Subject deep dive · ELA

Open · the honest worry

Do Quick poll: "Since September, how often have you gotten a piece of writing you suspected was AI?" Name the worry out loud, don't dismiss it.

Say "Every student now has a competent ghostwriter in their pocket. We're not going to out-detect it. We're going to teach around it."

Focus · writing instruction that AI can't shortcut

Three concrete moves, each demoed briefly:

• Critique the AI's draft. Have AI write a mediocre paragraph live, then show how you'd have students improve and critique it. This teaches analysis and revision, and it's the same skill whether the draft is theirs or a machine's.
• Make the process visible. Assignments built on drafts, conferencing, in-class writing, and reflection make outsourced final products pointless, because the grade lives in the process, not the polished artifact.
• Voice. Put a student's real sentence next to AI's generic version and name the difference. Students protecting their own voice is a stronger motivator than any rule.

Watch Don't sell AI-detectors. Say plainly they're unreliable and can falsely accuse, especially multilingual students. The strategy is instructional design, not surveillance.

Hands-on · redesign one assignment (laptops open)

Each teacher takes one writing assignment they brought and rebuilds it into an "AI-aware" version, choosing at least one of:

• A critique-the-AI component — students receive an AI-written draft and improve/critique it as part of the task.
• A visible-process component — a required draft, a conference, or an in-class writing portion, plus the disclosure line from September.
• A protected version — convert the highest-stakes piece to an in-class, AI-free assessment.

Do Teachers draft the redesigned assignment in the shared doc. Non-ELA teachers adapt to any writing they assign (lab write-ups, DBQs, reflections). Pairs swap and sharpen.

Watch Push for one finished redesign, not a list of ideas. The win is something they can assign next week.

Share-out + commit

Do Two or three teachers show their redesign. Commitment: run it this month and note what happened. Preview: November is Math, where AI is confidently wrong about numbers.