A leadership guide to balancing AI with human intelligence in K-12 schools
How K-12 principals balance AI with human intelligence: where AI helps, where people must decide, decision rights, staffing, and a 2026 checklist.
Updated with current data and sources. Originally published October 7, 2025.

Balancing AI with human intelligence in a K-12 school means deciding, in writing, which work the software may do and which decisions stay with a person. Most schools have not made that decision yet: 79% of districts now report AI guidelines (CoSN, 2026), but only 18% of teachers say they have received formal guidance on how to use AI (Gallup, 2026). This guide is for the principal or superintendent who has to close that gap in 2026.
The temptation is to treat balance as a dial: more AI means less humanity, so turn it down. That is fear dressed up as strategy. The schools getting this right treat balance as a set of decision rights. AI drafts, sorts, and flags. A named person approves anything that touches a child, a colleague, or a budget. The rest of this guide shows where AI helps, where humans must lead, who owns what, and what to check before the end of the school year.
- Decision rights
- A written register that names, for each use of AI in the school, what the tool may do on its own, what it may only draft or recommend, who approves the result, and how often the arrangement is reviewed. It is the leadership tool that turns a vague AI policy into daily practice.
Why balance is a leadership problem, not a technology problem
The tools arrived before the decisions. CoSN's 2026 survey of 607 district technology leaders in 44 states found 88% of districts have some AI initiative underway, while 19% say their overall approach to AI is still undefined. Six in ten teachers used AI for their work in the 2024-25 school year, according to Gallup, and 32% use it at least weekly.
Now look at what those teachers were told. In Gallup's 2026 survey of 2,069 public school teachers, 18% had received formal guidance from administrators on how AI should be used, 48% had received only informal guidance, and 34% had received none at all. Fewer than one in ten had formal guidance on any specific task.
That is the leadership problem in one picture: the district has guidelines, the teacher has a tool, and nothing connects the two. Writing a policy is not the same as leading people through a change. The policy says what is allowed. Decision rights say who decides, and that is what a teacher needs on a Tuesday afternoon with a stack of essays.
Where AI helps a school right now
Be specific about the upside, because your teachers already know it. Gallup found that teachers who use AI weekly estimate they save 5.9 hours a week, roughly six weeks over a school year. The most common uses are preparing to teach (37% of teachers at least monthly), making worksheets and activities (33%), and modifying materials for student needs (28%). Majorities of users say the work is better, from 57% for grading and feedback to 74% for administrative work.
The office is moving even faster than the classroom. CoSN found 64% of districts now use AI in operations, up from 37% a year earlier, and 41% have AI initiatives inside instructional platforms. Ninety-six percent of technology leaders believe AI could benefit education. None of this is the problem. The problem is doing it without deciding who checks the output.
AI in schools, 2025 to 2026
Where humans must lead: the decision rights register
Here is the register I ask leadership teams to write before they buy or approve another tool. It fits on one page. The rows below are the ones every K-12 school has, whether or not anyone has written them down.
| Decision | AI may | AI may not | Owner who approves |
|---|---|---|---|
| Grades and feedback | Propose a rubric score, draft comments | Enter a final grade or return feedback unsigned | The teacher of record |
| Placement, intervention, discipline | Flag patterns in attendance, grades, behavior | Assign a student to a program or consequence | Counselor or principal, with the family |
| Special education and 504 decisions | Summarize records, draft meeting notes | Determine eligibility or write the plan | The IEP or 504 team |
| Hiring, evaluation, staffing | Screen for required credentials, draft schedules | Rank candidates or score a teacher | Principal and HR |
| Family communication | Draft the message, translate it | Send anything a person has not read | The sender, by name |
| Tools and student data | Nothing until approved | Receive student data without a signed agreement | Technology director, with legal review |
| Student well-being flags | Surface a concern from monitoring tools | Respond to the student | Counselor, same day |
Two outside anchors make this register easier to defend to a board. The U.S. Department of Education's July 2025 letter to grantees asks that AI use be educator-led, ethical, accessible, and transparent, and that it protect student data. Ohio's December 2025 model policy, which every district in the state had to adopt or adapt by July 1, 2026, covers the same ground: appropriate use by students and staff, academic integrity, data privacy, curriculum integration, and review of third-party tools. The TeachAI toolkit, the most widely borrowed framework, states the principle behind every row: AI-supported decisions must allow a person to intervene and must rest on human approval.
Balance is not a feeling about technology. It is a page that says who decides.Ernie Delgado, BeyondK12
What the perception gap tells a leader
RAND's 2025 national survey put a number on something principals feel in parent meetings. Asked whether more AI use could harm students' critical thinking, 61% of parents said yes, along with 55% of high school students and 48% of middle school students. Among district leaders, 22% agreed.
Who worries that more AI could harm critical thinking
Leaders are the most optimistic group in the building, by a wide margin. That is not a reason to slow down. It is a reason to check your own read of the room before you present an AI plan to families, and to put the human-led decisions in the first slide rather than the last.
50%
of students agree that using AI in class makes them feel less connected to their teacher.
Center for Democracy and Technology, 2025
57%
of teachers think weekly student AI use would reduce students' independent thinking; 52% say critical thinking.
Gallup and Walton Family Foundation, 2025
59%
of teens say students at their school use AI to cheat at least somewhat often.
Pew Research Center, 2026
1 in 10
teachers has been trained on what to do when a student's AI use may be harming their well-being.
Center for Democracy and Technology, 2025
Each of those numbers is an argument for the register above. Connection, independent thinking, integrity, and well-being are what a person in the room protects and a tool cannot. When the policy names a person for each, parents hear a school that has thought about their child.
Staffing the balance: who owns what
Balance fails on staffing before it fails on philosophy. CoSN found 58% of districts are understaffed for the instructional use of technology, 65% name cybersecurity staffing and the lack of a dedicated budget as the top barriers, and 52% cite a lack of staff training or expertise. The most common AI initiative in districts, reported by seven in ten technology leaders, is training staff on generative AI tools. It is also the one teachers say is not reaching them.
Teacher training on generative AI
EdWeek Research Center's winter 2026 survey found 42% of teachers had received no training on generative AI, down from 50% in fall 2025 and almost 60% in fall 2024. Only 22% had attended more than one session, and 9% described their training as ongoing. Gallup's 2025 survey found 68% of teachers got no AI training in 2024-25 and roughly half taught themselves.
You do not need a new department. You need four names on the register:
- An AI lead, usually the curriculum or instruction director, who owns the decision rights register and the semester review.
- A data steward, usually the technology director, who approves every tool and signs every data agreement before a student logs in.
- A teacher leader per grade band who runs the training, collects the questions, and reports what is working.
- A board liaison who brings the review to the board twice a year, so AI is a standing item and not a surprise.
In a school of 300 students, two of these are the same person. That is fine. What is not fine is a policy with no names in it. Professional development for these roles is a normal use of existing Title II-A funds in many districts; our professional development page describes the sessions we run for leadership teams and teachers, and the funding guide explains how schools pay for them.
How students learn the balance
Students are already running their own experiment. Common Sense Media's 2026 survey of 1,017 teens found 70% use AI for schoolwork. Only 44% said a teacher had discussed when AI should and should not be used, 27% had a teacher explain what AI is and how it works, and 26% had a teacher discuss how to judge whether AI information is accurate. Thirty-five percent did not understand their school's AI rules or were unsure what they were, and 28% said the rules varied substantially from teacher to teacher.
RAND found the same gap from the leadership side: by spring 2025 only 35% of district leaders said their schools give students any training on how to use AI. So the student half of the balance is mostly missing. The fix is a literacy sequence, not a ban. Teach what the tool is, when it is allowed, how to check it, how to disclose it, and how to explain the work without it.
That last skill is the one we build the TechPlus AI curriculum around. Every project in TK-8 ends with the student explaining what they made, which tools helped, and what they would change. A student who can explain the work owns it, and a teacher who hears the explanation has the human check that no detector can provide.
The 2026 checklist for principals
Five moves, in order. A leadership team that meets twice a month can finish them in one semester.
- Write the decision rights register. Use the table above. Fill in the owner column with real names, not titles. Have the board adopt it as an appendix to the acceptable use policy so it carries the same weight.
- Inventory every AI tool already in use. Ask each teacher and office to list what they use and for what. Check each against the data steward's approval list. Anything receiving student data without a signed agreement stops until it has one.
- Put training on the calendar with dates. One session for everyone on the register and the rules, then grade-band sessions on the tasks teachers actually do: preparing to teach, modifying materials, drafting feedback. Count attendance and report it.
- Sequence student AI literacy by grade. What AI is, when it is allowed, how to verify it, how to disclose it, and how to explain your own work. Make the explanation the assessment.
- Review once a semester and tell families. Pull five AI-assisted decisions from the register and check them: was the owner involved, was the output corrected, did anyone complain? Send families a one-page summary. Our 90-Day AI Readiness Accelerator walks a leadership team through all five moves with templates your board can adopt.
Where to start this week
Balancing AI with human intelligence is not a mood you set or a product you buy. It is a page that says who decides, a calendar that says when people are trained, and a review that proves both are real. Write the page first. Everything else in this guide follows from it.
If you want to know how far your school is from that page, the free AI Readiness Assessment takes five minutes. Eight questions give you a score out of 24 across Policy, People, Curriculum, Guardrails, and Funding, and show which of the five moves to make first. The report is yours whether or not you ever call us.
Related reading: Hybrid intelligence in schools: why AI alone will not save education and Why school leaders must consider human in the loop AI for future-ready strategies.
Questions school leaders ask
What does balancing AI with human intelligence mean for a school?
It means the school has decided, in writing, which tasks AI may do and which decisions a person must make. AI drafts, sorts, and flags. A teacher, counselor, or principal reviews and decides anything that touches a grade, a placement, a discipline record, a hire, or a family. TeachAI's toolkit puts it as a principle: AI-supported decisions must allow human intervention and rest on human approval.
Where does AI help a K-12 school the most right now?
In drafting and operations. Gallup found teachers who use AI weekly save about 5.9 hours a week, roughly six weeks a year, mostly on preparing to teach, making worksheets, and modifying materials. CoSN found 64% of districts now use AI in operations such as scheduling and communications, up from 37% a year earlier.
What should AI never decide on its own in a school?
Final grades, placement in programs or interventions, discipline, special education decisions, hiring and evaluation, and any response to a student well-being concern. Federal guidance from July 2025 asks that AI use be educator-led and transparent, and Ohio's 2026 model policy requires districts to spell out appropriate use, academic integrity, data privacy, and vendor review.
How many staff does a school need to run AI well?
Fewer than you fear, but every role needs a name. One AI lead who owns the decision rights register, one data steward who approves tools and data agreements, a teacher leader per grade band, and a board liaison. In a small school two of these are the same person. CoSN found 58% of districts are understaffed for classroom technology support, so write the roles into existing jobs rather than waiting for a new hire.
How do we know whether our teachers are ready?
Ask them, and look at the training calendar. EdWeek Research Center found 42% of teachers had no generative AI training in early 2026, and only 9% had ongoing training. Gallup found 34% of teachers get no guidance at all on AI. If your school has a policy email but no training dates, you are in the majority and not ready.
Where does a principal start?
With the free AI Readiness Assessment. Eight questions give you a score out of 24 across Policy, People, Curriculum, Guardrails, and Funding, and show which of the five moves in this guide to make first.
Sources
- Gallup and Walton Family Foundation, most teachers receive no formal guidance on AI (May 2026) (opens in a new tab) (checked 2026-11-05)
- Gallup and Walton Family Foundation, three in ten teachers use AI weekly, saving six weeks a year (June 2025) (opens in a new tab) (checked 2026-11-05)
- The 74, survey: 60% of teachers used AI this year (Gallup and Walton Family Foundation, July 2025) (opens in a new tab) (checked 2026-11-05)
- CoSN, U.S. State of EdTech District Leadership 2026 (opens in a new tab) (checked 2026-11-05)
- GovTech, CoSN report: cybersecurity is top concern, AI guardrails needed (May 2026) (opens in a new tab) (checked 2026-11-05)
- EdSurge, report: school IT officials worried about AI adoption, cybersecurity (June 2026) (opens in a new tab) (checked 2026-11-05)
- RAND, AI use in schools is quickly increasing but guidance lags behind (2025, via eSchool News) (opens in a new tab) (checked 2026-11-05)
- RAND, AI use in schools growing as policies lag (October 2025, via KMPH) (opens in a new tab) (checked 2026-11-05)
- EdWeek Research Center, more schools are providing AI training for teachers, is it any good? (May 2026) (opens in a new tab) (checked 2026-11-05)
- EdWeek Research Center, teacher AI training is rising fast but still has a long way to go (November 2025) (opens in a new tab) (checked 2026-11-05)
- Center for Democracy and Technology, Hand in Hand survey (October 2025) (opens in a new tab) (checked 2026-11-05)
- Pew Research Center, how teens use and view AI (February 2026) (opens in a new tab) (checked 2026-11-05)
- Common Sense Media, Teens in the AI Era: Schoolwork and Skills That Matter (2026, via EdTech Innovation Hub) (opens in a new tab) (checked 2026-11-05)
- U.S. Department of Education guidance on AI and federal grant funds (July 2025, via THE Journal) (opens in a new tab) (checked 2026-11-05)
- Ohio Department of Education and Workforce, model AI policy for schools (December 2025) (opens in a new tab) (checked 2026-11-05)
- TeachAI, AI Guidance for Schools Toolkit principles (opens in a new tab) (checked 2026-11-05)

See where your school stands on AI, in five minutes
Eight questions. A score out of 24 across Policy, People, Curriculum, Guardrails, and Funding, with the three fixes that move it most. The report is yours whether or not you ever call us.