Algorithm Charter 2020 and Kaha Create

How Kaha Create Complies: The Six Commitments

1. TRANSPARENCY

Maintain transparency by clearly explaining how decisions are informed by algorithms, including plain English documentation, making information about the data and processes available, and publishing information about how data are collected, secured and stored Data.

Kaha Create's Implementation:

✅ Plain Language AI Disclosure

• Clear labelling: "AI-generated summary" / "AI-detected cultural content"

• User-facing documentation explaining what each AI feature does

• No hidden algorithmic decision-making

✅ Process Transparency

• Published documentation on how Claude AI processes video content

• Clear explanation of AWS Bedrock infrastructure

• Open communication about AI model versions and capabilities

✅ Data Collection Disclosure

• Privacy policy clearly states what data is processed

• Transparent about AWS/Supabase/Stripe data handling

• Clear retention policies for video, transcripts, and user data

2. PARTNERSHIP (Te Ao Māori Perspective)

Deliver clear public benefit through Treaty commitments by embedding a Te Ao Māori perspective in the development and use of algorithms consistent with the principles of the Treaty of Waitangi.

Kaha Create's Implementation:

✅ Cultural Advisory Integration

• Mātauranga Māori detection system developed with Te Wānanga o Raukawa input

• Traditional Knowledge Labels integration (Local Contexts)

• Ongoing consultation with iwi partners on AI behaviour

✅ Te Tiriti Principles Application

• Partnership: Co-design of cultural safeguards with Māori organisations

• Protection: AWS Guardrails prevent inappropriate AI handling of cultural content

• Participation: Māori organisations control their own data sovereignty settings

✅ Te Reo Māori Integration

• Te Hiku Media partnership for accurate reo transcription

• Bilingual UI and content generation

• Cultural concept recognition across languages

3. PEOPLE (Active Engagement)

Focus on people by identifying and actively engaging with people, communities and groups who have an interest in algorithms, and consulting with those impacted by their use

Kaha Create's Implementation:

✅ User Persona-Specific Consultation

• Regular feedback loops with Cultural Knowledge Stewards

• Community Hub Connectors advisory group for feature development

• Pilot programmes with Te Matarau a Māui, Te Wānanga o Raukawa

✅ Impacted Community Engagement

• Rangatahi feedback on AI-generated learning content

• Kaumātua consultation on cultural content handling

• Creator surveys on AI accuracy and usefulness

✅ Transparent Development Process

• Roadmap shared publicly with "why this matters" explanations

• Beta testing with diverse user groups before features go live

• Open channels for algorithm concerns (Gleap support system)

4. DATA (Fit for Purpose & Bias Management)

Make sure data is fit for purpose by understanding its limitations and identifying and managing bias.

Kaha Create's Implementation:

✅ Known Limitations Documentation

• AI transcription accuracy rates disclosed (typically 85-95% for clear audio)

• Cultural concept recognition limitations acknowledged

• Language-specific performance variations documented (te reo Māori vs English)

✅ Bias Identification & Mitigation

• Training data bias: Claude trained predominantly on Western/English content - we supplement with Te Hiku Media for reo

• Cultural bias: Mātauranga Māori detection system reviewed by cultural advisors

• User testing: Diverse pilot groups prevent feature design bias

✅ Data Quality Standards

• Video quality requirements for optimal AI processing

• Audio clarity standards for transcription accuracy

• Metadata validation for content categorisation

5. PRIVACY, ETHICS & HUMAN RIGHTS

Ensure that privacy, ethics and human rights are safeguarded by regularly peer reviewing algorithms to assess for unintended consequences and act on this information

Kaha Create's Implementation:

✅ Regular Algorithm Review Process

• Quarterly cultural advisor review of mātauranga Māori detection accuracy

• Monthly analysis of AI-generated content quality metrics

• User feedback integration into AI prompt engineering

✅ Unintended Consequences Monitoring

• Track: False positives in cultural content detection

• Monitor: AI generating culturally inappropriate learning materials

• Assess: Over-reliance on AI reducing creator agency

✅ Ethical AI Framework

• No AI training on user content without explicit consent

• AWS Guardrails prevent harmful AI outputs

• Human-in-the-loop for all cultural content decisions

6. HUMAN OVERSIGHT

Retain human oversight by nominating a point of contact for public inquiries about algorithms, providing a channel for challenging or appealing decisions informed by algorithms, and clearly explaining the role of humans in decisions informed by algorithms.

Kaha Create's Implementation:

✅ Point of Contact

• Algorithm Oversight Contact: Pera Barrett (Founder), [email protected]

• Public-facing AI transparency page on website

• Clear escalation path for AI-related concerns

✅ Challenge & Appeal Process

• Users can flag AI-generated content as inaccurate

• Mātauranga Māori false positives can be contested

• Content access restrictions reviewable by human moderators

✅ Clear Human Role Explanation

• AI Role: Generates drafts, suggests structures, detects patterns

• Human Role: Reviews, approves, adds context, applies tikanga

• Critical Decisions: All cultural protocols determined by creators, not AI