The backlash against technology in school is here. The advocates want a five-year pause, and honestly, they aren’t wrong to be skeptical. Most tech in schools earns skepticism. But ours is designed for learning, and the evidence is emerging:

Values-based technology works.

Most AI products in classrooms today cannot meet the basic standard of good learning design. Schools should not accept any that cannot. Here is how SchoolAI answers each test.

1. Learning outcomes, not cognitive offloading

Our deepest concern about AI in education is that it does the thinking for students. In 2023 we built against it.

Dot, our AI learning assistant, does not produce finished answers; instead, it asks better questions. It’s the “warm demander” stance a good coach would take. The practical result is simply that students have to think harder, not less.

Students don’t always like that. “Just give me the answer!” “Tell me!” and “I hate this AI” pop up sometimes. But after the initial boundary testing, they settle into critical thinking.

We have two years of externally validated evidence that this works. Jordan School District in Utah runs 68 schools and 55,000 students. A two-year study tracked 13,882 student-AI conversations across 82 teachers. Between October 2023 and October 2025, critical thinking rose 28%. Conversations reaching the highest levels of Bloom's Taxonomy (analysis, evaluation, and creation) more than doubled. Gains held across every subject and grade level studied. The study received ESSA Tier 3 certification following external review by researchers at Instructure.

The most important finding is what drove those gains. Teachers who actively designed learning experiences on the platform showed significant improvement. Teachers with minimal usage did not. AI did not raise the cognitive bar by itself. Teachers raised it, with AI as the instrument. That distinction matters more than any feature we could name.

2. Safer than the realistic alternative

No technology in a child's life is absolutely safe. Not paper books, not pencils, not classroom pets. The right question is does supervised, teacher-mediated classroom AI make students safer than the unsupervised consumer AI they already use at home? The answer is yes, by a wide margin.

The chatbot harms that have rightly alarmed the public have happened with consumer products outside school oversight, where no adult sees the conversation. Inside SchoolAI, the architecture is the safeguard. Our Critical Alerts system flags concerning conversations to school staff through real-time dashboards and alert emails. Every alert is monitored with humans responsible at each step and content is age-banded for K-7 and 8-12. Students cannot say a harmful image request is a school assignment to get around our guardrails. We built against that specifically.

These are not features bolted onto a chatbot. They are the conditions under which we believe AI should ever interact with a child. Banning supervised AI from schools does not make students safer. It pushes them toward the consumer AI in their pockets that alerts no one, even when the AI itself is concerned. That's where the harms keep happening.

3. Banning AI doesn’t stop cheating. It hides it.

Removing supervised AI from classrooms does not reduce student AI use. It moves it home. Unsupervised, unmediated, with no teacher in the loop. That is the worst of both worlds for academic integrity.

SchoolAI is built against this by design. Spaces give teachers full real-time visibility into every student conversation: who is working, what they are asking, where they are stuck, where they are looking for shortcuts. Educators set the scope, the context, the rules, and the resources of every session. Dot is built as a teaching assistant that helps each student think, not do their homework for them.

Academic dishonesty in education is real, and it predates AI. What is new is that for the first time, schools can put student AI use under direct teacher observation rather than driving it underground. That is a feature of supervised classroom AI no consumer product offers, and no AI ban can deliver.

The path to academic integrity in the AI era is not less supervised AI. It is student AI use that happens where a teacher can see it and guide learning moments.

4. Privacy, equity, and the things we cannot bluff

The four areas in this test are also the easiest places in AI marketing to make claims that do not survive contact with reality. Privacy, civil rights, ethics, climate. We’ll claim only what we can defend.

We publish our compliance and privacy notices publicly at trust.schoolai.com. Anyone can check it. It covers US COPPA and FERPA, but also meets rigorous international standards. Unlike big tech, we work to use less student data, not more.

On equity, our IEP drafting tool exists because special education teachers spend too much time on compliance documentation and not enough on students. It produces faster IDEA-compliant drafts for the educators serving neurodivergent students. Our work on bias and fair representation is an active program, not a completed one. Surface-level stereotypes have no place in instruction, and we work to make sure they do not get there.

Climate is where we owe the most honesty. We are an application-layer company and do not train the underlying models, so the largest portion of AI's environmental cost is not within our direct control. What is within our control is how we design the learning experience. Pedagogy that scaffolds focused work produces less inference per learning outcome than the sprawling sessions consumer AI is built to encourage. Efficiency and good teaching pull in the same direction here. We will meet credible disclosure standards before they are required of us.

5. Built around teachers, by architecture

This is the test we hold most strongly, and the principle the product is built around. Most AI in education is consumer AI wrapped in a school skin: a chatbot students plug in and talk to. SchoolAI is the inverse. A teacher-facing platform with student-facing surfaces, where the educator orchestrates, monitors, circulates the room, and intervenes.

Spaces put teachers in real-time view of student work. Our tools do not replace educator judgment. Critical Alerts route concerning conversations to school staff so a human, not an algorithm, decides what happens next. Teachers partner with Dot to develop high-quality materials drawing on evidence-grounded pedagogy, not AI slop. 

This architecture is the reason a SchoolAI classroom is an extension of a teacher's classroom. The platform does not run in parallel to instruction. It sits inside it. For the most vulnerable students, a teacher in the loop is non-negotiable. For advanced students, the technology unlocks learning at their pace. That is how we built it. That is the only way we believe it should be built.

The AI children experience in schools should meet a higher standard than the AI in their pockets. They also deserve to graduate ready for a workforce where AI is a baseline expectation, not a foreign concept. We welcome regulations that hold every vendor to that standard.

The two-year critical thinking study referenced above, "SchoolAI Makes Students Think," is available at schoolai.com/research, along with our broader evidence portfolio.