A Director of AI for Your School Is a Policy Hire Before It Is a Technology Hire

Why Your Director of AI Hire Starts With a Grade Dispute

A parent emails on a Tuesday. Their daughter's essay came back with a zero because a teacher ran it through a detector, and the family wants to see the policy that authorized the grade. Nobody in the building can produce one. That is the job you are hiring for: someone who has already written the answer and can defend it to a school board.

So screen for artifacts rather than enthusiasm. A real candidate arrives carrying a dated acceptable-use rule with a version number, a vendor review that ended in a no, and the name of the department chair who fought them and later signed on. A performed candidate arrives with a slide deck of tools. The gap the hire has to close is already measurable in your sector: 53 percent of district recruiters told the EdWeek Research Center they use AI tools, while roughly one district in ten has set any policy on AI use at all ¹.

One tell should end the interview early. A candidate who offers a detector as the enforcement backbone is promising a certainty the tool cannot give, and the policy built on it collapses the first time a family asks how the conclusion was reached. Listen instead for how they would redesign an assignment so that authorship stops carrying all the weight: drafts collected in sequence, writing done in class, a five-minute oral defense. That answer is harder to give and much harder to fake.

Three traits show up in every good one. They write, because policy is prose and someone who cannot draft a clear two-page rule cannot hold this job. They convene, because the work is mostly meetings with people who disagree and the director has no line authority over a single faculty member. And they can say no to a vendor in a room where the superintendent has already said yes.

Which Backgrounds Produce a Director of AI Who Can Face a School Board?

Most strong candidates come from inside education rather than from technology. Instructional technology directors, curriculum and instruction leaders, academic integrity officers, and directors of teaching and learning centers spend their days doing the two things this role requires: writing rules for adults who did not ask for them, and getting those rules adopted without formal authority over anyone.

The unexpected backgrounds are worth a first call. Research librarians have taught source evaluation, citation practice and licensing for decades, which turns out to be most of what AI literacy actually is. Special education compliance leads read federal rules for a living and know exactly what a documented process has to look like when it gets challenged. Registrars and institutional research staff already own student privacy agreements and data governance. Assessment directors know, from long experience, why a number attached to a student invites a dispute.

Candidates from industry can work, but check the sector translation before you fall in love. The chief AI officer pattern in companies assumes budget authority and a chain of command no district cabinet will hand over. Ask an industry finalist to explain shared governance, or how a collective bargaining agreement constrains a mandatory training day. A vague answer means the first semester gets spent learning that in public, at your expense.

Expect thin résumé lines and read them fairly. Johns Hopkins' School of Education now publishes a career path for AI leadership in schools, and its framing puts the emergence of the title after 2022 ². That is one school's account on a page that did not open to an automated check, so weigh it as a signal rather than as a date, and lean on the part your own applicant pool will confirm within a week of posting. Nobody has ten years in this job. What a serious candidate has instead is two or three years of doing it under a different title, usually without the mandate.

Ask the Director of AI Candidate How They Learned to Distrust a Model

The people who are good at this got good by using the tools on their own work long enough to be embarrassed by them. Ask it directly: what did you build with a model, what did you throw away, and what were you wrong about in public? A candidate who has only read about AI answers in principles. A candidate who has practiced answers with a specific Tuesday.

Good answers have a recognizable shape. The person built something real, such as a lesson sequence, a first draft of a board memo, or a set of rubrics. They noticed the model producing a confident citation that did not exist. They went and checked it against a source outside the conversation, and they changed how they work as a result: state the constraint before generating anything, ask for the reasoning and not just the output, keep the judgment that should not be delegated.

Then make them do it rather than describe it. A live exercise beats a portfolio: hand the candidate a real complaint from your inbox with names removed, an AI assistant, and forty minutes to draft both the reply and the policy paragraph standing behind it. Watch where they verify, where they accept a claim without checking, and what they refuse to hand to the model at all. Assessing how someone works with AI in the open tells you more than any question about their philosophy of it, and the same exercise belongs in front of your AI governance lead candidates.

One caution about the exercise. Score the moments, not the artifact. The finished memo mostly measures typing speed; the useful evidence is the point where the candidate stopped, doubted a sentence, and went to look.

Find Your Director of AI Where District AI Policy Already Gets Written

They are mostly not on job boards, because they still hold a different title. Look where this policy work already happens in public: state educational technology director associations, EDUCAUSE and CoSN member communities, ISTE and SXSW EDU session lists, and the acceptable-use policies that districts publish on their own sites. A district that published a good one has an author, and that author is a candidate.

Adjacent roles feed this one reliably. Instructional technology director, director of digital learning, academic integrity officer, and the head of a university teaching and learning center all sit one step away. So do the learning leads inside educational technology vendors and testing organizations, who have watched fifty institutions make the same three mistakes and can tell you which ones they are. University AI institutes and the certificate programs that have appeared since 2022 produce candidates who arrive with a written point of view ².

Closing them is about mandate rather than money. What they will raise, roughly in this order: who they report to, whether they can stop a procurement, whether training carries its own budget line, and whether the superintendent or provost will stand behind an unpopular rule in front of a board. What kills the offer: a reporting line buried inside IT with no academic standing, a title with no procurement authority, and any hint that the real assignment is policing students. Anyone worth hiring has turned that version of the job down at least once, and will recognize it in your first phone screen.

What Does a Director of AI Cost, and Should the Role Sit on Campus?

No published salary series covers this title as of mid-2026. The federal wage survey reports established administrator occupations rather than this one, and the salary aggregators do not yet carry a role page with enough postings behind it to quote honestly. Anyone handing you a confident national number for a Director of AI in education is extrapolating from something, and the useful question is what.

So price it against a band you already publish. Public districts post salary schedules and public universities post administrator ranges, which makes the honest anchor the band this role genuinely sits in at your institution: a district director or assistant superintendent line in K-12, an associate provost or assistant vice president line in higher education. Pull the schedules of three comparable institutions in your state, since they are public records, and hire inside that band. Budget for the top of it. The pool is small, and the corporate AI policy manager market is bidding for several of the same people.

The work is mostly on site, and that is a constraint rather than a preference. Faculty training happens in rooms. Cabinet meetings and board meetings happen in person, and the vendor conversation that decides a contract usually happens in a hallway after the demo. Policy drafting, vendor review and reading travel fine, so a hybrid arrangement of two or three days on campus works well, with heavier presence in the weeks before a term opens.

Multi-campus systems should budget travel instead of describing the role as remote. A fully remote Director of AI can write a perfectly good policy and will struggle to get it adopted, because adoption in an institution runs on trust that accumulates in person.

See what gets scored >