Read a sample: 1. The Teacher Stays in Charge
A teacher sits down with a stack of short written responses. The lesson is fresh in mind, the learning goal is clear, and the students would benefit from hearing what they did well and what to try next. The difficulty is not knowing that feedback matters. It is finding enough time to make each comment accurate, useful, and connected to the work.
An AI tool may be able to help with part of that process. It can suggest ways to organize criteria, produce possible comment starters, or draft a response to an example. Those suggestions can give a teacher something to work with instead of a blank page. They can also be wrong, generic, poorly timed, or more confident than the evidence warrants.
The distinction for this book is straightforward: AI may assist with organizing and drafting; the teacher remains responsible for interpreting the work, choosing feedback, and making grading decisions. A polished sentence does not become sound feedback merely because a tool produced it. It becomes usable only after a teacher checks that it fits the assignment, the student’s work, and the purpose of the lesson.
The workflow in this chapter uses a fictional response. It shows how a teacher might make decisions before, during, and after using a tool—and why the most helpful decision is sometimes to reject a suggestion.
Before opening an AI tool, identify the feedback task you actually want to accomplish. “Help me grade these” combines several activities that deserve separate attention. You might need to clarify the criteria, notice patterns across responses, draft a concise comment, or decide how a performance relates to a rubric. Those are not interchangeable jobs.
Suppose the immediate problem is that your comments take too long to phrase. The relevant task is drafting, not evaluation. You may already know that a response contains a clear claim but needs a stronger connection between evidence and reasoning. In that case, you can ask a tool for possible wording without asking it to determine what the student earned.
Or suppose the assignment directions are lengthy, and you want a shorter list of feedback criteria. A tool can propose a structure, but you still need to compare it with the learning goal and remove criteria that do not belong. A neatly organized list is useful only if it represents what you intended to teach and assess.
Starting with a narrow purpose also makes it easier to judge whether the tool helped. If your goal was to produce one clear next step, ask whether the result contains one clear, appropriate next step. If you spend more time correcting the draft than you would have spent writing the comment yourself, a different approach may serve you better.
The examples in this chapter are invented for practice. They do not represent a real student, even when they resemble ordinary classroom work. Fictional material lets a teacher experiment with instructions and review habits without needing to submit a student’s writing.
Real student work requires a separate decision. Before entering any of it into a tool, follow applicable protections and your school’s current rules concerning approved tools, accounts, permissions, and use of student information. Removing a name alone does not necessarily make a response appropriate to share. Details in the writing, the surrounding context, or information combined with it may still identify a student.
If you do not know whether a particular use is permitted, pause and seek guidance through the appropriate school channel. You can still practice with a fictional response or develop general criteria that contain no student information. A convenient drafting process is not a reason to bypass the rules that govern the material.
Consider a teacher who has just taught a short lesson on using evidence in an argument. The class read an invented passage about whether a school garden should include a small pond. For the lesson, the passage explains that a pond could attract insects and birds, but it would also need regular maintenance.
The assignment asks students to take a position and use one detail from the passage to support it. The teacher’s learning goal is not to assess everything that makes an argument effective. It is narrower: students should state a position, select a relevant detail, and explain how that detail supports the position.
The teacher chooses this fictional response for practice:
> The school should add a pond to the garden because it could bring birds. The passage says birds might visit the pond. Birds are interesting to watch.
The response has a recognizable position and a relevant detail. Its final sentence gives a reason the writer values birds, but it does not yet explain clearly why attracting birds supports adding a pond to the school garden. That distinction matters: the teacher is looking for a connection between the passage detail and the position, not merely another sentence about birds.
The teacher starts by translating the lesson goal into three working criteria:
1. States a position about adding the pond.
2. Uses a relevant detail from the passage.
3. Explains how the detail supports the position.
These criteria are deliberately limited. The teacher is not ignoring spelling, paragraph length, or other possible features forever; they are simply not the focus of this feedback task. If a tool suggests additional criteria, the teacher can consider them, but their appearance in a response does not make them part of the assignment.