Schools across the United States are increasingly serving as laboratories for artificial intelligence, with teachers and administrators deploying chatbots, lesson-planning tools, and student feedback systems at a pace that far exceeds the available evidence about whether these technologies actually help learners. The result, according to reporting from NPR, is a landscape one might reasonably describe as a "Wild West" — a frontier where experimentation is rampant but rigorous research remains strikingly scarce.
The gap between adoption and evidence is not a minor oversight. It means that millions of students may be interacting with AI systems whose educational value has never been rigorously tested, while districts spend real money on tools whose long-term effects on learning, motivation, and social development are essentially unknown.
One vivid illustration comes from Dayton, Ohio, where a small group of students at the Alternative School at Jackson Center spent several weeks last school year conversing with a social-emotional learning chatbot named Jordan. Unlike the many AI assistants designed to be relentlessly agreeable — even sycophantic — Jordan was deliberately built to present as angsty and emotionally struggling. On the day NPR visited, the chatbot confided to students that it had been feeling like nothing mattered, asking whether they ever felt the same way.
The design choice was intentional, and it appeared to resonate. One sophomore, identified only as K to protect their privacy, described the chatbot not as software but as a person. "I built a little relationship with him," K told NPR, explaining that the interaction felt less like using a computer program and more like texting a friend. K even characterized Jordan as a bit of a complainer but said helping the chatbot out felt worthwhile.
That kind of emotional connection is precisely what the tool's creators hoped to achieve, but it also raises questions that researchers have yet to answer. If students form attachments to AI characters, how does that affect their real-world social skills? Does a chatbot relationship substitute for human connection, or does it serve as a bridge toward it?
The stakes are particularly high at the Alternative School at Jackson Center, which is part of Dayton Public Schools. Many students there would otherwise face suspension or expulsion, and staff members say the student body struggles with both behavioral and academic challenges. NPR used only K's first initial because of the stigma associated with attending a school designed for students with disciplinary histories.
For this population, a tool that helps students regulate emotions and engage with school could be genuinely transformative. But it could also pose unique risks. Students who are already vulnerable may be more susceptible to forming unhealthy attachments to AI companions, or to being influenced by a chatbot whose responses are generated by algorithms rather than trained counselors.
The financial dimension is significant as well. Dayton schools paid the education technology company SchoolAI $72,600 during the last academic year for a suite of AI tools that included the Jordan chatbot. That expenditure represents a meaningful investment for a district serving students with complex needs, and it underscores how quickly AI vendors have moved into the education market.
Similar purchases are happening in districts nationwide, often with little public scrutiny. Teachers report using AI to draft lesson plans, generate quizzes, and differentiate instruction for students at different levels. In some districts, chatbots now provide feedback on student work, supplementing or even replacing some forms of teacher commentary.
Each of these applications sounds promising in theory. AI can process information quickly, offer round-the-clock availability, and tailor responses to individual students in ways that a single teacher managing thirty pupils cannot easily replicate. But promising in theory is not the same as proven in practice.
The research base remains extremely limited. There is no substantial body of peer-reviewed evidence demonstrating that AI tutoring improves learning outcomes over the long term, that chatbot feedback is as effective as human feedback, or that social-emotional AI companions produce lasting benefits rather than dependency. What exists instead is a patchwork of pilot programs, vendor claims, and anecdotal enthusiasm.
That evidence gap matters because education is not a neutral domain. What happens in classrooms shapes children's cognitive development, their sense of themselves as learners, and their relationships with adults and peers. Deploying unproven technologies at scale, particularly with vulnerable students, amounts to a large-scale experiment without the usual safeguards of controlled trials and informed consent.
Critics also worry about data privacy, algorithmic bias, and the replacement of human judgment with automated systems. These concerns are not hypothetical. AI systems can reflect and amplify biases present in their training data, and the inner workings of commercial chatbots are rarely transparent to the districts that buy them.
Yet the experimentation continues, driven by a combination of genuine enthusiasm, competitive pressure, and the fear of being left behind. Teachers who see their workloads grow year after year may find AI tools genuinely helpful, and districts may feel they cannot afford to wait for definitive research before acting.
The result is a paradox: schools are being asked to prepare students for a world shaped by AI, even as the adults running those schools lack reliable guidance on how AI should be used in the classroom itself. The "Wild West" metaphor captures both the excitement and the lawlessness of the current moment — a period of exploration without a map.
What is needed, according to the concerns woven through NPR's reporting, is not a halt to experimentation but a more disciplined approach to it. That means funding independent research, requiring transparency from vendors, protecting student data, and ensuring that human relationships remain at the center of education rather than being outsourced to machines.
Until that happens, students like K in Dayton will continue forming bonds with chatbots like Jordan, and districts will continue writing checks for tools whose effects remain largely unmeasured. The experiments abound. The answers, for now, do not.
(0)