In classrooms around the world, a quiet revolution is taking place in how children learn to understand their own feelings and navigate their relationships with others. This area of education, known as social-emotional learning, teaches students to manage emotions, set goals, show empathy, and make responsible decisions. For years, educators have relied on human teachers to guide this delicate work. But in recent years, a new partner has entered the room: artificial intelligence. These are computer systems capable of recognizing emotions, simulating conversations, and offering personalized feedback to help students develop these crucial life skills. The promise is immense: technology that could offer endless patience, instant support, and tailored guidance to every child, regardless of their background.

Yet, as these tools become more sophisticated, a critical question remains unanswered. While researchers are busy building and testing these intelligent systems, they have largely forgotten to ask who should be in charge of them, what rules should govern their use, and how schools should prepare for their arrival. The technology is racing ahead, but the map for how to use it safely and fairly is missing. This disconnect creates a dangerous gap between what is possible and what is practical, leaving schools to navigate complex ethical and legal challenges without a guide.

A team of researchers set out to investigate this gap by looking closely at the scientific literature itself. They gathered sixty-five peer-reviewed studies that examined the intersection of artificial intelligence and social-emotional learning. Their goal was not to test the technology again, but to read what the scientists who built it had to say about the rules for using it. They wanted to see if these studies offered clear, actionable advice for policymakers, school leaders, and teachers, or if the topic of governance was being ignored.

What they found was a startling silence. Nearly three-quarters of the studies they reviewed did not mention policy implications at all. When the researchers looked at the remaining studies that did touch on the subject, they discovered that the advice was often vague, high-level, and disconnected from the reality of a classroom. The authors of these papers frequently praised the potential of their tools but stopped short of explaining who should act, what specific actions were needed, or when and where those actions should happen. It was as if a group of architects had designed a magnificent new bridge but left the engineers to figure out how to build the supports and who should be allowed to cross it.

The researchers used a simple but powerful method to uncover this pattern. They treated every mention of policy in these papers as a short story and asked five basic questions: Who is being asked to do something? What exactly should they do? Why is it necessary? When and where does it apply? And how strongly is this being recommended? By breaking down the text this way, they could see exactly where the stories fell apart. They found that while some studies mentioned the need for ethical guidelines or better training, they rarely specified which government agency should write the rules, which schools should get funding, or how to handle the specific risks of collecting emotional data from children.

This lack of detail points to a deeper issue in how science is currently practiced. The field seems caught in a trap where the excitement about technical innovation overshadows the hard work of governance. Researchers are often rewarded for showing that a new tool works in a small experiment, but they are not encouraged to think through the complex steps required to use that tool safely in a real school system. The study found that papers published in academic journals were slightly more likely to discuss policy than those published in conference proceedings, but even in the journals, the conversation was thin. The incentive structure of science appears to prioritize the "what" of technology over the "how" of implementation.

The few studies that did offer policy advice tended to focus on broad, abstract ideas like "protecting privacy" or "ensuring fairness." While these are important goals, they are not instructions. A school principal cannot build a policy based on the idea of fairness alone; they need to know who is responsible for checking the data, what specific software features are allowed, and how to train teachers to use the tools without overwhelming students. The researchers noted that when these details are missing, the result is a "pilot-study paradox." A tool might work perfectly in a controlled test, but when it is released into the messy reality of a school, it fails because there are no regulations to support it, no budget to train staff, and no clear rules to protect vulnerable children.

The analysis also revealed that the language used in these papers often lacked urgency or clarity. Some suggestions were written as weak possibilities, using words like "could" or "might," while others were presented as absolute demands without explaining the reasoning. This inconsistency makes it difficult for decision-makers to know how seriously to take the advice. Furthermore, the studies rarely explained the specific conditions under which their recommendations would work. A suggestion that applies to a university in one country might not make sense for a primary school in another, yet the papers often failed to make these distinctions.

To fix this, the researchers propose a shift in how scientists write about their work. They argue that policy thinking should not be an afterthought, added as a few sentences at the end of a paper just to satisfy a requirement. Instead, it should be part of the method itself, woven into the research from the very beginning. They suggest that every study should clearly identify the specific actors who need to act, the concrete steps they should take, and the reasons why those steps matter. This approach would turn vague ethical horizons into practical roadmaps.

The study concludes that the future of artificial intelligence in education depends on closing this gap. The technology is ready, and the need for social-emotional support is urgent, but the framework for using these tools responsibly is still being built. By demanding that researchers provide clear, specific, and actionable guidance, the scientific community can help ensure that these powerful tools serve students effectively and safely, rather than leaving them to navigate a complex new landscape without a map. The path forward requires a cultural change where the question of "how do we govern this?" is asked with the same intensity as "how does this work?"