The survey does not establish that generative AI has caused a decline in critical thinking. It records teachers’ professional concerns about a rapidly changing classroom toolset. But the policy record shows their larger complaint has substance: B.C. has issued a provincial framework, yet it has left the actual rules, approved products, age thresholds, privacy controls, and assessment practices to individual districts.
That means the practical question is no longer whether a chatbot belongs in a school. It is whether a district can show that its approved AI service, data safeguards, classroom rules, and teacher training fit together. On that measure, B.C.’s approach is still a patchwork.
The provincial guidance is a framework, not a classroom rulebook
B.C.’s Ministry of Education and Child Care published its “Considerations for Using AI Tools in K-12 Schools” framework in 2024. The document addresses ethical use, privacy and data security, accessibility, teaching and learning, inclusion, integration, and local needs. It is a serious starting point, particularly because it directs districts to consider privacy obligations before selecting tools.
But the ministry’s own online guidance makes the division of responsibility clear: school boards, district leaders, school administrators, teachers, and staff are expected to develop local procedures and policies. The province provides resources; it does not prescribe a single approved chatbot, a common student age requirement, or a uniform system for declaring AI use in assignments.
That decentralization has consequences. A student’s access to AI, the protections around their prompts, and the standard used to judge AI-assisted work can differ depending on which B.C. district they attend. For teachers who transfer between districts, and IT teams asked to support staff working across shared services and cloud platforms, the difference is operational rather than theoretical.
The BCTF’s March 2026 research note found publicly accessible AI guidance, policies, or related resources in 26 of B.C.’s 60 districts. The wording matters. The federation did not claim that only 26 districts have formal AI policies. Its count includes a mixture of guidelines, formal policies, and other resources posted online. Districts without a public-facing item were omitted from the appendix.
So the number is evidence of uneven public documentation, not a definitive audit of every district’s internal practices. Still, it is difficult for families, teachers, or vendors to verify expectations when policies are either absent from district websites or folded into broader acceptable-use documents.
Copilot approvals show why “AI in schools” is too broad a category
Microsoft Copilot is already appearing in district AI plans, but its status varies materially. Langley School District’s published guidance identifies Microsoft Copilot as its primary approved generative-AI tool, while saying Copilot is currently approved for staff use and that a 13-and-over version would be approved for Grades 9 through 12 upon release. Its guidance requires staff to use the district’s Copilot account when handling private or identifying information.
Vancouver School Board takes a more restrictive published position: its May 2026 AI page says that Microsoft Copilot with Enterprise Data Protection, accessed through the district login, is its only approved AI chat platform.
Those are not interchangeable implementations. A consumer chatbot account, a school-managed account, and a service covered by enterprise data protections present different governance questions. Districts must establish which account type is permitted, whether conversation content may be retained or used for service improvement, who can enable plugins or connected services, and what teachers may enter into prompts.
The B.C. ministry framework calls for attention to data security and privacy, but a framework cannot substitute for a procurement record. School IT administrators need a written answer to a basic question: what information may never be pasted into an AI prompt? Student names, individualized education-plan details, health information, behavioural notes, parent communications, unpublished assessment responses, and identifiable classroom records should not be left to each staff member’s judgment.
This is where the BCTF’s demand for teacher involvement overlaps with the work of district technology teams. Procurement and privacy review can establish a permitted service, but teachers decide whether a particular use supports instruction or sidesteps it. A tool can be technically approved while remaining educationally inappropriate for a specific task.
The assessment problem is more immediate than the critical-thinking debate
The headline survey finding about critical thinking will attract attention, but the teachers’ concern about assessment is the problem districts can act on now. If a student can generate a polished response in seconds, an assignment completed outside class no longer demonstrates the same thing it demonstrated before widespread chatbot access.
That does not mean every take-home essay is worthless, nor does it mean AI detectors can restore certainty. Detection software produces probabilistic signals rather than proof of authorship, and false accusations carry real consequences. The reliable response is to redesign assessment around evidence of process and understanding.
Districts that allow student use should define what counts as permitted assistance. Brainstorming, language translation, accessibility support, outline generation, feedback on clarity, coding help, and full-text generation should not all be treated as one category. Each activity changes what the assignment measures.
A workable policy should require teachers to state, in the assignment itself, whether AI is prohibited, permitted with disclosure, or expected. When AI is permitted, students should identify the tool used, the type of assistance received, and the portions of work influenced by it. Where the goal is to assess independent writing or reasoning, schools need complementary evidence: in-class work, drafts, oral explanations, short conferences, source annotations, or demonstrations of how a student checked a model’s output.
Langley’s published guidance already points in that direction. It tells educators to communicate when AI is or is not allowed and says students must acknowledge AI use, while prohibiting them from presenting AI-generated work as their own. The larger gap is consistency. A rule posted on a district website does little if individual students receive contradictory expectations between subjects.
Teacher concern is widespread, but it is not a verdict on the technology
The BCTF survey is not an isolated expression of professional anxiety. A June 2026 NPR/Ipsos poll of U.S. K-12 teachers also found a majority concerned about AI’s effect on students’ critical-thinking skills, with many saying it complicates judging what students know. Common Sense Media reported on August 18 that many U.S. teens who use AI for schoolwork say they use it in ways that hand off critical thinking.
These surveys describe perceptions, reported behaviour, and policy gaps; they do not settle a causal question about learning outcomes. Students may use generative AI to evade effort, but they may also use it to compare explanations, translate material, receive practice questions, or explore ideas they could not otherwise access. The outcome depends on task design, supervision, and whether students are taught to challenge the system rather than accept its answers.
The B.C. government’s own guidance makes this distinction when it calls AI a supplementary tool rather than a substitute for human processes. The missing link is implementation. A district cannot claim to teach AI literacy simply because students have access to a chatbot. AI literacy requires students to recognize hallucinated information, spot weak sourcing, understand bias and privacy risks, compare outputs with reliable materials, and explain their own reasoning.
That is more demanding than teaching students how to write a better prompt. It asks schools to preserve the productive struggle that generative AI can so easily remove.
The next policy test is whether districts publish usable rules
The BCTF is calling for a provincial-level table that includes classroom teachers in AI policy decisions. Its request should not be read as a demand to ban the technology. Teachers surveyed by the federation are themselves using AI for routine administration and lesson planning, according to AM1150’s report. The issue is governance: who determines acceptable use, who carries the accountability when things go wrong, and whether those rules are clear enough to use in a classroom on Monday morning.
For B.C. districts, the immediate work is concrete. Publish the approved tools and account types. State the student age and grade limits. Set a data-handling rule that staff can understand. Require assignment-level disclosure expectations. Give teachers time and training to redesign assessments. Review the rules on a schedule that recognizes how quickly services such as Copilot change.
Without those basics, schools will continue to have an AI policy in the abstract and an AI reality defined by individual teachers, student workarounds, and whatever chatbot is easiest to reach.