Artificial intelligence is moving into schools quickly. In places with available resources, it is arriving through district-approved platforms. In others, it is entering quietly through student phones, classroom tools, tutoring apps, writing supports, and vendor promises about personalization.
The question is no longer whether students will encounter AI. They already are.
The more important question is whether AI will deepen student agency or weaken it.
By student agency, we mean a student’s ability to take ownership of their learning. Their ability to ask questions, make choices, revise ideas, explain reasoning, and see themselves as capable of shaping outcomes. Students with agency are not treated as passive users of technology. They are prepared to think critically, use tools ethically, and understand when to question what a tool produces.
An Intentional Tool, Not Experimental Shortcut
AI can support that kind of learning when it is used as a scaffold. It can help a student brainstorm, translate, organize ideas, receive feedback, practice reading, or access support that may not otherwise be available. For multilingual learners, students with disabilities, rural students, and students in under-resourced schools, those supports could matter.
AI can also weaken autonomy when it becomes an answer machine, a surveillance system, or a substitute for student thought and teacher judgment. If a tool helps students move through assignments faster but does not expand critical thinking, we should ask what kind of learning is actually taking place.
This is especially important for Black, brown, immigrant, low-income, rural, and under-resourced communities. Too often, new educational technologies are introduced first in communities that have the least power to shape them. Families are told that the tool will close gaps, accelerate learning, personalize instruction, or prepare children for the future. But they are not always told who built the tool, what data it collects, how student information will be used, or whether the product has been meaningfully tested with children like theirs.
That is not innovation. That is experimentation without enough consent.
Schools and districts should not adopt AI simply because a company says it supports learning. They should ask harder questions before these tools become part of children’s daily school experience.
Who controls the tool? Who owns the data? How are student responses stored? Can families opt out? Were students, teachers, and communities involved before the product was launched? Has the tool been tested for bias? Does it support students’ ability to question and revise, or does it reward quick answers and compliance?
Our questions are not anti-technology. They are pro-child.
Understanding and Evaluating Artificial Intelligence
AI was not built first as an education tool. Much of it was designed outside classrooms, outside student support systems, and outside the relationships that make learning possible. That matters because learning is not only about information. Learning is about trust, identity, culture, language, belonging, feedback, and human connection.
A student does not become more powerful simply because a tool gives them an answer. A student becomes more powerful when they learn how to question the answer, challenge the source, revise their thinking, and connect knowledge to their own life and community.
That is why AI literacy must be more than teaching students how to use prompts. Students need to understand how AI systems work, where their information comes from, what they get wrong, whose knowledge they may leave out, and how bias can be reproduced through technology. They need to know that AI can sound confident and still be incorrect. They need opportunities to compare, critique, verify, create, and use AI ethically.
In other words, students should not only learn how to use AI. They should learn how to evaluate it.
Stakeholder Voice Must Be Included
This work cannot fall on teachers alone. Educators are already being asked to do more with less. If AI companies want access to classrooms, they should not design tools in isolation and then expect schools to absorb the risks. Teachers, principals, counselors, multilingual learner specialists, special educators, students, families, and community members should be meaningful partners in design, piloting, and evaluation.
Their role should not be symbolic. They should help decide whether a tool actually strengthens inquiry, feedback, revision, creativity, and student voice.
Districts also need to slow down before adopting AI at scale. A tool should be piloted in real classrooms, with real students and educators, before it becomes part of a district’s technology system. The measure of success should not be how many students logged in, how quickly assignments were completed, or how much staff time a vendor claims to save.
The better question is: Did the tool help students think, ask, revise, explain, and communicate more deeply?
Families and communities also deserve a voice. If student data is being collected, families should understand what is being collected and why. If AI is being used to support reading, writing, discipline, assessment, counseling, or intervention, communities should know how those decisions are made. If a tool is being introduced in a school, students should be able to ask what it does, what it cannot do, and how it affects their learning.
AI should not become another system that happens to students without them.
The future of AI in schools should not be decided only by technology companies, consultants, or district procurement offices. It should be shaped by educators, students, families, and communities before the tools shape the classroom.
If schools are serious about student agency, then students and communities must be more than users. They must be co-authors of the future being built around them.
