Use AI for draft feedback, but keep judgment human
AI can support parts of the assessment process, but it should not replace a teacher’s direct evaluation of student work or final judgment. An LLM can assist with repetitive, time-consuming tasks such as checking for grammatical issues, identifying patterns in writing mechanics, or comparing student work against a rubric you have already created. In these cases, the AI serves as a support tool that helps surface information efficiently, not as the grader.
What AI should not do is carry the full weight of assessment. Students often spend significant time and effort on their work, and assigning a grade without personally reading and engaging with it is not fair to the learner. Professional judgment is especially important when evaluating student voice, originality, growth over time, and nuanced thinking that does not align neatly with automated criteria. AI can inform assessment, but responsibility for interpretation and final decisions should remain human.
Always remember that unless you are using an LLM approved for Educational Institutions, use No-PII Prompting—never enter student-identifying details or confidential grades. When collecting assessment results, you must only use official, district-approved systems (like an LMS or SIS) that are contractually bound to protect data, and never store results in the AI chat logs or history. Review all feedback before it is shared with a student.
