I Let My Staff Vote on AI and Eight Hands Stayed Down

The quietest voices mattered most.

The names are gone. The vote count is not. Fifty teachers packed into a media center four days before the first bell and decided, by paper ballot, whether artificial intelligence would have a place in next year’s classrooms. I called that vote. I have not stopped thinking about the eight hands that never went up.

Tuesday, 3:45 p.m.

Four days before the first bell, I stood in front of my staff with a slide deck I’d built over the summer and a decision I could not put off any longer. The district had cleared an AI writing and reading platform for building-level adoption. Not required. Optional, if the principal wanted it and the staff would use it.

I wanted it. I had spent June and July reading the research, sitting through two vendor demos, and running the tool myself against a stack of last year’s essays. I believed it could close a gap I had watched my at-risk readers carry for years.

I also knew I could not walk in and tell 50 teachers to trust a new tool most of them had never touched.

The pitch.

I laid out what the platform could do. Draft feedback in seconds instead of days. Reading passages that adjusted to a student’s level while she read them. A dashboard flagging which kids were stalling out weeks before report cards made it official.

I kept my voice even. I didn’t oversell it. I have sat through enough vendor pitches to know what oversell sounds like, and Idid not want to be the guy doing it to my own staff.

I told them the district was not mandating anything and told them I would not either. Then I told them the part that mattered most to me, which is that three of our lowest readers had gained almost a full grade level on the pilot data the vendor sent over, and I could not sit on that number and pretend I hadn’t seen it.

The room did not stay even for long.

The room.

Sharon stood before I finished my last slide. Twenty-two years in that building, English before laptops were standard issue. She said what I suspect half the room was thinking and the other half was afraid to say out loud. If a machine drafts the feedback, what happens to the conversation she has with a kid in the hallway about a sentence that almost worked?

She was not against technology. She was against losing the part of teaching that has made her stay twenty-two years.

A teacher two years into the job said the opposite. She teaches the students furthest behind grade level, and she told the room she was drowning by October every year. A tool giving her real feedback on twenty-eight essays overnight was not a threat to her craft. It was oxygen!

A third teacher, somewhere in the middle of her career, asked the question I had been avoiding all summer. What happens to the kid who leans on the tool instead of learning to think through a hard paragraph himself? Nobody in that room had a clean answer, including me. I said something about guardrails and teacher oversight. It sounded much thinner out loud than it had on my slide.

Both were right.

The count.

I had a choice. Mandate the platform and own every consequence myself. Or put it to the staff and live with a result I could not fully control. I chose the vote. Paper ballots, no names, folded into a manila folder on the back table. I told them I would abide by the majority either way.

Forty-two yes. Eight no.

I read the count out loud once, then again, because the room had gone quiet enough that I was not sure they had heard me the first time. Sharon did not vote yes. Neither did four others I had expected to see on that side.

Consensus is not the same as comfort.

After.

I told the staff the platform would roll out in September for any teacher who wanted it, and that no one would be required to touch it this year. That part was true. It is also true that a building does not stay divided in half forever. The eight who voted no will watch forty-two colleagues save an hour a night, and most of them will ask for access before Thanksgiving, not because they changed their mind about the machine, but because nobody wants to be the last one grading essays by hand at eleven at night.

I told the staff this was a positive step for kids who are behind, and the research backs that up. I believe it. I did not tell them the other thing I was thinking, which is that thirty-seven affirmative votes do not erase ten people’s fear.

I found Sharon in her room the next morning, unpacking a bin of paperbacks for the shelf by her window. I did not ask her to change her vote, but asked her what would need to be true in October for her to feel like this had not cost her something. She thought about it longer than I expected, then said she wanted to see the tool’s feedback before a single student did, every time, for the whole first semester. I told her that was fair. I wrote it down.

That is the piece I almost missed. A vote gives you a number. It does not give you a plan for the people who lost.

I wasn’t thrilled when I saw the count. I’m keeping my fingers crossed that feeling will hit me soon.

THE TRANSFERABLE PRINCIPLE

A vote settles the decision. It does not settle the people. When a school leader brings AI or any new technology into the building, the temptation is to treat a majority as proof that the fear was unfounded and move on. It was not unfounded. It was simply outvoted. The eight who said no are still teaching your students tomorrow, and the trust they place in the building depends on whether the leader keeps checking in with them after the vote, not just before it. The systemic risk is not the tool. The systemic risk is a staff that learns dissent gets a hearing and then gets forgotten. Build the follow-up into the rollout itself: name the concerns out loud in the next staff meeting, track what the no-voters said would go wrong, and report back on whether it did. A school leader who does that turns a divided vote into an ongoing conversation, and an ongoing conversation is the only thing that keeps a vote split from hardening into two very different staffs sharing one building.

If you have run a staff vote that split closer than you expected, tell me in the comments how you handled the people on the losing side. If you have not faced this yet, send this to one department chair who will be building the agenda for that meeting soon.

The Principal’s Chair lands twice a week: real dilemmas from twenty-five years in the seat, told the way they happened. If a colleague forwarded this to you, subscribe below.

Until next time — the chair is yours.

theprincipalschair.substack.com