Part 1 — The Risks of Using AI
The five real risks of AI in schools
These risks are real and worth taking seriously. They are also manageable — each one has a pedagogical fix, not a ban.
Data privacy exposure
Student data sent to consumer AI tools may be stored, trained on, or leaked. This is a real risk — and it is solved by using tools that do not train on student data and that follow FERPA and COPPA.
How it is managed
The Secondary AI does not train on student data and follows FERPA, COPPA, and PIPEDA. Student chatbot sessions are tied to the teacher's account, not sold.
Algorithmic bias
AI models carry the biases of their training data. A lesson generated without teacher review can reproduce stereotypes or present a narrow perspective.
How it is managed
The teacher review pass catches bias before students see it. The platform never auto-assigns — a teacher signs every output.
Over-reliance and erosion of critical thinking
If students use AI to do the thinking for them, they stop developing the skill. This is the most cited risk — and it is a pedagogy problem, not a tool problem.
How it is managed
Socratic and error-based pedagogical approaches make the AI ask questions instead of giving answers. The student does the thinking; the AI coaches.
Academic integrity
AI-generated essays and unattributed AI work undermine assessment. AI detectors do not work and harm the wrong students.
How it is managed
The platform designs AI-proof assignments — tasks that require the student's own thinking, context, and presence — rather than trying to detect AI after the fact.
The equity gap
Schools that can afford good AI tools pull ahead. Schools that ban AI or cannot afford tools fall behind. The digital divide widens.
How it is managed
The Secondary AI offers a free tier so any teacher can access pedagogically grounded AI, not just well-resourced districts.
Part 2 — The Bigger Risk
The five risks of not using AI in schools
The conversation about AI risks almost always frames AI as the danger. But there is a second column in the risk ledger — the cost of inaction — and it is larger.
Preparation debt
Students graduate into a workforce where AI is standard, having never been taught to use it ethically or critically. They learn it from TikTok instead of from a teacher.
The divide widens
Students in schools that use AI pedagogically gain a literacy advantage. Students in schools that ban it fall behind — not because they are less capable, but because they were never given the chance.
Teacher burnout continues
Teachers are leaving the profession at record rates. AI is a tool that can reduce the planning and differentiation workload. Not using it means accepting a status quo that is already breaking teachers.
Differentiation stays impossible
A teacher with 30 students at 5 different readiness levels cannot manually create 5 versions of every lesson. AI can. Not using it means differentiation stays a theory instead of a practice.
Students use it anyway — unguided
A ban does not stop students from using AI. It stops them from learning to use it well. They use it at home, unsupervised, with no framework for ethics or accuracy.
The Middle Path
The pedagogical middle path: guided use, not ban or abdication
The two extremes — ban AI entirely, or let students use it unsupervised — both fail students. The ban fails because students use AI anyway, unguided. Abdication fails because students never learn to think critically about what the AI produces.
The middle path is older than AI. It is how teachers have always introduced powerful tools: model the use, scaffold the practice, gradually release responsibility. We did it with calculators, with the internet, with search engines. AI is the next tool in that line.
The principle: a tool is only as good as the pedagogy around it. AI used with a Socratic approach, a learning objective, and a teacher review pass is a teaching tool. AI used with none of those is a shortcut. The risk is not the AI. The risk is the pedagogy.
This is Part 1 of 3