The missing piece in AI grading: the task sheet
Most AI graders mark essays in a vacuum. Grading against the actual task you set is what makes feedback fair — and trustworthy.
A common failure of AI grading is that it marks writing in isolation. It can tell you an essay is well-structured and fluent — while completely missing that it never answered the question you actually asked.
Good writing isn't the same as a good answer
A polished essay that ignores the stimulus, the source material, or the specific prompt should not score well. Marking without the task sheet rewards fluency over relevance, which is the opposite of what most rubrics intend.
Context is what makes a grade defensible
When the model can see the task and stimulus you set, its feedback can point out where a student drifted off-brief, misused a source, or answered a different question. That is feedback a teacher can stand behind in a parent meeting.
Keep the teacher in the loop
Even with the right context, AI should draft, not decide. The teacher reviews the highlights, adjusts anything that needs a human eye, and releases the result. Context plus oversight is what turns AI grading from a novelty into a tool you can trust.
CoMarker grades against the actual task and stimulus you upload, so feedback reflects what you asked for — and you always have the final say.