Our Commitments & Design Principles.

How CARL approaches equity, Indigenous sovereignty, and responsible AI in education.

Our Perspective

CARL is built primarily with Canadian K–12 educators in public education contexts in mind. While grounded in British Columbia on the traditional and unceded territory of the Syilx Okanagan Nation, we engage with questions that extend across Canada and beyond: equity, accessibility, responsible AI, Indigenous sovereignty, and teacher control.

This page explains how CARL approaches responsible AI in education, including equity, accessibility, Indigenous sovereignty, and teacher control. It exists to be transparent about what CARL is, what it is not, and how our values shape our design choices.

We are educators and technologists whose work is shaped by experiences with inequity in education, along with research and classroom-informed practice. We are not Indigenous. We do not speak for Indigenous peoples, Nations, or communities. We work on land where First Peoples have lived and continue to live, and we understand that this carries responsibility, not authority.


What CARL Is, and What It Is Not

CARL is:

  • A teacher-controlled lesson planning platform designed to reduce workload while supporting inclusive, curriculum-aligned instruction

  • A structured system that offers optional enhancements (UDL, accessibility, EDI, student voice, etc.) to help teachers meet diverse needs without starting from scratch

  • A tool that supports professional judgment through editable outputs, clear constraints, and guardrails

CARL is not:

  • A replacement for teacher expertise, professional judgment, or relational teaching

  • A curriculum authority, evaluator, or “one right way” teaching engine

  • A source of Indigenous knowledge or a system that generates or interprets Indigenous teachings


Our Commitments

Equity as Structure, Not Add-Ons

For CARL, equity is not a checklist or an optional enhancement. It is built into lesson structure, prompts, and safeguards.

In practice, this means CARL is designed to:

  • Prompt for perspectives beyond dominant norms, including non-Western viewpoints and experiences that are often excluded

  • Flag common representation gaps (whose voices are centered, who is missing, what assumptions are showing up)

  • Support bias-aware instructional decisions, including age-appropriate language, respectful framing, and avoidance of stereotypes

  • Make inclusive practice more accessible, especially for teachers who care deeply about this work but face real time constraints

Equity in CARL is tied to lesson content and instructional decisions, not generic statements.


Teacher Control and Professional Judgment

CARL is designed to support teacher expertise, not replace it.

Teachers:

  • Control the inputs

  • Decide what to use, adapt, or skip

  • Retain full authority over instructional decisions

CARL includes guardrails to mitigate harm, but it does not prescribe “correct” teaching moves. Supports are optional, and teacher judgment always comes first. Because CARL uses an integrated lesson plan editor, teachers remain in full control of what they create, revise, and publish.


Accessibility and Universal Design for Learning (UDL)

Every CARL lesson includes multiple, concrete suggestions for increasing accessibility.

These supports:

  • Address both how students access learning and how they demonstrate understanding

  • Maintain equivalent rigor across options

  • Are framed as universal supports, not accommodations for “some” students

Teachers have the option to use CARL’s Lesson Enhancements for deeper dives with more detailed strategies, allowing teachers to choose the level of support that fits their context. Paired with easy-to-follow facilitator guides and rubrics, CARL aims to support teachers in reaching every student without creating additional work.


Student Voice, Agency, and Consent

CARL is designed with the belief that students are not passive recipients of learning.

Student voice and agency show up through:

  • Meaningful choices in process, format, and audience

  • Opportunities for consent, especially around sharing and personal disclosure

  • Clear “Why are we doing this?” explanations in student-facing activities, written in age-appropriate language

  • Low-pressure participation pathways that respect students as individuals

Agency is not treated as an engagement trick. It is a matter of respect.


Responsibilities of AI in Education

We believe responsible AI in education requires clear limits, transparency, and accountability.

That includes:

  • Avoiding harm and the reinforcement of harmful narratives

  • Designing systems that respect complexity rather than flatten it

  • Being explicit about what we know, what we don’t, and where boundaries exist

AI tools can produce errors, bias, or confident-sounding misinformation. CARL is intentionally constrained: it follows pedagogical rules, ethical guardrails, and validation checks rather than “freestyling” content. Teachers should review and adapt outputs, especially for sensitive topics, local context, and student-specific needs.


Indigenous Voice Inclusion

Respecting Sovereignty and Self-Determination

Indigenous peoples hold the inherent right to govern their own knowledge, cultures, and histories.

In keeping with principles of Indigenous data sovereignty and self-determination, AI systems, including CARL, must not generate, interpret, or prescribe Indigenous knowledge. Doing so risks appropriation, loss of context, and the erosion of cultural protocols.

CARL’s Indigenous Voice Inclusion feature is designed to support respectful classroom inclusion by pointing educators to Indigenous-authored works and verified, Indigenous-led resources, rather than providing content itself. This approach ensures that Indigenous voices remain with their rightful stewards and are encountered in their original context.

Key design principles of Indigenous Voice Inclusion

  • CARL is not a source of Indigenous knowledge

  • The feature surfaces specific authors, works, and Indigenous-led databases where authorship and provenance are clear

  • The Indigenous Voice Inclusion option is teacher-initiated and opt-in, designed to support respectful inclusion, not automate it
  • When certainty is not possible, CARL points teachers to trusted Indigenous-led collections rather than guessing

  • If no authentic connection exists, CARL names that gap rather than forcing a connection

This approach avoids extraction, appropriation, and unpaid educational labour, while still supporting educators in locating authentic Indigenous voices when they belong in the learning.

Giving Back

We are building a giving-back workflow connected to the Indigenous Voice Inclusion option. Our intention is to support community-led healing and remembrance work by donating to the Na-mi-quai-ni-mak Community Support Fund at the National Centre for Truth and Reconciliation.

This donation is not a substitute for relationship-building, guidance, or consent from Indigenous communities. It does not give CARL authority to generate or interpret Indigenous knowledge, and it does not replace the need for Indigenous-led leadership in this space.

Because CARL is not yet fully operational at scale, we are still finalizing the donation structure (amount, timing, and reporting cadence). Once implemented, we will publish a clear explanation of how donations are triggered and how totals are reported.

As we grow, we are committed to building relationships with Indigenous educators who can guide this work, ensuring it serves communities authentically. We will not claim partnerships or consultation we have not yet undertaken.


Learning, Limits, and Accountability

This work is ongoing.

We are continually learning, refining, and revisiting our design decisions. Meaningful consultation requires time, relationship-building, and resources. As a small team, we are working toward that future engagement and are careful not to overstate where we are today.

We welcome respectful dialogue and feedback at team@hicarl.ai or use our contact page.