School AI Pilot Program Examples: What Districts Learned, and How to Run One That Earns Trust

A district leadership team of five adults around a conference table in a school administration office, mid-conversation, one person pointing at a printed timeline on the table, a laptop open but pushed aside.

In February 2025, three months into a year-long pilot of two AI tools, the AI committee at Everett Public Schools in Washington asked for more time before choosing. Forty teachers had accounts. The committee’s planning slide put it in one line: “Any good tech implementation is at least a three-year cycle.”

A pilot is a question a district asks itself. Below are school AI pilot program examples from districts that asked it in public over the last two years, three pilots that paused, and a six-step sequence for how to pilot AI tools in a district so the answer can go into a district AI rollout plan.

School AI pilot program examples from 2024 to 2026

Connecticut’s state pilot, evaluated from the start. In January 2025 the Connecticut State Department of Education launched a pilot of approved AI tools for grades 7 to 12, after the state legislature asked for an investigation of AI tools in those grades. The evaluation, by the University of Connecticut and Southern Connecticut State University, was published in October 2026. It surveyed 735 students and 58 teachers across six districts and found the pilot “generally successful in helping students and teachers develop skills using AI tools.” Teachers found AI most useful for preparing and differentiating materials, and both groups were wary of using it for math. The evaluators’ strongest finding was about adults: “Professional development and peer collaboration for teachers was vital to the success of the pilot.”

Indiana’s grant pilot, and the funding cliff. Indiana’s Department of Education put about $1.8 million of one-time federal relief money into AI platforms in 112 schools across 36 districts for the 2023 to 2024 school year. Of 625 teachers surveyed, about 53 percent rated the platforms’ impact on students as positive. When the grant ended, some districts kept their platforms through a separate state grant and others did not. A state official said at the time: “AI is only going to be successful if teachers are the ones making decisions about how to implement it in the classroom.”

Charlotte-Mecklenburg, listening before launch. Before any districtwide launch, Charlotte-Mecklenburg Schools gathered feedback from more than 10,000 families, students, and staff, then named 30 AI Champion Schools and 60 central office champions to lead peer training. The champion schools were chosen for their “growth mindset and innovative thinking,” so those results come from schools that wanted to go first.

Everett, Washington, the committee that asked for more time. Everett gave 40 teachers year-long accounts in two AI tools in November 2024. The final survey in May heard from 173 students and 10 teachers who had used both tools.

Dayton, Ohio, a chatbot at one alternative school. Dayton Public Schools piloted a student-facing chatbot at one alternative school and told families who could read the chat logs: school employees, an external evaluator, and the vendor’s staff. The evaluator said the pilot “isn’t ready for district-wide rollout,” adding, “I can only get as far as perception.”

Three pilots that paused

Los Angeles Unified launched a family-facing chatbot in 2024 on a five-year, $6 million contract. About three months later the vendor furloughed most of its staff and the district took the tool offline, having paid about $3 million.

Salamanca, New York, paused a classroom robot pilot “while we work through enhanced student data privacy agreements” after teachers, parents, and state education leaders raised concerns.

Portland, Oregon’s school board voted unanimously in June 2026 to pause generative AI contracts until an oversight framework exists, after a parent-led group pushed for it. The resolution requires an inventory within 120 days of every AI tool, contract, cost, and the data shared with vendors. I would start there in any district.

How to pilot AI tools in a district: a sequence that holds

1. Listen before you choose a tool. Wisconsin’s Department of Public Instruction tells administrators to “hold staff and community listening sessions to surface concerns, instructional interests, and practical ideas” before deployment. In the AI Community Playbook labs I facilitate, students, families, educators, and staff are heard in the same process, and what they say can change which problem the pilot takes on.

2. Write the success statement before launch. Digital Promise’s pilot framework asks districts to “co-create a success statement” with benchmarks for learning outcomes and for student, teacher, and family experience. Wichita’s staff told NPR they “haven’t settled on how to define or measure success.” Without a success statement, the results come down to how people felt about the tool.

3. Settle data privacy before anyone logs in. Digital Promise’s guidance is that data privacy agreements should be “firmly established before a tool enters the classroom.” Wisconsin recommends privacy impact assessments for new tools. Salamanca’s pause, above, was about these agreements. If your district needs the checklist, the student data privacy guide on this site walks through it.

4. Choose participants openly, and keep teachers choosing. Nye County, Nevada, made its current pilot optional for teachers, some of whom have ethical objections. The Indiana official quoted above put teacher decision-making at the center. Wisconsin asks for a “human review checkpoint before AI-assisted work is used.” Volunteers show what a tool can do in willing hands, and the results report can say that.

5. Measure small, and say what you measured. Everett’s final teacher survey heard from 10 people and Connecticut’s from 58 teachers and 735 students. Both measured skills and perceptions. None of the pilots here reported a controlled learning outcome.

6. Set the decision date and the money. Nye County’s grant-funded pilot has a February decision written into it. Wisconsin asks districts to review each AI use “at least annually” and decide whether to keep, improve, or restrict it. Digital Promise’s advice is to “consider long-term funding, capacity, and scalability from the initial planning stages.” Name the date and the budget line after the grant, and decide who will take the answer to the board.

From pilot to district AI rollout plan

A pilot is ready to inform a rollout plan when the district can say what it learned and from whom. Everett’s committee called a good implementation a three-year cycle, and Connecticut’s evaluators put teacher professional development at the center. A pilot sits inside that longer stretch. I wrote the sequence out in Building an AI Roadmap for Schools, and the decisions a strategy rests on are in AI Strategy for School Districts.

Families can join that sequence earlier. RAND found that 61 percent of parents think greater AI use will harm students’ critical thinking, against 22 percent of district leaders. That gap is a good reason to bring families in before the tool is chosen.

In Montgomery County, Ohio, between February and April, 72 educators and leaders from 10 districts worked through an AI Community Playbook Lab over three sessions. Of 115 survey responses, 94.3 percent said they were more confident about the work ahead. One participant wrote: “I have never in my career built something like this with community input for policy, and I’m really excited to be able to share with our community that they had input on this.”

Frequently asked questions

What is a school AI pilot program?

A school AI pilot program is a time-limited trial of an AI tool or practice with a defined group of teachers, students, or schools. It is run to answer a specific question before a district decides whether to adopt, change, or drop the tool. The pilots in this guide ran from a semester to a school year. Many set a decision date, and the guidance cited here asks for success measures and privacy agreements first.

How do you pilot AI tools in a school district?

Listen to staff, families, and students about the problem first. Write a success statement with measures for learning and for people’s experience. Sign data privacy agreements before classroom use. Recruit participants openly and keep teachers in charge of classroom decisions. Measure on a small scale and say what you measured. Set a date when the district will decide what happens next, with funding identified beyond the pilot.

How long should a school AI pilot run?

Most of the district and state pilots in this guide ran for one semester to one school year. Indiana’s ran for a school year, and Nye County’s has a decision written in for February. A pilot needs enough time for teachers to build skill with the tool. Connecticut’s evaluators found professional development “vital” to success.

Should families be part of a school AI pilot?

Yes, and earlier than a disclosure notice. Charlotte-Mecklenburg gathered feedback from more than 10,000 families, students, and staff before launch. Dayton told families who could read chat logs. Portland’s board paused generative AI after parents organized. Family input before the tool is chosen gives the pilot a different starting point.


A thought partner for the pilot

Districts can run a good pilot on their own. What they tend to want is someone who has watched other districts run them and who has no stake in which vendor wins. The Community AI Playbook is the listening and alignment work before a pilot. AI Strategic Advisory sits with the superintendent across the whole sequence, including vendor-neutral evaluation.

If your district has a pilot underway, or a yes that now needs a plan, I would welcome the conversation: brett@aeieducation.org, or book a conversation.