Steve Swanson built an AI grading tool — and then deleted the feature that would have returned AI-generated grades straight to students. He's a high school engineering teacher who was staring down 150 assignments after a field-trip week, and he built ClassLens because he needed it for his own gradebook. But when he got to the point where he could push one button and have everything graded and sent back automatically, it felt wrong. So he took the button out.

In this Tech Tip Tuesday conversation, Steve makes the case for "human in the loop" — not as a buzzword, but as the actual job of teaching. He also answers a question he pitched to me by email, and it's the right one: what is the single question every teacher should ask before letting an AI tool grade student work?

In this episode, you'll learn:

The 30-second checklist to run before you let AI grade a stack of student work
The one question to ask any vendor about where student data goes — and what "they hold none of your data" actually means
Why writing a strong rubric changes the quality of what AI hands back to you
Whether it's a violation of student trust to send AI-written feedback without disclosing it
The teaching-assistant analogy that reframes the whole AI grading debate

Show notes, resources, and the full transcript: https://www.coolcatteacher.com/e945