In the Lake Washington School District in Redmond, Wash., a veteran English teacher worries that students who turn bullet points into essays aren’t really thinking for themselves.
At an independent school in nearby Kirkland, one technology leader sees opportunity as teachers increasingly use AI to improve their lesson plans and classroom activities.
Across the Seattle region, student leaders say they appreciate the personalized feedback and guidance AI can provide, even as they wonder whether it shortcuts the very struggle that makes learning meaningful.
And south of the city, a math teacher has watched her students more than double their annual growth benchmarks with the support of AI tools developed by a University of Washington team.
Together, these moments capture the mix of promise, unease, and transition defining education as generative AI continues to make its mark on schools.
Nearly three years after the release of ChatGPT, most schools have moved past the impulse to simply ban its use. Instead, students and teachers are rethinking classroom dynamics, and districts are shaping new frameworks, even as researchers continue to push the technology’s limits.
“Nobody really knows how to do this,” conceded Jonathan Briggs, director of technology, innovation, and partnerships at Eastside Prep in Kirkland.
There’s still not enough information about what works, Briggs explained. But with AI boosting human capabilities in quantifiable subjects, like math, perhaps schools should focus more on qualities that are harder to measure, like collaboration, communication, and creativity.
“So much of it,” he said, “is actually about people in the end.”
As AI returns to the classroom for another school year, there are still more questions than answers about how to navigate this new era of technology. But in the midst of this uncertainty, some administrators, teachers, and students are discovering how AI can enhance, rather than replace, education and learning.
A national turning point
This fall marks a turning point in education as the first full school year in which artificial intelligence has officially become a nationwide education priority.
In April, the White House issued Advancing Artificial Intelligence Education for American Youth, an executive order that placed AI at the top of the U.S. education agenda. It established a White House Task Force on AI Education, launched a Presidential AI Challenge to spur innovation, and mandated new funds for teacher training and workforce apprenticeships.
The policy aims to ensure America remains a “global leader in this technological revolution.” But the real story is far more complex. Interviews with educators and administrators across the Seattle region show that many students, teachers, and administrators are still figuring it out as they go.
In the Lake Washington School District, Chad Marsh is entering his 20th year as an English teacher. He worries that AI lets students avoid the struggle that makes learning meaningful.
Instead of developing their own voice, he says, students let the machine think for them, and their writing doesn’t evolve. Although AI’s results might look convincing or be “close enough” to their own ideas, he said, “someone has basically generated your thoughts for you.”
Meanwhile, teachers spend extra time trying to spot AI use with imperfect detection software. Marsh said AI tools sometimes flag students unfairly, making the classroom environment feel “adversarial” — the last thing he wants.
At the same time, he advocates for integrating AI into classrooms rather than ignoring it. “I think the worst thing we can do is pretend that it doesn’t exist and not use it,” he said.
Meanwhile, many teachers are finding benefits in using the technology themselves.
According to a report released by the Gallup and Walton Family Foundation in June, 60% of US teachers used an AI tool during the 2024-25 school year. Weekly users reported saving nearly six hours per week, representing time reinvested into “more personalized instruction, deeper student feedback, and better parent communication.”
As Eastside Prep’s Briggs put it: “The ideas you have as a teacher are no longer limited by the number of hours you have.”
Leading the way in Washington state
Washington state is positioning itself as a national leader in managing new technologies in schools. It was the fifth state to release AI guidance and the first to publish a human-centered framework for its use in education.
For the 2025-2026 school year, the Office of Superintendent of Public Instruction (OSPI) continues to update its guidance, pairing digital citizenship with responsible AI use while working to minimize classroom distraction.
In January 2024, OSPI released its “Human-AI-Human” framework, which stresses that AI should enhance, not replace, human learning. It says students should initiate and interpret AI use, aiming to keep critical thinking and reflection at the core.
At the time, State Superintendent Chris Reykdal explained, “We already know that possibly tens of thousands of students and educators are using AI both in and out of the classroom.” He added: “We now get to put some shape and definition around this usage by embracing it with a human-centered approach.”
The goal, Reykdal wrote, is to elevate “human control and inquiry,” using AI for production but “never as the final thought, product, or paper.”
District-level adoption
After making headlines in 2023 as one of many school districts to ban ChatGPT and other AI tools, Seattle Public Schools (SPS) has shifted course. The district is now framing AI as a resource that can enrich learning when approached thoughtfully. SPS students have access to AI tools supported by new guidance on ethical use, academic integrity, and digital citizenship. SPS did not respond to an interview request.
In Bellevue, a district serving 19,000 students, leaders are incorporating the state’s Human-AI-Human guidance into an evolving strategy aimed at enhancing, rather than replacing, core educational practices. Thomas Duenwald, director of educational technology and assessment, emphasized Bellevue’s focus on the “real power” of AI for formative assessment and personalized learning.
AI tools free teachers from tasks like “planning, grading, assessment,” Duenwald said, allowing them to focus more on relationships and classroom instruction.
The district has chosen not to ban the technology, but instead to model its use transparently. Duenwald is frank: AI “will definitely let you do dumb things, and maybe even worse than dumb things,” but he says this challenge pushes educators to design assignments that machines can’t just shortcut.
Bellevue is also preparing its students for an AI-driven future. Chief of Communications Janine Thorne points to the broader goal: students should be “creators and critical thinkers, and not just consumers of technology.”
Parents are actively included through the district’s Educational Technology Advisory Committee, which addresses concerns about the social-emotional impact of technology and AI.
Student voices in the AI debate
In a May 2025 seminar, students with the Legislative Youth Advisory Council and the Washington Student Engagement Network (WA-SEN) described actively integrating AI into their learning as a supplement, not a substitute. They offered firsthand accounts of AI use, such as:
- Brainstorming, clarifying complex concepts, summarizing topics, and refining their writing.
- Tools like PhotoMath assist with math problems by providing step-by-step solutions.
- Generating practice quizzes for studying.
Overall, students said they appreciated the personalized guidance and immediate feedback AI offers for understanding complex issues.
They also cited challenges:
- AI can be inaccurate, rigid, and repetitive.
- It sometimes “takes away from the beauty of learning” by skipping over the struggle essential for critical thinking, said Paree Raval, a junior at Sammamish High School in the Bellevue School District.
- Inconsistent AI detection software sometimes leads to false positives on human-written essays, fostering distrust between teachers and students.
- AI output “can’t be used as a final version of anything, and it always has to be fact-checked,” said Jordan Verkh-Haskell, a recent graduate of the University of Puget Sound with a degree in computer science and the South Sound Region Coordinator of WA-SEN.
AI as a ‘third agent’ in the classroom
Min Sun is the co-founder and CEO of Colleague AI, and a professor in the College of Education at the University of Washington. With her team at the UW, she offers one example of where things might be headed, seeing AI as a third actor in the classroom, working alongside teachers and students.
Her team’s platform, Colleague AI, helps teachers save time on lesson planning while offering adaptive tools to support students’ individual needs. In one example, English learners used the system to switch seamlessly between Chinese and English during a geography lesson, while teachers monitored their progress in real-time.
For Sun, this responsiveness highlights AI’s potential to make classrooms more inclusive for students with disabilities, multilingual learners, or those behind grade level.
The outcome can be striking. In Federal Way, 7th-grade math teacher Hayley Spohn used Colleague AI to support her students. By midyear, her class had already more than doubled growth benchmarks, averaging 108% toward annual goals. One student, who had shown no growth for two years, advanced from a 4th-grade to a 6th-grade level in just six months.
That promise is drawing interest at Eastside Prep in Kirkland, where teachers are also using Colleague AI to experiment with lesson plans and simulations. One history teacher, for example, used AI to generate governance scenarios for the state of Washington — a complex lesson plan that might otherwise have been too time-consuming.
To balance innovation and caution, Eastside Prep has built a culture of transparency around AI use. Faculty work groups, listening sessions with students, and parent focus groups help to set these norms.
The school has a rule of thumb, Briggs explained: if students feel excited to tell a teacher about how they used AI, it was probably fine; if they feel the need to hide it, it probably wasn’t.
What matters is intent. Used well, AI can sharpen inquiry or make lessons more inclusive. Used poorly, it shortcuts the struggle that makes learning meaningful.
Addressing inequality is another advantage. Access to reliable internet and devices remains unequal, so Sun argues that banning AI would worsen the divide. She sees integration as essential to ensuring equal access, complemented by infrastructure investment and free or subsidized tools such as Colleague AI.
Looking ahead, she imagines classrooms where teachers, students, and AI collaborate as “three agents.” The form of that AI, she concedes, is still unknown, whether embedded in software or something more tangible. What is certain is that the core challenge will be determining which skills matter most in an AI-driven world.
In the end, Sun balances her optimism with caution. She worries about over-reliance, both from teachers who might let AI shape too much of their work as educators, and from students tempted to outsource their thinking.
The key is balance: AI, she said, shouldn’t do the work for them. It should work with them.
