If you keep seeing advice about helping K–12 educators build practical AI skills, the easy mistake is to think the next step is better prompting. That sounds efficient. It can also send your time, budget, and attention at the wrong problem.
The sharper call is this: the real AI skill gap for K-12 educators is not prompting. It is rubric design. A rubric is the scoring guide that spells out the criteria, performance levels, and feedback standard. If that structure is fuzzy, AI does not fix the problem. It just produces faster, cleaner misalignment.
That is why the workshop result with 25 K-12 teachers matters. AI could draft a rubric and help make vague standards more visible, but many drafts missed the intended grade level or drifted away from the lesson focus. Teachers still had to do major rewrites. The bottleneck was not writing a clever prompt. The bottleneck was giving the model a scoring guide precise enough to follow.
A second study points the same way from the feedback side. When a system used teacher-supplied rubrics, materials, and exemplars, students received more specific writing feedback and teachers saved time. But automated scoring still needed human calibration. That boundary matters: this supports AI for planning and feedback support, not hands-off grading.
A useful update is not the one with the longest feature list. It is the one that changes your next decision. Share this with the teacher or school lead who is starting with prompts instead of the scoring guide: build the rubric before you chase better prompts, and do not mistake rubric support for automatic grading.