offers a way to make draft review realistic.

The purpose of formative feedback is to help a student improve, not to assign a grade. That means comments should point toward the next draft rather than justify a score. Focus on the biggest opportunities, such as sharpening the thesis, adding commentary after quotations, or reorganizing paragraphs around ideas instead of line order.

An AI tool can produce a first pass of this feedback quickly by comparing the draft against your rubric. It can note whether the thesis makes an arguable claim, whether quotations are explained, and whether the essay addresses the poem's form. This gives students immediate guidance while you reserve your time for the conversations and judgments that matter most.

Set boundaries so feedback stays useful

The value of formative feedback depends on how it is framed. Tell students that the comments are meant to guide revision, that the final grade will reflect the improved draft, and that they are responsible for deciding which suggestions to apply. Students who understand this treat the feedback as a resource rather than as an answer key.

• Limit draft feedback to three or four priorities per essay
• Focus on thesis, evidence, and analysis before sentence-level issues
• Ask students to write a short revision plan after receiving comments
• Review a sample of AI-drafted comments before releasing them to the class
• Keep the final grade and final comments under the teacher's control

Review comments before students see them

No automated comment should reach a student without a teacher's review, at least at first. Spot-checking a sample tells you whether the tool applies your rubric accurately and whether its tone suits your students. When you find comments that miss the mark, adjust your rubric language or edit the comments directly.

Over time, you will develop a sense of which types of comments are reliable and which need more attention. Structural feedback on thesis and evidence tends to be dependable, while comments on subtle interpretive choices may need your judgment. Knowing these strengths and limits lets you use the tool wisely.

Build revision into the assignment

Feedback is worthless if students do not revise. Building a revision requirement into the assignment, with a specific deadline and a grade component, ensures that comments get used. Asking students to submit a brief note explaining what they changed and why also encourages reflection.

You can compare drafts to see whether the feedback led to real improvement. If students consistently ignore a certain type of comment, it may be unclear or too difficult to act on. Adjusting your approach based on this evidence makes the feedback cycle more effective each time.

Keep the human relationship at the center

Students still want to know that their teacher cares about their ideas. A brief personal note on a draft, mentioning something interesting the student said about the poem, goes a long way. Automated feedback can handle the routine observations, but it cannot replace that sense of being read by someone who is paying attention.

Used thoughtfully, AI feedback expands the number of students who receive meaningful draft comments. Teachers gain time, students gain guidance, and the final essays improve as a result. The key is to treat the technology as an assistant that serves your teaching goals rather than a substitute for them.