Schools Need a Better Way to Think About AI

For the last few years, schools have spent a great deal of time trying to decide what to do about AI.

Much of that work has centered on policy. Schools have debated student use, academic integrity, teacher practice, data privacy, acceptable tools, and the role of generative systems in assignments and assessment.

Policy has become the default response, partly because policy feels concrete. It can be written, approved, published, and revised. It creates the sense that the institution has acted. The problem is that AI is changing too many parts of school life (let alone life outside of school) for most policy documents to carry the weight schools are placing on them.

AI now appears in search, writing tools, productivity platforms, learning systems, communications software, student services, assessment tools, and the software schools already depend on. It enters through formal adoption and through individual experimentation. It shows up in classrooms before committees have met and in vendor products before contracts are renewed. The result is more than a simple yes or no about whether or not to use a tool. Rather, the ubiquity of AI makes it an institutional challenge that reaches into teaching, governance, technology, ethics, the lives of students and their families, and the integrity and trust of the learning institutions themselves.

One of the central weaknesses in the current response is that schools often treat AI as a single issue when it is several issues layered together.

Academic integrity is one issue. Faculty use is another. Student formation is another. Then there is information literacy and technology governance. Community communication becomes an issue. And then there is the fact that AI is constantly being integrated into everyday digital products across the web. As a result, ideas of either “banning AI” or “hyping AI” end up looking foolish. Because AI is not one issue, but a variety of overlapping issues. They overlap, but they do not collapse into one another. There is no black-and-white.

That distinction matters because different parts of a school experience the same technology in different ways. A teacher may encounter AI through student writing. A technology leader may encounter it through a platform update. A parent may encounter it through questions about standards and judgment. A student may encounter it as part of the ordinary environment in which schoolwork now happens. Leadership has to make sense of all of these experiences at the same time.

There is no credible way to tell a student simply not to use AI.

Frankly, that would mean telling the student not to use the internet. AI is already being built into search, writing tools, productivity software, phones, browsers, cybersecurity solutions, and the platforms students use every day. The real question is how students should use these systems, how they should decide when not use them, and what intellectual responsibilities still belong to the student. Schools would be wise to consider the challenges of AI as an opportunity to explore a conversation about agency with students.

But the absence of shared language makes that harder. A school can have an AI policy and still have wide variation in what teachers permit, what students understand, and what families believe the school expects. The policy may exist while the culture remains unsettled. In that situation, the document can create the appearance of coherence without producing much coherence at all.

The deeper challenge is educational. As machines become more capable of producing text, images, analysis, and code, schools have to become more precise about what they want students to learn and what kinds of work they believe will produce positive learning outcomes. That question reaches into authorship and judgment. It creates challenges around what we expect from memory, effort, and interpretation. It forces us to define responsibility. It also exposes a weakness in any approach that begins with the tool instead of the educational purpose.

Students do not need to learn how to use AI or how not to use AI. They need to learn what success and failure look like and feel like in an environment where AI has become pervasive.

Faculty face a similar problem. Teachers are being asked to make decisions about systems that can generate materials, summarize student work, produce feedback, organize information, and shape instruction. Some uses may save time. Some may improve access. Some may weaken professional judgment or create new dependencies that are difficult to see at first. Administration and faculty could likely spend weeks just discussing the ways that AI’s integration into search engines changes expectations about what it means to access information online. A school that leaves every teacher to figure this out alone is not preserving professional autonomy. It is outsourcing institutional responsibility.

Information awareness belongs in the same conversation. Students now work in an environment where synthetic media, generated text, fabricated sources, and plausible falsehoods are easier to produce and harder to detect. Schools have long taught students how to evaluate sources. That task has changed. The issue is no longer limited to whether a source is credible. It now includes whether the source exists, whether the evidence is authentic, and whether the apparent authority behind a claim has any connection to reality. If social media was a megaphone, AI is an army of millions of people carrying megaphones. Each person on the internet in charge of their own information shock troops. And no one knows exactly what this means.

Schools need to be okay with not knowing. But there is a difference between not knowing and not engaging.

Not engaging is not an option. Because there is also a technology management problem that receives less attention. Schools often adopt tools one department or one teacher at a time. AI speeds up that pattern. New functions appear inside existing platforms, staff members create new workflows, and small experiments can become part of everyday operations. Over time, the school can accumulate dependencies that few people understand. This is where the discussion of AI begins to overlap with technical debt and technology stewardship.

It won’t be long before teachers and students using AI to create new applications or automated workflows to do tasks for them will be as commonplace as anything else. The education space has always adapted to new technological realities. There was a time when the idea of video-conferencing seemed far-fetched. And there were always new technologies that bent the norm until they became the norm. Like being on social media. Or listening to a podcast. Or the days when blogs were king. Or starting a web page. Or joining a bulletin board a million years ago. Every time a new thing came along, it was initially distrusted or even scoffed at. And before long it became ordinary, expected, and even depended upon. Sometimes, as in the case of social media, maybe it became depended upon too much. We cannot risk that happening with AI.

Because there is a difference. In the past, we talked about digital capabilities as tools. And there are plenty of people talking about AI as though it is just a tool. That might be presumptuous. It might be more useful to resist calling AI anything at all. Resist putting it into a familiar box. Instead, we should consider the ways that it might have an impact on our lives. How it may have an impact on the future families and society supported by the students who are now in our classrooms getting mixed messages from all of the adults and authority figures. Maybe we need to spend less time categorizing and more time talking.

Technologies are built on top of one another.

That overlap deserves more attention than it gets. It sounds like a boring problem. Technical debt. It is something you clean up when you have to. Or are forced to. Technical debt in schools is not limited to old code or aging systems. It can also take the form of undocumented workflows, unclear ownership, fragmented decision-making, vendor dependence, and technologies that become operational before anyone has decided how they should be governed. AI makes these conditions easier to create and harder to unwind.

And if the prediction holds that at some point in the near future, teachers and students are using AI to build their own applications and process workflows, then it stands to reason that the amount of technical debt potentially accruing in any school ecosystem will grow exponentially.

That becomes an issue of institutional capacity. Schools need clear decision paths, shared language, stronger faculty practice, better communication, and a more disciplined way to evaluate technology as it enters the institution. To evaluate not that AI is being used, but rather to evaluate what AI is making.

None of that requires certainty about where AI is heading. It requires a school to know what it values, how it makes decisions, and how those decisions connect to teaching and learning.

The schools that navigate this period well will be the ones that develop the habits and structures needed to keep making sound decisions as the technology changes. They also need to realize that we are transitioning into a new era and there are going to be reams of difficult questions we are going to have to deal with. Difficult questions. Difficult conversations. Schools need to face this head-on and not insist on trying to shelter both students and faculty from something entirely outside of their control. Because schools do not control AI. Cat’s out of the bag. Instead, successful educational environments will see this as an opportunity to be there for their students when they needed it the most. That is not a conversation about whether or not you are allowed to use AI. It is a conversation about what it means to learn. And that is the sort of thing a school environment, if well prepared, can offer in a way no other institution can.