It’s 9:47 PM on a Thursday. You’ve got a family newsletter due tomorrow morning, three emails you’re avoiding because you don’t have the energy to word them carefully, and one lesson tomorrow that already needs a differentiated version you haven’t built. This is where AI actually earns its keep — not as the impressive conference demo, but as the thing that quietly gives Sunday night back. Here’s how to use AI as a teacher without touching student data, without district drama, and without losing the parts of the work that matter.
Start here, in the next fifteen minutes
Pick one email you’ve been putting off, plus one weekly parent message whose content is already in your head but hasn’t made it into a Google Doc. Open whatever AI tool your district has approved, or a public tool if none is specified for adult use. For the email, describe the situation to the tool the way you’d describe it to a colleague in the hallway — that this parent tends to hear feedback as criticism, that you’re frustrated, that you need the language to land softly without going soft on the substance. For the parent message, paste your bullet points and ask for a short weekly note in your voice. Then read what comes back, cut what isn’t yours, and keep what is. That’s the whole first move, and you just used AI without a single student’s name in it.
The non-negotiable: don’t put student information into AI tools unless you have a school-approved platform. Names, grades, discipline notes, and IEP details all count. This isn’t a technicality — it’s the line between a productivity tool and a policy problem.
Skill 1: Context building — what to tell AI before you ask
A common first attempt goes something like “Give me a differentiated lesson plan on fractions.” What comes back is generic, because the tool has no way to know your students, your standards, or what you already tried yesterday. Compare that to: “I teach fifth grade in a mixed-ability classroom. Yesterday we introduced equivalent fractions using pattern blocks. About a third of my students got it, a third are getting there, and a third are still confused about what makes fractions equivalent. Give me three activity variations for tomorrow, each targeting one of those groups.” The second prompt gets you something usable because the tool didn’t get smarter — you did.
The rule underneath all of this is that your assumption should be that the tool knows nothing, and what you put in matters because what you put in is the data it’s drawing from. This is the Cognitive layer of the Human Intelligence framework in action: the creativity and critical thinking that make AI more effective in your hands, not less.
Skill 2: Treat the first draft as 80% done
The first draft AI gives you is almost never the one you’ll use, and even after four or five iterations you’re still editing. Khushali Narechania, Director of Learning at aiEDU and a faculty member in the William Jewell College M.S.Ed. in Transformative Teaching & Learning, in partnership with Breathe for Change, described the practice this way:
“It is like the best thought partner ever, if I’m actually just talking to it like it’s a colleague constantly and giving it hard feedback. I don’t like that. Not like that. This I would change like that. Shorten it here. This doesn’t sound like my voice.”
Treat the output as an 80% draft, ask it what you might be missing, give it feedback the way you’d give a colleague feedback, and refuse the answers that don’t sound like you.
Skill 3: The person-to-person test
Before you use AI for anything student-facing, run one check — does this task require direct human interaction? If it does, AI can prepare you for it or debrief you afterward, but it doesn’t replace it. A hard conversation with a struggling student isn’t a task to hand off; writing a note to that student’s family in a voice that actually lands is a different matter entirely.
For more on what AI genuinely can’t replicate in a classroom, and why that changes the case for educators, the research is worth reading before you build this test into your routine.
Four AI moves you can make tomorrow
- The email in a different voice. You know what needs to be said, you just don’t have the energy to say it the way this person needs to hear it — paste your rough draft, tell the tool who it’s for, let it reformat.
- The weekly family message. Bullets in, formatted note out, in your voice; you already have the content, the typing is the friction.
- The one lesson, differentiated. Give it your existing lesson plan, tell it the profiles of learners you’re serving, ask for three variations of a single activity, and keep the ones that fit.
- The instructional coach on curriculum you already own. Paste in a lesson you’re about to teach and ask, “What might I be missing?” — not to replace your judgment, but to surface the angle you were too tired to see.
For elementary teachers whose students aren’t in the tool at all, the same principle applies: sandwich any AI use with clear learning before and after, so students interact with the output rather than the tool itself. Narechania described a second-grade teacher who used AI to generate short two-minute skits their students then rehearsed and performed — the students never touched the tool, but they practiced reading, writing, and performance, and got something the teacher wouldn’t have had time to build otherwise.
See how other educators are navigating AI in today’s classrooms.
The guardrail: protect productive struggle
AI should never replace the hard thinking moments in a lesson, because productive struggle is where learning moves from short-term into long-term memory. Research on desirable difficulties by cognitive scientists Robert A. Bjork and Elizabeth L. Bjork shows that effortful processing during learning produces stronger, more durable retention than passive or errorless instruction (Bjork & Bjork, 2011), and that’s the exact place educators have to reclaim their agency rather than offload it. The purpose test is simple — is the task one where the effort is the point, or one where the effort is the obstacle? Use the tool on the second kind, and protect the first.
This is also what rethinking professional development looks like when you build it around the moves teachers actually need. If you want to go deeper on all three of these skills in a structured program, the AI Teaching & Learning course lives inside the M.S.Ed. — learn more here.
It’s still 9:47 PM on a Sunday, but the newsletter is written, two of the three emails are drafted, and tomorrow’s lesson has three versions instead of one. The tool didn’t teach the class for you — you did that this week, and you’ll do it again tomorrow. It just gave you back the part of the evening that was never supposed to belong to work.
Want to go deeper on any of these moves? Listen to the full conversation between Dr. Ilana Nankin, co-founder and co-CEO of Breathe for Change, and Khushali Narechania on A Work of Heart, the Breathe for Change podcast. They cover the three meta skills, the classroom examples they came from, and the guardrails that keep AI a tool rather than a shortcut.
