Kevin Korte

The New AI Classroom Power Problem

For years, marketing has defended proprietary software with the same claims. The software was supposedly polished, supported, and easy to deploy. AI makes that compromise much harder to justify because educational systems now operate on far more than static content delivery. Instead, the systems analyze student behavior, generate recommendations, and shape decisions in ways that can affect opportunities and outcomes. Thus, the vendor is not just selling software anymore. It is mediating instructional judgment, data access, and sometimes even the interpretation of student performance.

This expansion of responsibilities creates an unhealthy asymmetry. Schools are expected to accept opaque systems on trust, while vendors retain control over models, data flows, and business logic. In education, that is a structural problem, not a procurement detail. When the technology can influence attention, assessment, or intervention, the people using it need more than a user interface. Teachers and students deserve transparency and accountability.

Why AI Opacity Becomes Harmful

The case against proprietary software is not just ideological. It is operational. AI in education raises documented concerns around privacy, bias, unfairness, low accuracy, surveillance, and weak accountability. Those risks are amplified when the software is closed, because educators and administrators cannot meaningfully evaluate what data the system collects, how it weighs variables, or why it produces a given recommendation. If districts cannot hire experts to examine systems independently, its errors and biases can persist longer and spread faster.

That matters especially in K-12 settings, where the users are minors and the power imbalance is already severe. A proprietary AI tool can quietly become part of a child’s educational profile, influencing how teachers, schools, or future systems see them. Once that profile is created, the child has little leverage over how it is reused, retained, or repurposed. In that environment, opacity is not neutral. It is a risk multiplier.

Data Rights Matter

Education should never require students to surrender their data to private systems they do not control. Yet AI-powered tools routinely depend on large-scale collection and processing of personal information, including behavioral signals, performance data, and sometimes biometric or surveillance-like inputs. Privacy, consent, and data ownership must be core concerns in AI-enabled education, especially when data allows for racial profiling.

Open Source software does not magically solve every privacy issue, but it changes the power relationship. Schools can inspect code, limit telemetry, remove unneeded dependencies, and choose how data is stored or processed. With proprietary platforms, those choices are constrained by vendor contracts and technical black boxes. If classrooms are going to use AI at all, the default should be systems that can be scrutinized, adapted, and governed locally.

Bias Is Not Theoretical

One of the strongest reasons to reject proprietary AI in classrooms is that bias is not an edge case. Humanity carries thousands of years of racial and gender biases, which enter these models as part of the training data. Thus, it is a predictable outcome that training data, model design, and deployment context misalign. As a result, past unfairness, discrimination, and low representativeness can actively harm marginalized learners today. Closed systems make it harder to detect these problems, and even harder to correct them.

That is especially dangerous in classrooms because educational technology can become self-reinforcing. If a model misreads a student’s behavior, labels them inaccurately, or recommends less ambitious content, the system can help turn a temporary issue into a durable disadvantage. A school should not have to guess whether a vendor’s model treats dialect, disability, culture, or socioeconomic background fairly. It should be able to verify these requirements.

The Open Alternative

Naturally, some might scream: “No Technology!” Yet, in our modern world, schools cannot wholly avoid any technology. Instead, schools need software they can understand, adapt, and govern in line with educational values rather than vendor incentives. That is where Open Source has a clear advantage: it gives institutions the ability to audit systems, control data flows, and reduce dependency on opaque third parties. It also supports local alignment, which matters because ethical and privacy concerns in education are contextual, not one-size-fits-all.

This is also a strategic issue. Proprietary AI tools lock schools into external roadmaps and pricing models, while open systems create room for community oversight, shared improvement, and policy alignment. In a world where AI will increasingly shape learning experiences, schools should not be consumers of invisible infrastructure. They should be stewards of digital environments built for education, not extraction.

A Change In Technology

Replacing proprietary software in classrooms does not mean banning all digital tools. It means drawing a boundary around control. If an AI system is important enough to influence teaching or assessment, it is important enough to inspect, govern, and, if necessary, modify. That is difficult or impossible with closed platforms. Schools already understand this logic in other contexts: we do not hand over curriculum governance to a black-box publisher and call that educational leadership.

AI has made that boundary urgent. The more software can predict, personalize, and persuade, the less acceptable it is to treat software procurement as a convenience purchase. Education depends on trust, but trust without transparency is just dependency with better branding. That is why proprietary software has no place in classrooms that want to remain accountable to students, parents, and the public.