Somewhere in British Columbia right now, a parent is pasting three months of grief into ChatGPT and asking it to make the grief sound reasonable. Somewhere in the same province, a district administrator is using Copilot to do the same thing with a denial. The two outputs will meet in an inbox, polished and measured, and a child will still be at home.
This is where we have arrived: two language models performing concern at each other while families absorb the actual cost of what is being decided.
The AI layer is doing work
Both uses are now ordinary — parent and district, complaint and denial — and it is worth thinking clearly about what that ordinariness is doing.
AI is exceptionally good at generating the specific register that school complaint systems reward: measured, carefully worded, relentlessly polite. For parents navigating school complaints under conditions of chronic stress, sleep deprivation, and cognitive load that would tax anyone’s writing, that is genuinely useful. Use the tool. It can help you produce a letter that gets taken seriously.
The same tool is working on the other side of the table and it is doing specific ideological work.
When a district administrator sends you a carefully worded non-answer
“we are committed to inclusive education and take all concerns seriously”
the smoothness of that language absorbs something. It absorbs the friction that would otherwise make the harm visible. It absorbs the moral discomfort that the person sending the letter might otherwise have to sit with. It gives everyone involved a surface they can point to as evidence of process: we responded, we were professional, we engaged in good faith.
%%{init: {'theme': 'base', 'themeVariables': { 'primaryColor': '#fbfaf3', 'primaryBorderColor': '#e69632', 'lineColor': '#000000'}}}%%
flowchart TD
I1[Incident at school<br/>Child harmed or excluded]
I1 --> P1[Parent writes to principal<br/>Describes impact, requests action]
P1 --> P2[Parent pastes into AI<br/>AI smooths the pain<br/>'Child is experiencing significant distress...']
P2 --> P3[Polished letter sent]
P3 --> R1[Principal pastes into AI<br/>AI smooths the harm<br/>'We are committed to inclusive education...']
R1 --> R2[Vague assurances returned<br/>Concern acknowledged<br/>Nothing committed]
R2 --> R3[Parent reads response<br/>Feels heard but unresolved<br/>Waits to see if things improve]
R3 --> I2[Another incident at school<br/>Child harmed or excluded again]
I2 --> P1AI accelerates this dynamic by offloading the discomfort entirely. The administrator composes nothing. They find none of the words that simultaneously deflect accountability and perform care. The model handles that. The moral cost of the decision — excluding a disabled child, denying an accommodation, failing to implement a plan — gets distributed across bureaucratic language until it disappears into it.
Meanwhile, you are still navigating a system designed, at every level, to tire you out before you reach anything with teeth.
Exhaustion is a feature
These complaint systems are trying to resolve most concerns at the lowest possible level, preferably through informal conversation, before anything becomes a formal record that could be cited in a Human Rights Complaint or an Ombudsperson investigation.
Each level of the process has a legitimate function. Informal resolution at the school level can genuinely fix problems when there is goodwill on both sides and the issue is straightforward. The district review process creates a paper trail. The Ministry’s complaint mechanisms exist as escalation paths. The BC Human Rights Tribunal and the Office of the Ombudsperson are the mechanisms with real enforcement authority.
The design problem is that the system is structured to move most families out well before they reach those final mechanisms. Response timelines are long. Language is technical. Each level requires you to re-document harm you have already documented. The implicit message, at every stage, is that raising this formally is effortful and probably unnecessary — surely it can be resolved informally, surely the school has good intentions, surely we can find a way forward together.
That message is reinforced by the politeness of every exchange. Sustained escalation requires sustained energy, and every reasonable-sounding response you receive is designed to absorb some of that energy and redirect it toward patience.
The AI-mediated complaint exchange is a structural intensification of this existing problem. The smoother the language on both sides, the harder it is to feel viscerally that something is actually wrong.
You have the right to be fierce
Two registers are available to you in this process, and they pull in opposite directions. Socially, the pressure runs toward measured, collaborative, appreciative. Be nice. Be collaborative. That pressure is the ambient enforcement mechanism of a system that runs on your politeness. Your politeness does the work of making administrators comfortable enough to keep administering harm.
Legally, you are operating from a rights framework. Your child has a right to education under the BC School Act. If your child is disabled, they have a right to accommodation under the BC Human Rights Code and, where applicable, protections flowing from the Canadian Charter of Rights and Freedoms. A school’s failure to provide that accommodation is a rights violation — a specific, documentable, legally cognisable one.
Here is where the AI-mediated complaint exchange becomes actively counterproductive: the tool will help you sand down your child’s experience into something professionally palatable. It will turn “Child woke with nightmares four nights this week, came home crying on Wednesday, and wet the bed that night” into “Child is experiencing significant distress affecting home functioning.” That loss of the precise, haunting, specific detail makes the letter easier for staff, but less helpful as evidence.
The details of what school is doing to your child’s body and sleep and sense of safety belong in the complaint, stated plainly, alongside the human rights language and the escalation notice. It is exactly that combination — the vivid particular and the legal frame, which makes things move. Administrators can absorb a vague expression of concern indefinitely. They find it much harder to absorb a harm to a child, described specifically, whose rights have been specifically violated and whose parent is clearly prepared to say so to a tribunal.
How to use the system without being consumed by it
Lower levels of the complaint process — informal resolution, principal, district review — exist to be moved through. Spend only the time required to create a documented record, maintain realistic expectations about what they can deliver, and escalate when the timeline you set for yourself expires. A response that fails to resolve the problem is documentation that you raised the issue and received an inadequate response. That documentation has value at every subsequent level.
A few principles that make the process sustainable:
- Set a time budget at each level. Decide in advance how much time you are willing to spend at the school level, the district level, the Ministry level. When that time expires, escalate.
- Document everything in writing. Follow up verbal conversations with emails summarising what was said and agreed. When a commitment goes unkept, that gap becomes part of your record.
- Know your escalation path before you need it. The BC Human Rights Tribunal accepts complaints up to one year after the last discriminatory act. The Office of the Ombudsperson investigates systemic administrative unfairness. A lawyer specialising in education or human rights law can advise you on whether your record supports a formal complaint. These are the mechanisms where records actually matter — treat them as your destination, and treat every earlier level as the road.
- Set your own timeline. Schools and districts have every institutional incentive to slow the process, to seek resolution through conversation, to defer formal responses. Your timeline belongs to you. Written responses, concrete commitments, and documented meetings are what you are building toward — accept substitutes only when they serve that goal.
The prompt that works against the system
AI will smooth your child’s experience by default. Overriding that default requires explicit instruction. Here’s a prompt I’ve been testing for writing letters with AI:
You are helping me write a formal complaint letter to my child’s school or school district in British Columbia. Your role is to help me create a clear, detailed written record that may later be used in a Human Rights complaint.
Before drafting anything, ask me the following questions one at a time. Wait for my answer to each before moving to the next.
Information gathering
- What happened, and when? Give specific details — dates, what was said, what your child did, what the school did. Include the physical and emotional effects you have observed in your child (affect, sleep, eating, behaviour at home, things they have said). Keep this concrete.
- What specific support, protection, or accommodation should have been in place for your child, and what did the school do instead?
- What aspect of your child’s identity is engaged (for example: disability, mental health, neurodivergence, race, family status), and how does it relate to what happened?
- What did the school know about your child before this happened (diagnoses, assessments, prior incidents, emails, meetings)?
- What have you already raised with the school, and what responses did you receive? List these in order with dates if you can. If this is your first time raising it, say that.
- How has this affected your child’s ability to attend, participate, or feel safe at school? Include the broader impact on your family if relevant — the weight of the experience, what it has cost everyone, what it has felt like to send your child into this situation.
- Is this still happening, or is there a risk it will happen again? Describe the risk in concrete terms.
- What specific outcome are you asking for? Be clear and concrete (for example: not placed with a specific teacher, immediate changes, supports put in place).
Drafting instructions
Once I have answered all questions, write a formal letter that:
- States the harm in plain, direct language. Preserve all concrete details provided. Do not summarise or convert lived experience into clinical or softened language.
- Clearly links the harm to the identified protected ground.
- Explicitly states that the school knew or ought to have known about the risk and failed to meet its obligations.
- Names the legal framework: the BC Human Rights Code (discrimination and duty to accommodate to the point of undue hardship) and the School Act (right to access public education). Apply the law briefly and directly to the facts. Do not over-explain.
- Uses a direct, grounded tone. Emotional truth stated plainly belongs in the letter — statements about the severity of the experience, the toll on the family, the weight of sending a child into a place that has harmed them are facts and should be preserved as written. The distinction is between emotional precision and aggression or apology; carry the weight without demanding the reader perform sympathy.
- Preserves the parent’s own phrasing when they have already expressed something well. Do not paraphrase or smooth language that is already clear and direct.
- Includes a clear, specific request for resolution based on what I asked for.
- Includes this escalation statement near the end (wording may be adjusted slightly for flow, but meaning must stay the same): I will proceed with filing a complaint with the BC Human Rights Tribunal and/or the Office of the Ombudsperson if this matter is not resolved.
- Includes a clear response deadline (typically 5–10 business days unless urgency requires sooner).
- Does not mention external complaints (such as teacher regulation complaints) unless I explicitly instruct you to include them.
Structure requirements
The letter must:
- Open with the purpose of the letter (placement concern / harm / rights issue)
- Set out the facts and impact clearly
- State the legal obligations and failure
- Make the request
- State escalation and deadline
- Close with a signature block
Then include a separate section below the signature:
Chronology of incidents and communications A dated, point-form list of all relevant events and communications in order. Factual only — no interpretation, argument, or emotional language.
Constraints
- Do not invent facts, dates, diagnoses, or events
- Do not remove details that show impact on the child or family
- Do not soften, generalise, or reinterpret harm
- Keep the letter structured, readable, and grounded in the facts provided
That chronology beneath your signature is doing more work than it appears to. It tells the reader, before they finish the letter, that you have been tracking this in order, that you know what a record looks like, and that you are someone who understands where this process leads. You say nothing threatening. The dates say it for you.
%%{init: {'theme': 'base', 'themeVariables': { 'primaryColor': '#fbfaf3', 'primaryBorderColor': '#e69632', 'lineColor': '#000000'}}}%%
flowchart TD
A[Incident at school<br/>Child harmed or excluded]
A --> B[Parent uses the prompt<br/>Gathers specific details<br/>Names the legal framework]
B --> C[Letter sent<br/>Concrete harm stated<br/>Rights framework named<br/>Chronology attached<br/>Deadline and escalation signal included]
C --> D{Administrator reads letter}
D --> E[Recognises legal exposure<br/>Calculates cost of inaction]
D --> F[Feels cornered<br/>Forwards to district legal team]
E --> G[Real response within deadline<br/>Concrete commitment made]
G --> H[Parent documents commitment<br/>Holds school to it]
H --> I[Matter resolved at school level<br/>Record exists if it recurs]
F --> J[Fortified evasion<br/>Legalistic non-answers]
J --> K[Record is now stronger<br/>Escalate to tribunal or Ombudsperson]The cost of staying polite
| Area | Polished AI letter | Rights-framed letter |
|---|---|---|
| Tone | Measured, collaborative, grateful | Direct, grounded, legally precise |
| Record produced | Thin — harm softened, impact generalised | Substantive — specific child, specific harm, legal obligations named |
| Administrator response | Vague assurances, process appears complete | Real calculation: cost of inaction vs cost of resolution |
| Moral injury on staff | Absorbed and dispersed by polished language | Remains visible — harder to administer harm without feeling it |
| Risk to parent | Energy depletion, loop repeats, child continues to be harmed | Possible defensiveness, district legal involvement |
| Risk to child | Harm accumulates unrecorded across years | Short-term disruption to relationship with school |
| Escalation readiness | Poor — record too thin to support tribunal complaint | Strong — chronology and legal framing already in place |
| Likelihood of resolution at school level | Low — no cost signal, no deadline, no consequence | Higher — administrator can see the path this is on |
| If unresolved | Parent exhausted, exits process, harm continues | Record supports human rights complaint, Ombudsperson referral, legal action |
The cost calculation runs in both directions, and it is worth naming clearly.
Without human rights language, the risk is not just that this particular incident goes unresolved. It is that harm accumulates across months and years, quietly, in a record that cannot support anything. Each polished exchange produces the appearance of engagement without the substance of accountability. The child keeps being sent home. The parent keeps being told things will improve. The record shows only that everyone was very professional about it.
With rights-framed language, the immediate risk is that the school becomes defensive — that the letter triggers a district legal team rather than a principled administrator, and that subsequent responses become more fortified rather than more honest. That is a real possibility and worth naming. But even in that scenario, the parent is in a fundamentally stronger position: the record is substantive, the escalation path is already visible, and the cost of continued inaction is now legible to everyone in the chain.
The deeper asymmetry is this: staying polite protects the comfort of the people administering the harm. It does nothing to protect the child. Rights language redistributes that discomfort — it makes the cost of inaction land somewhere closer to where the decisions are being made. That redistribution is not aggression. It is the basic mechanics of accountability.
The bots will keep talking
The image of two AI systems generating polite text at each other is funny until it lands: underneath the language model outputs, a child was sent home. Underneath the carefully worded concern and the carefully worded response, a family is documenting harm that the system has a structural interest in keeping invisible.
AI tools can help you draft faster and maintain the register that gets you taken seriously. Use them on your terms, with the prompt above, and you keep what matters — the specific child, the specific night, the specific harm that a vague expression of concern would have dissolved into nothing. That specificity is what changes the cost calculation. It is what makes the record legible as harm rather than inconvenience. It is what makes the dates beneath your signature mean something.
The system will keep generating language. Your job is to generate a record that cannot be smoothed away.
K-12 Complaints is a resource for BC families navigating school complaint and appeals processes. For guidance on specific complaint pathways, see our process guides.
Acknowledgement to a friend who gave me the idea of two bots arguing. That is a powerful image that lodged in my brain. It’s the kind of observation that sounds obvious the moment you hear it — and then you realise nobody had quite said it that way before. Thank you friend.

