Decolonizing AI Ethics in Education: A Systematic Review and Glocalized Framework (FGAIEE)

Imagine a world where the rules for how we treat each other are written by a single group of people, using only their own history and language, and then applied to everyone else. This is the current reality for artificial intelligence in schools. As computers become smarter, they are increasingly used to grade tests, recommend lessons, and even decide who gets into a university. But the ethical guidelines meant to keep these systems fair are mostly created in wealthy, Western nations. These guidelines assume that everyone thinks, speaks, and learns in the same way. When these rules are applied to students in Africa, Asia, or Latin America, they often fail. They might misunderstand a student's accent, ignore local cultural values, or simply not work because the internet connection is too slow. The result is a system that claims to be fair but actually pushes certain groups further to the edge.

This imbalance has sparked a new question among researchers: who gets to decide what is right? A team of scholars from Indonesia has set out to answer this by analyzing how AI ethics are actually used in classrooms around the world. They did not just read the rules; they examined the 84 published studies that discuss the people who wrote the guidelines, the students they affect, and the places where they are used. Their goal was to find a way to make AI ethics work for everyone, not just the few who hold the power to write the guidelines. They call this approach "glocal," a blend of global and local, suggesting that good rules must be made with the whole world in mind but built to fit the specific needs of each community.

The researchers began by gathering 84 different studies published between 2015 and 2025. They looked for any research that discussed how AI is used in education and whether it was fair. They sorted through thousands of articles, keeping only those that offered real evidence or deep analysis of how these systems work in practice. They paid special attention to studies from the "Global South," a term used to describe regions that have often been left out of the main conversations about technology, including countries in Africa, Latin America, and parts of Asia. By focusing on these voices within the existing literature, the team hoped to see a different picture of what ethical AI should look like.

What they found was a clear divide. The vast majority of the rules and guidelines come from institutions in Europe and North America. These rules are often written as a simple checklist of values, like "be fair" or "be transparent." However, the study shows that these checklists rarely change how things actually work in schools in other parts of the world. In many cases, schools in the Global South are told to adopt these Western rules without being asked for their input. The researchers found that when AI systems are built using only English data and Western ideas, they make mistakes more often for students who speak other languages or come from different cultural backgrounds. For example, automated grading tools often struggle with non-native English speakers, giving them lower scores not because their work is worse, but because the computer does not understand their way of speaking.

The study also revealed that the people who are supposed to be protected by these rules—teachers, students, and local leaders—are rarely part of the decision-making process. In the few cases where local communities were involved, the involvement was often just a formality, a box to be checked rather than a real partnership. The researchers noted that when local voices are ignored, the technology often fails to solve the problems it was meant to fix. Instead of helping everyone learn, it can deepen existing inequalities, making it harder for those who are already struggling to keep up.

However, the story does not end with failure. The researchers also discovered a growing movement of local solutions that are working much better. In Ghana, for instance, schools are using a concept called "Ubuntu," which emphasizes that a person is a person through other people. This idea is used to design AI systems that focus on community trust and shared responsibility rather than just individual scores. In Southeast Asia, another approach called "gotong royong," which means working together for the common good, is being used to make sure that everyone has a say in how technology is used. These local methods are not just small adjustments to the big global rules; they are completely different ways of thinking about what is right and wrong. They treat local knowledge as just as important as Western science.

The team concluded that the current way of doing things is broken because it treats ethics as a one-size-fits-all product. They argue that we need to stop trying to force a single set of rules on the whole world. Instead, they propose a new framework called the Framework for Glocalized AI Ethics in Education. This approach suggests that ethical rules should be built from the ground up. It means that when a school in a different country wants to use AI, they should be able to shape the rules to fit their own culture, language, and needs. It also means that the companies selling these tools should be required to prove that their systems work fairly for local students before they can sell them.

The researchers suggest three main steps to make this happen. First, the knowledge that guides these systems must come from many different places, not just one. Second, the people who use the technology must have a real voice in how it is designed and checked. Third, there must be real consequences if the technology fails to be fair, such as refusing to buy tools that do not work for everyone. This is not just about being nice; it is about making sure the technology actually works. The study shows that when local communities are involved in the design, trust in the system goes up, and the technology becomes more useful for everyone.

This new way of thinking does not mean throwing away all the old rules. Instead, it means changing how we make them. It is about recognizing that there are many ways to be ethical, just as there are many ways to speak a language or teach a lesson. The researchers believe that if we can build a system that respects these differences, we can create AI that truly helps all students learn. They are not saying that this is easy or that it will happen overnight. But they are clear that the current path, where a few powerful groups decide for everyone, is not working. The future of AI in education depends on listening to the people who are actually using it, and making sure their voices are heard in the design of the tools that will shape their futures.