What AI Readiness Actually Asks of Your School Board

For almost every school board in the country, AI readiness is currently asking for a policy, not a purchase. Ohio is among the first states to require one: state law there directs every district, community school and STEM school to adopt an artificial intelligence policy no later than July 1, 2026. Most states advise, a growing few require, and Ohio is the clearest and most recent example. In most districts the pressure is loud and the instruction is absent, which is why the item keeps sliding to the next meeting.

The short version: Thirty-four states had issued statewide AI guidance by the end of 2025, and nearly all of it advises rather than requires. That leaves most trustees with real constituent pressure and no compelled action. The board-sized piece hiding inside the AI question is the input layer. Every AI tool a district approves assumes a student can produce text on a keyboard, and the standards assume the same thing well before they measure it: the grade three writing standard already has students using technology to produce writing, while grade four is the earliest grade that sets anything checkable. Setting an elementary keyboarding expectation sits squarely inside board authority, starts with a one-page audit that costs nothing, and produces evidence within one school year.

What is actually being asked of boards right now

The National Association of State Boards of Education published a policy update in February 2026 by program director Winona Hao that puts a number on it: by the end of 2025, 34 states had developed statewide AI guidance. The same piece names Ohio among the first states to require every district to adopt a formal AI use policy, either the state model or a local one aligned with it. Read that as an as-of-2025 picture rather than a permanent national count, because legislatures keep moving on this, and a trustee who says "the only state" from the dais can be corrected on the spot.

Ohio's requirement is worth reading in the original, because it is the clearest picture available of what a legislature thinks a board should do. The Department of Education and Workforce page reproducing the statute says each district, community school and STEM school "shall adopt a policy on the use of artificial intelligence" not later than July 1, 2026, and that a district may adopt the state model verbatim and be compliant. The requirement is procedural. It does not tell a district to teach AI, to use AI, or to buy anything.

The federal layer adds urgency without adding instruction. The White House materials on AI education describe an April 2025 executive order that created a task force on AI education, a Presidential AI Challenge, educator training and public-private partnerships. Earlier, in May 2023, the Department of Education's Office of Educational Technology issued a 71-page report, Artificial Intelligence and the Future of Teaching and Learning, whose recommendations are addressed to policymakers and leaders rather than to any specific governing board.

In most states nobody is going to tell your board what to vote on, and the absence of a mandate is not the absence of exposure.

The assumption every AI plan makes before any standard measures it

Read the student technology standards your district already cites and look for the word keyboarding. It is not there. The current ISTE Standards for Students, marked 2024 on ISTE's own page, describe an Empowered Learner who understands "fundamental concepts of how technology works" and can choose and use current technologies effectively. Operation is presumed. The mechanics of getting words into a machine are never specified.

The one place a typing expectation is written down in plain language is the writing strand of the Common Core, still in force in many states, and it arrives in two very different forms one grade apart. W.3.6 already asks grade three students, with guidance and support from adults, to "use technology to produce and publish writing (using keyboarding skills)". That is an assumption with nothing in it a district can check. California's posted version of the grade four standard, W.4.6, is the first to put a number on it, requiring students to "demonstrate sufficient command of keyboarding skills to type a minimum of one page in a single sitting."

So the sequence a board is actually governing looks like this. AI tools and typed digital work reach classrooms in grades two and three, and typed state testing commonly begins in grade three: California, for one, administers its Smarter Balanced summative assessments by computer from grade three. The standard assumes the skill in grade three, first measures it in grade four, and assigns no one responsibility for teaching it in between. The gap is not that no standard mentions typing. It is that the assumption arrives before the instruction and before anything checkable, and that is the part of AI readiness a board can close.

Why keyboarding passes the three tests a board action has to pass

Trustees are not short of good ideas. They are short of good ideas that survive the three questions a competent board asks of any motion: is this ours to decide, what does it commit us to financially, and when will we know whether it worked.

Possible AI-readiness action	Whose decision	Recurring commitment	Evidence available
Adopt an AI use policy	Board	Staff time to draft and review	Immediate, but procedural only
Set an elementary input-fluency expectation in the program of studies	Board sets it, staff implements	Instructional minutes; any software line quoted in writing first	Within one school year, per student
Purchase an AI instructional platform	Superintendent and purchasing	Multi-year license, often escalating	Two to three years, contested
District-wide AI professional development	Superintendent	Stipends, substitute coverage, calendar days	Indirect, hard to attribute

Only one row belongs to the board, can begin at no cost at all, and returns evidence before the next budget cycle. Keep the lane clean when you write the motion: the board sets the expectation and funds it, the superintendent's staff selects the method and the vendor, and purchasing runs whatever process the district's thresholds require. A motion that names a product has crossed into administration, and someone in the room will say so.

Sequence it so that no trustee is ever asked to vote with a blank fiscal note. Step one is free and board-sized: a one-page instructional audit of who teaches keyboarding in this district, in which grades, for how many minutes a week, and who owns the outcome. It costs nothing, needs no vendor, and is defensible in open session. Step two, only if step one shows a gap, is a costed conversation about school access, with a written per-student quote in the packet before anyone votes on money.

Sample motion language: "That the Board direct the Superintendent to include an explicit keyboarding expectation in the elementary program of studies for grades one through four, to report baseline and end-of-year words-per-minute and accuracy data to the Board by June, and to bring any associated software agreement forward with a written per-student quote attached, under the district's existing procurement and student data privacy procedures."

The strongest argument against doing this

Voice input and multimodal interfaces are improving quickly, and a trustee who says typing may matter less in ten years is making a serious argument, not a lazy one. Part of it lands. For casual consumer use, dictation already works well enough that many adults have stopped typing long messages.

Where it stops landing is the classroom. Twenty-four children dictating in one room is not a workable acoustic environment. Revision, which is where most of the learning in writing happens, is done by hand at a keyboard rather than by re-dictating a paragraph. And the assessments and assignments a district already administers accept typed entry.

One honest limit belongs here as well. No one has published a controlled study showing that keyboarding instruction improves student outcomes with AI tools specifically, and any vendor who tells you otherwise should be asked for the citation. The argument for acting is a prerequisite argument, not an efficacy claim, and a board should record it that way in the minutes.

The risk your board actually carries here

It is not that keyboarding fails. It is that the district buys a tool nobody opens, and a trustee has to explain the line item three years later.

That risk is measurable. An analysis by the University of Pennsylvania's Center for Learning Analytics, covering 48 districts and reported by Government Technology, found a median of 70 percent of student software licenses going completely unused, with a median of 97.6 percent not used to their intended levels. The supply side has not thinned out since. LearnPlatform by Instructure reported that in the 2024-25 school year districts accessed an average of 2,982 distinct edtech tools, close to a nine percent increase over the prior year.

The defense against that is written into the motion, not discovered at renewal.

Checkpoint	Question the board asks	Evidence staff brings
Week 8	Are the accounts actually being used?	Active students as a share of enrolled students
Mid-year	Is anyone moving?	Median words per minute and accuracy, by grade
June	Do we renew, change or stop?	Year-over-year comparison plus teacher time cost

The three questions you will be asked in open session

Does this duplicate something we already have? Frequently, yes. Most districts already own Google Workspace for Education and a rostering system, and some states supply a practice platform. Ask staff to name what a new tool does that the existing stack does not, in one sentence, before the item reaches you.

What does it cost per student? This is where vendors, including the one publishing this piece, get evasive, so here is our own answer without the softening. Meta Typing Club is free for individual learners and families, and an independent teacher on a personal account can run one class of up to five students. Anything larger than that is school use, and school use is commercial: priced on request, with no published per-seat figure, no self-serve district purchase, and a school route that is planned rather than built today. The free personal tier is not a way to start a class of your students around the review this piece is telling you to demand. The moment real students in your schools are on a platform, it is school use and it belongs behind your data privacy agreement first, free or not. Require a written quote before any vote, from us and from every competitor on the list.

What happens to student data? Ask for the data privacy agreement your state or consortium uses, and ask each vendor to sign it. On our own side, the useful and checkable fact is that no age or date of birth is ever collected, and a teacher or parent creates the student account directly, so a pilot can run on class identifiers with the roster mapping kept inside the school. We do not publish a storage region or a retention period, so no board should take a compliance claim from us on trust, and account deletion is handled by email request rather than a self-service button. Those are the questions to put to every vendor in this category, and a vendor that will not answer them in writing has answered them.

Key takeaways for trustees

• 34 states had statewide AI guidance by the end of 2025; most of it advises rather than compels, and only a growing few states require an adopted policy.
• Ohio Revised Code requires an adopted AI policy from every district by July 1, 2026, and permits the state model verbatim.
• The current ISTE student standards presume technology operation and never name keyboarding.
• Common Core W.3.6 already assumes grade three students type; W.4.6 in grade four is the earliest standard that sets a measurable expectation, and typed state testing commonly starts in grade three.
• Keyboarding is one of the few AI-adjacent actions that sits in board authority, starts with a no-cost audit, and reports evidence within a year.
• A median of 70 percent of student software licenses went entirely unused across 48 districts studied, so build a usage checkpoint into the motion.
• No published study links keyboarding instruction to better AI-tool outcomes; the case is a prerequisite case and should be minuted as one.
• Ask every vendor, including ours, for a written quote and a signed data privacy agreement before you vote.

Questions trustees ask

What would this cost our district per student per year?

There is no published per-student price for school use of Meta Typing Club, there is no self-serve district purchase, and the school route is planned rather than built. Free covers individual learners and families; an independent teacher on a personal account can run one class of up to five students. Your district's use is commercial and priced on request, so require a written quote in the board packet rather than a figure from a blog.

Do we have to vote at all, or can the superintendent handle this?

Software selection under the bid threshold is administrative in most districts. Changing what the elementary program of studies expects of students is a board matter. If you want the expectation to survive a change of principal, it belongs in a board action rather than a building decision.

Are we mandating something teachers will resent?

Only if the motion adds minutes without removing any. Direct staff to identify the instructional time this displaces before implementation, and to start with volunteer classrooms. Adoption in the early grades fails in classrooms rather than in offices.

What if the state issues binding AI requirements later?

An input-fluency expectation is compatible with any AI policy a state is likely to write, because it concerns student skill rather than tool use. Ohio's statute is a policy adoption requirement and does not conflict with a curricular expectation.

How would we know within a year whether it worked?

Baseline and end-of-year words per minute and accuracy, reported by grade with participation rates alongside. Meta Typing Club reports words per minute only, with no characters-per-minute option, so the metric your staff brings you will be in those units. Insist on the participation number in the same table, because a strong average from a quarter of the students tells you nothing.

What about student data privacy in the early grades?

Send it to the same review any child-facing tool gets. The specific asks are a signed data privacy agreement on your state or consortium template, a named list of subprocessors, a written retention and deletion period, and confirmation of what is collected from a child account. Score every vendor against those four, ours included.

Is this really an AI item, or are we relabeling a typing purchase?

Fair challenge, and worth stating plainly in the meeting. The AI framing matters because it explains the timing: conversational tools moved typed text production down into grades where nothing in the standards requires it. The action itself is old-fashioned, and that is an argument in its favor rather than against it.

Where a board can start

Ask for one page at the next meeting: which grades in this district currently receive keyboarding instruction, from whom, for how many minutes, and what the June data looks like. In most districts that page cannot be produced, and the silence is the finding. From there, a single motion setting an expectation for grades one through four, with a reporting date attached, gives you something concrete on AI readiness that does not depend on predicting what the technology does next. You can see the structure of a sequenced elementary course on the Meta Typing Club lesson sequence, which runs to more than 2,400 lessons across multiple international languages, and the help documentation shows what teacher-facing reporting looks like before you ask staff to evaluate anything.