docs: Rewrite Village case study — sovereign AI, polycentric governance, updated facts
Reframe page for AI Forum NZ audience. Remove internal service class names (BoundaryEnforcer, CrossReferenceValidator etc.), replace with plain English. Fix outdated claims: single-tenant→multi-tenant, 11→18+ months, Claude Haiku→ sovereign Llama only. Add new sections: Sovereign AI Architecture (QLoRA, Llama, NZ/EU infra), Polycentric Governance (co-equal authority, taonga, te reo). Fix CTA link to community.myfamilyhistory.digital. Update all i18n translations (DE, FR, MI) via DeepL. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
TheFlow
parent
004202ff47
commit
772f32008b
5 changed files with 593 additions and 670 deletions
|
|
@ -1,187 +1,167 @@
|
|||
{
|
||||
"breadcrumb": "Village Fallstudie",
|
||||
"breadcrumb": "Fallstudie Dorf",
|
||||
"hero": {
|
||||
"badge": "FALLSTUDIE",
|
||||
"title": "Village-Plattform: Tractatus in Produktion",
|
||||
"subtitle": "Wie Home AI alle sechs Tractatus Governance Services auf jede Benutzerinteraktion in einer Live-Community-Plattform anwendet.",
|
||||
"title": "Das Dorf: Souveräne Gemeinschaft AI in der Produktion",
|
||||
"subtitle": "Eine mandantenfähige Gemeinschaftsplattform, die ihr eigenes Sprachmodell auf ihrer eigenen Infrastruktur betreibt und durch eine polyzentrische Architektur gesteuert wird, in der die Gemeinschaften gleichberechtigte Autorität besitzen.",
|
||||
"disclaimer": {
|
||||
"label": "Wichtig:",
|
||||
"text": "Dies dokumentiert eine frühe Phase des föderierten Einsatzes durch den Entwickler des Frameworks auf vier Mandanten. Die Metriken werden selbst berichtet. Ein unabhängiges Audit und eine umfassendere Validierung sind geplant, wurden aber noch nicht durchgeführt."
|
||||
"label": "Das ist wichtig:",
|
||||
"text": "Dies dokumentiert eine frühe Phase der mandantenfähigen Bereitstellung durch den Entwickler des Frameworks. Die Metriken werden selbst berichtet. Ein unabhängiges Audit und eine umfassendere Validierung sind geplant, wurden aber noch nicht durchgeführt."
|
||||
}
|
||||
},
|
||||
"overview": {
|
||||
"heading": "Übersicht",
|
||||
"platform_purpose": {
|
||||
"title": "Plattform-Zweck",
|
||||
"description": "The Village ist eine Community-Plattform im Besitz der Mitglieder, die souveräne Datenspeicherung, KI-gestützte Funktionen und eine Privacy-by-Design-Architektur bietet. Gemeinschaften arbeiten mit vollem Dateneigentum und Governance-geschützter KI-Unterstützung."
|
||||
},
|
||||
"deployment_facts": {
|
||||
"title": "Fakten zum Einsatz",
|
||||
"what_is": {
|
||||
"heading": "Was ist das Dorf?",
|
||||
"description": "The Village ist eine mitgliedergeführte Plattform für Whānau, Marae, Clubs und Gemeinschaftsorganisationen. Jede Gemeinschaft erhält ihren eigenen, isolierten Mieter mit souveräner Datenspeicherung, KI-gestützten Funktionen und durch die Verwaltung geschützter Privatsphäre. Die Plattform unterstützt durchgehend te reo Māori.",
|
||||
"description_2": "Die gesamte KI-Verarbeitung läuft über die plattformeigene Infrastruktur - ein lokal abgestimmtes Llama-Modell, ohne dass Daten an externe KI-Anbieter gesendet werden. Die Gemeinschaften sind Eigentümer der Daten und können ihre Zustimmung jederzeit zurückziehen.",
|
||||
"facts_title": "Fakten zum Einsatz",
|
||||
"facts": {
|
||||
"duration_label": "Dauer:",
|
||||
"duration_value": "11+ Monate in Produktion",
|
||||
"tenant_label": "Mandanten-Modell:",
|
||||
"tenant_value": "Single-Tenant (Multi-Tenant geplant)",
|
||||
"features_label": "KI-Funktionen:",
|
||||
"duration_value": "18+ Monate in der Produktion",
|
||||
"tenant_label": "Mieter-Modell:",
|
||||
"tenant_value": "Multi-Tenant (mehrere Gemeinschaften)",
|
||||
"ai_label": "AI-Modell:",
|
||||
"ai_value": "Souveränes Lama (QLoRA feinabgestimmt)",
|
||||
"features_label": "AI-Merkmale:",
|
||||
"features_value": "4 geregelte Funktionen live",
|
||||
"services_label": "Dienste/Antwort:",
|
||||
"services_value": "6 Governance-Prüfungen"
|
||||
"infra_label": "Infrastruktur:",
|
||||
"infra_value": "NZ + EU (keine US-Abhängigkeit)"
|
||||
}
|
||||
},
|
||||
"architecture_mapping": {
|
||||
"heading": "Architektur-Mapping",
|
||||
"description": "Jede Village KI-Funktion ist bestimmten Tractatus-Governance-Diensten zugeordnet. Die folgende Tabelle zeigt, wie die sechs Dienste für jede Funktion koordiniert werden.",
|
||||
"table": {
|
||||
"col_feature": "Village-Funktion",
|
||||
"col_service": "Primärer Dienst",
|
||||
"col_function": "Funktion",
|
||||
"rows": [
|
||||
{
|
||||
"feature": "Home AI-Antworten",
|
||||
"service": "BoundaryEnforcer",
|
||||
"function": "Blockiert Werturteile, überlässt es dem Menschen"
|
||||
"sovereign": {
|
||||
"heading": "Souveräne KI-Architektur",
|
||||
"intro": "Das Dorf führt sein eigenes Sprachmodell aus - kein API-Aufruf an einen US-Hyperscaler, sondern ein lokal abgestimmtes Modell, bei dem die Trainingsdaten, die Modellgewichte und die Inferenzpipeline unter der Kontrolle der Gemeinschaft bleiben.",
|
||||
"model_title": "Lokales Sprachmodell",
|
||||
"model_desc": "Llama 3.1 8B und Llama 3.2 3B, feinabgestimmt mit QLoRA auf gemeinschaftsspezifische Daten. Alle Inferenzen laufen auf der plattformeigenen GPU-Infrastruktur.",
|
||||
"infra_title": "Souveräne Infrastruktur",
|
||||
"infra_desc": "Produktionsserver in Neuseeland und in der EU. Keine Daten durchqueren die US-Gerichtsbarkeit. Die Daten der Gemeinschaft verlassen nie die Einrichtung, zu der sie gehören.",
|
||||
"training_title": "Von der Gemeinschaft kontrollierte Ausbildung",
|
||||
"training_desc": "QLoRA-Feinabstimmung auf bereichsspezifischen Daten mit Zustimmungsverfolgung und Herkunftsnachweis. Gemeinschaften können Trainingsdaten zurückziehen und eine Modellumschulung auslösen.",
|
||||
"link_note": "Eine ausführliche Beschreibung der Modellarchitektur, des Trainingsansatzes und der Governance-Integration finden Sie unter <a href=\"/home-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">Home AI / SLL: Sovereign Locally-Trained Language Model</a>."
|
||||
},
|
||||
{
|
||||
"feature": "Verarbeitung von Benutzeranfragen",
|
||||
"service": "CrossReferenceValidator",
|
||||
"function": "Verhindert Prompt-Injection, validiert Absicht"
|
||||
"polycentric": {
|
||||
"heading": "Polyzentrische Governance",
|
||||
"intro": "Der besondere Beitrag des Village ist seine Governance-Architektur. Anstatt dass ein einziger Betreiber alle Entscheidungen trifft, implementiert die Plattform eine polyzentrische Governance - mehrere gleichberechtigte Behörden, die sich die strukturelle Kontrolle darüber teilen, wie KI eingesetzt wird.",
|
||||
"coequal_title": "Gleichberechtigte Autorität",
|
||||
"coequal_desc": "Die Gemeinschaften behalten die architektonische Mitbestimmung - nicht nur das Recht auf Konsultation, sondern auch die strukturelle Autorität über die Verwendung ihrer Daten. Abgeleitet von te ao Māori-Konzepten wie rangatiratanga (Selbstbestimmung) und kaitiakitanga (Vormundschaft).",
|
||||
"nonparticipation_title": "Recht auf Nichtteilnahme",
|
||||
"nonparticipation_desc": "Mitglieder können sich von jeder KI-Funktion abmelden, ohne den Zugang zur Plattform zu verlieren. Die KI-Governance überlässt die Entscheidung über Wertefragen den Menschen und setzt sich niemals über die Entscheidungen der Gemeinschaft hinweg.",
|
||||
"taonga_title": "Taonga-zentriertes Design",
|
||||
"taonga_desc": "Kulturschätze (taonga) werden als Objekte erster Klasse mit Herkunftsnachweis, Rücknahmerechten und gemeinschaftlicher Autorität darüber, wie sie in AI-Kontexten erscheinen, verwaltet.",
|
||||
"tenant_title": "Mieter-bezogene Isolierung",
|
||||
"tenant_desc": "Jede Gemeinschaft arbeitet in vollständiger Datenisolierung. Keine gemeinsame Nutzung von Daten durch andere Mieter. Die Governance-Entscheidungen eines jeden Mieters gelten nur innerhalb seiner eigenen Grenzen.",
|
||||
"paper_note": "Die Forschungsgrundlage wird beschrieben in <a href=\"/downloads/taonga-centred-steering-governance-polycentric-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">Taonga-Centred Steering Governance: Polyzentrische KI für indigene Datensouveränität</a>."
|
||||
},
|
||||
{
|
||||
"feature": "Session-Management",
|
||||
"service": "ContextPressureMonitor",
|
||||
"function": "Verfolgt den Sitzungszustand, löst Übergaben aus"
|
||||
"flow": {
|
||||
"heading": "Wie Governance in der Praxis funktioniert",
|
||||
"intro": "Wenn ein Mitglied eine KI-Funktion nutzt, durchläuft die Anfrage sechs Governance-Prüfungen, bevor eine Antwort erzeugt wird. Jede Prüfung ist unabhängig und kann die Anfrage blockieren oder ändern.",
|
||||
"steps": {
|
||||
"1": {
|
||||
"title": "Antrag eines Mitglieds erhalten",
|
||||
"desc": "Ein Mitglied bittet um Hilfe, bittet um OCR oder nutzt die Hilfe von Geschichten."
|
||||
},
|
||||
{
|
||||
"feature": "Mehrstufige Operationen",
|
||||
"service": "MetacognitiveVerifier",
|
||||
"function": "Erkennt Scope-Creep, erfordert Überprüfung"
|
||||
"2": {
|
||||
"title": "Prüfung der Wertgrenzen",
|
||||
"desc": "Handelt es sich um eine Wertefrage, die ein menschliches Urteil erfordert? Wenn ja, zögert die KI, anstatt zu antworten."
|
||||
},
|
||||
{
|
||||
"feature": "Feature-Flags",
|
||||
"service": "InstructionPersistenceClassifier",
|
||||
"function": "Persistenzklassifizierung für Einstellungen"
|
||||
"3": {
|
||||
"title": "Validierung von Absichten",
|
||||
"desc": "Steht die Anfrage im Widerspruch zu den gespeicherten Governance-Regeln oder versucht sie eine Prompt Injection? Querverweise gegen bekannte Befehlssätze."
|
||||
},
|
||||
{
|
||||
"feature": "Einwilligungshandhabung",
|
||||
"service": "PluralisticDeliberationOrchestrator",
|
||||
"function": "Multi-Stakeholder-Entscheidungen"
|
||||
"4": {
|
||||
"title": "Kontext und Zustand der Sitzung",
|
||||
"desc": "Befindet sich die Sitzung in einem akzeptablen Rahmen? Überwacht den Kontextdruck und löst bei Bedarf eine ordnungsgemäße Übergabe aus."
|
||||
},
|
||||
"5": {
|
||||
"title": "Genehmigungsgefilterte Abfrage und Antwort",
|
||||
"desc": "Das souveräne Llama-Modell erzeugt eine Antwort unter Verwendung des RAG-Kontextes, gefiltert nach den Berechtigungen des Mitglieds. Die gesamte Verarbeitung bleibt in der Infrastruktur."
|
||||
},
|
||||
"6": {
|
||||
"title": "Überprüfung des Umfangs",
|
||||
"desc": "Entspricht die Antwort dem, was angefordert wurde? Erkennt die Ausweitung des Umfangs und blockiert Antworten, die über die ursprüngliche Anfrage hinausgehen."
|
||||
},
|
||||
"7": {
|
||||
"title": "Lieferung mit Namensnennung",
|
||||
"desc": "Die Antwort wird dem Mitglied mit Quellenangabe zugestellt. Jeder Schritt wird für die Prüfung protokolliert."
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
"home_ai_flow": {
|
||||
"heading": "Der Home AI Flow",
|
||||
"description": "Wenn ein Benutzer eine Anfrage an Home AI stellt, durchläuft diese sechs Überprüfungsstufen, bevor eine Antwort generiert wird. Dieser Ablauf erfolgt auf dem kritischen Ausführungspfad.",
|
||||
"steps": [
|
||||
{
|
||||
"title": "Benutzeranfrage empfangen",
|
||||
"description": "Benutzer stellt Anfrage über Help-Chat-Widget oder Story-Assistenz"
|
||||
"features": {
|
||||
"heading": "Was die Plattform bietet",
|
||||
"help": {
|
||||
"title": "Hilfe-Center",
|
||||
"benefit": "Mitglieder stellen Fragen in natürlicher Sprache und erhalten Antworten aus Hilfeinhalten, Beiträgen und Dokumentationen - gefiltert nach ihren Berechtigungen.",
|
||||
"governance": "Steuerung: Die Überprüfung der Wertgrenzen verhindert, dass die KI Urteile fällt; die Absichtsüberprüfung blockiert prompte Injektionsversuche."
|
||||
},
|
||||
{
|
||||
"title": "BoundaryEnforcer-Prüfung",
|
||||
"description": "Ist dies eine Wertefrage, die menschliches Urteil erfordert?"
|
||||
},
|
||||
{
|
||||
"title": "CrossReferenceValidator-Prüfung",
|
||||
"description": "Steht dies im Widerspruch zu gespeicherten Anweisungen oder versucht es Injection?"
|
||||
},
|
||||
{
|
||||
"title": "ContextPressureMonitor-Prüfung",
|
||||
"description": "Liegt der Sitzungszustand in akzeptablen Grenzen?"
|
||||
},
|
||||
{
|
||||
"title": "Abfrageverarbeitung",
|
||||
"description": "RAG-System ruft Kontext mit Berechtigungsfilterung ab"
|
||||
},
|
||||
{
|
||||
"title": "Antwortgenerierung",
|
||||
"description": "KI generiert Antwort (Claude Haiku für Nicht-EN, lokales Llama für EN)"
|
||||
},
|
||||
{
|
||||
"title": "MetacognitiveVerifier-Prüfung",
|
||||
"description": "Ist die Antwort dem Umfang der Anfrage angemessen?"
|
||||
},
|
||||
{
|
||||
"title": "Auslieferung",
|
||||
"description": "Antwort an den Nutzer mit Quellenangabe"
|
||||
}
|
||||
]
|
||||
},
|
||||
"governed_features": {
|
||||
"heading": "Geregelte Funktionen im Detail",
|
||||
"rag_help": {
|
||||
"title": "RAG-basiertes Hilfezentrum",
|
||||
"description": "Vektorsuche über indizierte Hilfeinhalte, Beiträge und Dokumentation. Ergebnisse werden vor Aufnahme in den Kontext nach Benutzerrechten gefiltert.",
|
||||
"services": "CrossReferenceValidator, BoundaryEnforcer"
|
||||
},
|
||||
"document_ocr": {
|
||||
"ocr": {
|
||||
"title": "Dokument-OCR",
|
||||
"description": "Automatisierte Textextraktion aus hochgeladenen Dokumenten. Arbeitet mit expliziter Einwilligungskontrolle und Audit-Protokollierung.",
|
||||
"services": "PluralisticDeliberationOrchestrator (Einwilligung)"
|
||||
"benefit": "Laden Sie ein Dokument hoch und lassen Sie den Text automatisch extrahieren. Nützlich für das Digitalisieren von Briefen, Urkunden und historischen Aufzeichnungen.",
|
||||
"governance": "Verwaltung: Vor der Verarbeitung ist eine ausdrückliche Zustimmung erforderlich. Alle Vorgänge werden mit vollständiger Provenienz protokolliert."
|
||||
},
|
||||
"story_assistance": {
|
||||
"title": "Geschichten-Assistenz",
|
||||
"description": "KI-unterstützte Schreibvorschläge für Familiengeschichten. Inhaltsvorschläge werden durch BoundaryEnforcer gefiltert, um unangemessene Empfehlungen zu verhindern.",
|
||||
"services": "BoundaryEnforcer, MetacognitiveVerifier"
|
||||
"stories": {
|
||||
"title": "Assistenz bei Geschichten",
|
||||
"benefit": "KI-unterstützte Schreibanregungen für Gemeinschaftsgeschichten und Familiengeschichten. Hilft bei der Strukturierung, bei Aufforderungen und beim sanften Editieren.",
|
||||
"governance": "Steuerung: Die Überprüfung der Wertgrenzen verhindert unangemessene Inhaltsvorschläge; die Überprüfung des Umfangs stellt sicher, dass die KI innerhalb der gestellten Fragen bleibt."
|
||||
},
|
||||
"ai_memory": {
|
||||
"title": "KI-Speicher-Transparenz",
|
||||
"description": "Benutzergesteuerte Zusammenfassung mit vollständigem Audit-Dashboard. Mitglieder können anzeigen, bearbeiten und löschen, was die KI über sie speichert.",
|
||||
"services": "InstructionPersistenceClassifier, PluralisticDeliberationOrchestrator"
|
||||
},
|
||||
"services_label": "Dienste:"
|
||||
"memory": {
|
||||
"title": "AI-Speicher-Transparenz",
|
||||
"benefit": "Mitglieder können sehen, bearbeiten und löschen, was sich die KI über sie \"merkt\". Das vollständige Audit-Dashboard zeigt jede KI-Interaktion an.",
|
||||
"governance": "Steuerung: Zustimmung mehrerer Interessengruppen erforderlich. Persistenzentscheidungen werden klassifiziert und sind überprüfbar. Die Mitglieder kontrollieren ihre eigenen Daten."
|
||||
}
|
||||
},
|
||||
"limitations": {
|
||||
"heading": "Ehrliche Einschränkungen",
|
||||
"intro": "Diese Fallstudie dokumentiert vorläufige Erkenntnisse aus einer frühen Phase der föderierten Bereitstellung bei vier Mietern. Wir sind uns über die folgenden Einschränkungen im Klaren:",
|
||||
"heading": "Ehrliche Beschränkungen",
|
||||
"intro": "Diese Fallstudie dokumentiert vorläufige Erkenntnisse aus einem mandantenfähigen Produktionseinsatz. Wir sind uns über die folgenden Einschränkungen im Klaren:",
|
||||
"items": [
|
||||
{
|
||||
"title": "Begrenzter Einsatz:",
|
||||
"description": "Village arbeitet über vier föderierte Mieter innerhalb einer Plattform. Die Übertragbarkeit auf andere Kontexte ist nicht bekannt."
|
||||
"title": "Kleiner Maßstab:",
|
||||
"description": "Das Dorf bedient derzeit eine kleine Anzahl von Gemeinschaftsmietern. Die Verallgemeinerbarkeit auf größere Anlagen oder andere Gemeindetypen ist unbekannt."
|
||||
},
|
||||
{
|
||||
"title": "Selbst berichtete Metriken:",
|
||||
"title": "Selbstberichtete Metriken:",
|
||||
"description": "Eine unabhängige Überprüfung der protokollierten Daten hat nicht stattgefunden."
|
||||
},
|
||||
{
|
||||
"title": "Überschneidung Betreiber-Entwickler:",
|
||||
"description": "Der Framework-Entwickler betreibt auch Village (Interessenkonflikt)."
|
||||
"title": "Überschneidungen zwischen Bedienern und Entwicklern:",
|
||||
"description": "Der Rahmenentwickler betreibt auch das Dorf (Interessenkonflikt)."
|
||||
},
|
||||
{
|
||||
"title": "Begrenzte Adversarial-Tests:",
|
||||
"description": "Es wurde keine formelle Red-Team-Bewertung durchgeführt."
|
||||
"title": "Begrenzte kontradiktorische Tests:",
|
||||
"description": "Es wurde keine formelle Bewertung des roten Teams durchgeführt."
|
||||
},
|
||||
{
|
||||
"title": "Freiwillige Nutzung:",
|
||||
"title": "Freiwillige Anrufung:",
|
||||
"description": "KI könnte theoretisch die Governance umgehen, wenn sie nicht dafür konfiguriert ist."
|
||||
}
|
||||
]
|
||||
},
|
||||
"demonstrates": {
|
||||
"heading": "Was dies demonstriert",
|
||||
"heading": "Was dies beweist",
|
||||
"supports": {
|
||||
"title": "Evidenz unterstützt",
|
||||
"title": "Beweise unterstützen",
|
||||
"items": [
|
||||
"Architektur-Governance kann in Produktion ohne übermäßigen Aufwand betrieben werden",
|
||||
"Die Koordinierung von sechs Diensten ist technisch machbar",
|
||||
"Governance-Verstöße sind nachweisbar und überprüfbar",
|
||||
"Das Framework lernt aus Fehlern (dokumentierte Vorfallreaktionen)"
|
||||
"• Der Einsatz von souveräner KI ist für kleine Gemeinschaftsorganisationen technisch machbar",
|
||||
"• Eine polyzentrische Verwaltung kann in der Produktion ohne übermäßigen Aufwand betrieben werden",
|
||||
"• Mandantenübergreifende Isolierung mit gemeinschaftsbezogener Governance ist möglich",
|
||||
"• Governance-Verstöße sind aufdeckbar und überprüfbar",
|
||||
"• Der Rahmen lernt aus Fehlern (dokumentierte Reaktionen auf Vorfälle)"
|
||||
]
|
||||
},
|
||||
"does_not_support": {
|
||||
"title": "Evidenz unterstützt NICHT",
|
||||
"title": "Beweise stützen NICHT",
|
||||
"items": [
|
||||
"Framework-Effektivität im großen Maßstab (Tausende gleichzeitiger Nutzer)",
|
||||
"Generalisierbarkeit über verschiedene KI-Systeme hinweg",
|
||||
"Widerstandsfähigkeit gegen ausgeklügelte Adversarial-Angriffe",
|
||||
"Regulatorische Eignung (EU-KI-Gesetz-Konformität ungeprüft)"
|
||||
"• Effektivität des Rahmens im großen Maßstab (Tausende von gleichzeitigen Nutzern)",
|
||||
"• Verallgemeinerbarkeit über verschiedene KI-Systeme oder Modellarchitekturen hinweg",
|
||||
"• Widerstandsfähigkeit gegen ausgeklügelte feindliche Angriffe",
|
||||
"• Angemessenheit der Vorschriften (Einhaltung des EU-AI-Gesetzes ungeprüft)"
|
||||
]
|
||||
}
|
||||
},
|
||||
"cta": {
|
||||
"heading": "Weiter erkunden",
|
||||
"description": "Sehen Sie die Village-Plattform in Aktion oder erfahren Sie mehr über die technische Architektur.",
|
||||
"visit_village": "Besuchen Sie Village →",
|
||||
"system_architecture": "Systemarchitektur →",
|
||||
"research_details": "Forschungsdetails →"
|
||||
"heading": "Weiter erforschen",
|
||||
"description": "Erfahren Sie mehr über die technische Architektur, lesen Sie die Forschungsergebnisse oder sehen Sie sich die Village-Plattform in Aktion an.",
|
||||
"visit_village": "Besuchen Sie das Dorf →",
|
||||
"home_ai": "Souveränes Sprachmodell →",
|
||||
"research_paper": "Forschungspapier →",
|
||||
"research_details": "Forschung → Details"
|
||||
}
|
||||
}
|
||||
|
|
@ -2,141 +2,119 @@
|
|||
"breadcrumb": "Village Case Study",
|
||||
"hero": {
|
||||
"badge": "CASE STUDY",
|
||||
"title": "Village Platform: Tractatus in Production",
|
||||
"subtitle": "How Home AI applies all six Tractatus governance services to every user interaction in a live community platform.",
|
||||
"title": "The Village: Sovereign Community AI in Production",
|
||||
"subtitle": "A multi-tenant community platform running its own language model on its own infrastructure, governed by a polycentric architecture where communities hold co-equal authority.",
|
||||
"disclaimer": {
|
||||
"label": "Important:",
|
||||
"text": "This documents an early-stage federated deployment across four tenants by the framework developer. Metrics are self-reported. Independent audit and broader validation are planned but not yet conducted."
|
||||
"text": "This documents an early-stage multi-tenant deployment by the framework developer. Metrics are self-reported. Independent audit and broader validation are planned but not yet conducted."
|
||||
}
|
||||
},
|
||||
"overview": {
|
||||
"heading": "Overview",
|
||||
"platform_purpose": {
|
||||
"title": "Platform Purpose",
|
||||
"description": "The Village is a member-owned community platform providing sovereign data storage, AI-assisted features, and privacy-by-design architecture. Communities operate with full data ownership and governance-protected AI assistance."
|
||||
},
|
||||
"deployment_facts": {
|
||||
"title": "Deployment Facts",
|
||||
"what_is": {
|
||||
"heading": "What Is the Village",
|
||||
"description": "The Village is a member-owned platform for whānau, marae, clubs, and community organisations. Each community gets its own isolated tenant with sovereign data storage, AI-assisted features, and governance-protected privacy. The platform supports te reo Māori throughout.",
|
||||
"description_2": "All AI processing runs on the platform's own infrastructure — a locally fine-tuned Llama model with no data sent to external AI providers. Communities operate with full data ownership and can withdraw consent at any time.",
|
||||
"facts_title": "Deployment Facts",
|
||||
"facts": {
|
||||
"duration_label": "Duration:",
|
||||
"duration_value": "11+ months in production",
|
||||
"duration_value": "18+ months in production",
|
||||
"tenant_label": "Tenant Model:",
|
||||
"tenant_value": "Single-tenant (multi-tenant planned)",
|
||||
"tenant_value": "Multi-tenant (multiple communities)",
|
||||
"ai_label": "AI Model:",
|
||||
"ai_value": "Sovereign Llama (QLoRA fine-tuned)",
|
||||
"features_label": "AI Features:",
|
||||
"features_value": "4 governed features live",
|
||||
"services_label": "Services/Response:",
|
||||
"services_value": "6 governance checks"
|
||||
"infra_label": "Infrastructure:",
|
||||
"infra_value": "NZ + EU (no US dependency)"
|
||||
}
|
||||
},
|
||||
"architecture_mapping": {
|
||||
"heading": "Architecture Mapping",
|
||||
"description": "Each Village AI feature maps to specific Tractatus governance services. The table below shows how the six services coordinate for each feature.",
|
||||
"table": {
|
||||
"col_feature": "Village Feature",
|
||||
"col_service": "Primary Service",
|
||||
"col_function": "Function",
|
||||
"rows": [
|
||||
{
|
||||
"feature": "Home AI Responses",
|
||||
"service": "BoundaryEnforcer",
|
||||
"function": "Blocks values judgments, defers to human"
|
||||
"sovereign": {
|
||||
"heading": "Sovereign AI Architecture",
|
||||
"intro": "The Village runs its own language model — not an API call to a US hyperscaler, but a locally fine-tuned model where the training data, model weights, and inference pipeline all remain under community control.",
|
||||
"model_title": "Local Language Model",
|
||||
"model_desc": "Llama 3.1 8B and Llama 3.2 3B, fine-tuned with QLoRA on community-specific data. All inference runs on the platform's own GPU infrastructure.",
|
||||
"infra_title": "Sovereign Infrastructure",
|
||||
"infra_desc": "Production servers in New Zealand and the EU. No data transits US jurisdiction. Community data never leaves the deployment it belongs to.",
|
||||
"training_title": "Community-Controlled Training",
|
||||
"training_desc": "QLoRA fine-tuning on domain-specific data with consent tracking and provenance. Communities can withdraw training data and trigger model retraining.",
|
||||
"link_note": "For a detailed account of the model architecture, training approach, and governance integration, see <a href=\"/home-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">Home AI / SLL: Sovereign Locally-Trained Language Model</a>."
|
||||
},
|
||||
{
|
||||
"feature": "User Query Processing",
|
||||
"service": "CrossReferenceValidator",
|
||||
"function": "Prevents prompt injection, validates intent"
|
||||
"polycentric": {
|
||||
"heading": "Polycentric Governance",
|
||||
"intro": "The distinctive contribution of the Village is its governance architecture. Rather than a single operator making all decisions, the platform implements polycentric governance — multiple co-equal authorities that share structural control over how AI is used.",
|
||||
"coequal_title": "Co-Equal Authority",
|
||||
"coequal_desc": "Communities maintain architectural co-governance — not just consultation rights, but structural authority over how their data is used. Drawn from te ao Māori concepts of rangatiratanga (self-determination) and kaitiakitanga (guardianship).",
|
||||
"nonparticipation_title": "Right of Non-Participation",
|
||||
"nonparticipation_desc": "Members can opt out of any AI feature without losing access to the platform. AI governance defers to human judgment on values questions and never overrides community decisions.",
|
||||
"taonga_title": "Taonga-Centred Design",
|
||||
"taonga_desc": "Cultural treasures (taonga) are governed as first-class objects with provenance tracking, withdrawal rights, and community authority over how they appear in AI contexts.",
|
||||
"tenant_title": "Tenant-Scoped Isolation",
|
||||
"tenant_desc": "Each community operates in complete data isolation. No cross-tenant data sharing. Each tenant's governance decisions apply only within their own boundary.",
|
||||
"paper_note": "The research foundation is described in <a href=\"/downloads/taonga-centred-steering-governance-polycentric-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">Taonga-Centred Steering Governance: Polycentric AI for Indigenous Data Sovereignty</a>."
|
||||
},
|
||||
{
|
||||
"feature": "Session Management",
|
||||
"service": "ContextPressureMonitor",
|
||||
"function": "Tracks session health, triggers handoffs"
|
||||
"flow": {
|
||||
"heading": "How Governance Works in Practice",
|
||||
"intro": "When a member uses any AI feature, the request passes through six governance checks before a response is generated. Each check is independent and can block or modify the request.",
|
||||
"steps": {
|
||||
"1": {
|
||||
"title": "Member request received",
|
||||
"desc": "A member asks for help, requests OCR, or uses story assistance."
|
||||
},
|
||||
{
|
||||
"feature": "Multi-Step Operations",
|
||||
"service": "MetacognitiveVerifier",
|
||||
"function": "Detects scope creep, requires review"
|
||||
"2": {
|
||||
"title": "Values boundary check",
|
||||
"desc": "Is this a values question that requires human judgment? If so, the AI defers rather than answering."
|
||||
},
|
||||
{
|
||||
"feature": "Feature Flags",
|
||||
"service": "InstructionPersistenceClassifier",
|
||||
"function": "Persistence classification for settings"
|
||||
"3": {
|
||||
"title": "Intent validation",
|
||||
"desc": "Does the request conflict with stored governance rules or attempt prompt injection? Cross-references against known instruction sets."
|
||||
},
|
||||
{
|
||||
"feature": "Consent Handling",
|
||||
"service": "PluralisticDeliberationOrchestrator",
|
||||
"function": "Multi-stakeholder decisions"
|
||||
"4": {
|
||||
"title": "Context and session health",
|
||||
"desc": "Is the session within acceptable bounds? Monitors for context pressure and triggers graceful handoff when needed."
|
||||
},
|
||||
"5": {
|
||||
"title": "Permission-filtered retrieval and response",
|
||||
"desc": "The sovereign Llama model generates a response using RAG context filtered by the member's permissions. All processing stays on-infrastructure."
|
||||
},
|
||||
"6": {
|
||||
"title": "Scope verification",
|
||||
"desc": "Is the response appropriate to what was asked? Detects scope creep and blocks responses that exceed the original request."
|
||||
},
|
||||
"7": {
|
||||
"title": "Delivery with attribution",
|
||||
"desc": "Response delivered to the member with source attribution. Every step is logged for audit."
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
"home_ai_flow": {
|
||||
"heading": "The Home AI Flow",
|
||||
"description": "When a user submits a query to Home AI, it passes through six verification stages before a response is generated. This flow operates in the critical execution path.",
|
||||
"steps": [
|
||||
{
|
||||
"title": "User Query Received",
|
||||
"description": "User submits query via Help Chat widget or story assistance"
|
||||
"features": {
|
||||
"heading": "What the Platform Delivers",
|
||||
"help": {
|
||||
"title": "Help Centre",
|
||||
"benefit": "Members ask questions in natural language and get answers drawn from help content, stories, and documentation — filtered by their permissions.",
|
||||
"governance": "Governance: Values boundary check prevents AI from making judgments; intent validation blocks prompt injection attempts."
|
||||
},
|
||||
{
|
||||
"title": "BoundaryEnforcer Check",
|
||||
"description": "Is this a values question requiring human judgment?"
|
||||
},
|
||||
{
|
||||
"title": "CrossReferenceValidator Check",
|
||||
"description": "Does this conflict with stored instructions or attempt injection?"
|
||||
},
|
||||
{
|
||||
"title": "ContextPressureMonitor Check",
|
||||
"description": "Is session health within acceptable bounds?"
|
||||
},
|
||||
{
|
||||
"title": "Query Processing",
|
||||
"description": "RAG system retrieves context with permission filtering"
|
||||
},
|
||||
{
|
||||
"title": "Response Generation",
|
||||
"description": "AI generates response (Claude Haiku for non-EN, local Llama for EN)"
|
||||
},
|
||||
{
|
||||
"title": "MetacognitiveVerifier Check",
|
||||
"description": "Is response appropriate to query scope?"
|
||||
},
|
||||
{
|
||||
"title": "Delivery",
|
||||
"description": "Response delivered to user with source attribution"
|
||||
}
|
||||
]
|
||||
},
|
||||
"governed_features": {
|
||||
"heading": "Governed Features in Detail",
|
||||
"rag_help": {
|
||||
"title": "RAG-Based Help Centre",
|
||||
"description": "Vector search over indexed help content, stories, and documentation. Results filtered by user permissions before inclusion in context.",
|
||||
"services": "CrossReferenceValidator, BoundaryEnforcer"
|
||||
},
|
||||
"document_ocr": {
|
||||
"ocr": {
|
||||
"title": "Document OCR",
|
||||
"description": "Automated text extraction from uploaded documents. Operates under explicit consent controls with audit logging.",
|
||||
"services": "PluralisticDeliberationOrchestrator (consent)"
|
||||
"benefit": "Upload a document and get the text extracted automatically. Useful for digitising letters, certificates, and historical records.",
|
||||
"governance": "Governance: Requires explicit consent before processing. All operations are audit-logged with full provenance."
|
||||
},
|
||||
"story_assistance": {
|
||||
"stories": {
|
||||
"title": "Story Assistance",
|
||||
"description": "AI-assisted writing suggestions for family stories. Content suggestions filtered through BoundaryEnforcer to prevent inappropriate recommendations.",
|
||||
"services": "BoundaryEnforcer, MetacognitiveVerifier"
|
||||
"benefit": "AI-assisted writing suggestions for community stories and family histories. Helps with structure, prompts, and gentle editing.",
|
||||
"governance": "Governance: Values boundary check prevents inappropriate content suggestions; scope verification ensures the AI stays within what was asked."
|
||||
},
|
||||
"ai_memory": {
|
||||
"memory": {
|
||||
"title": "AI Memory Transparency",
|
||||
"description": "User-controlled summarisation with full audit dashboard. Members can view, edit, and delete what AI \"remembers\" about them.",
|
||||
"services": "InstructionPersistenceClassifier, PluralisticDeliberationOrchestrator"
|
||||
},
|
||||
"services_label": "Services:"
|
||||
"benefit": "Members can see, edit, and delete what the AI \"remembers\" about them. Full audit dashboard shows every AI interaction.",
|
||||
"governance": "Governance: Multi-stakeholder consent required. Persistence decisions classified and auditable. Members control their own data."
|
||||
}
|
||||
},
|
||||
"limitations": {
|
||||
"heading": "Honest Limitations",
|
||||
"intro": "This case study documents preliminary evidence from an early-stage federated deployment across four tenants. We are transparent about the following limitations:",
|
||||
"intro": "This case study documents preliminary evidence from a production multi-tenant deployment. We are transparent about the following limitations:",
|
||||
"items": [
|
||||
{
|
||||
"title": "Limited Deployment:",
|
||||
"description": "Village operates across four federated tenants within one platform. Generalisability to other contexts is unknown."
|
||||
"title": "Small Scale:",
|
||||
"description": "The Village currently serves a small number of community tenants. Generalisability to larger deployments or different community types is unknown."
|
||||
},
|
||||
{
|
||||
"title": "Self-Reported Metrics:",
|
||||
|
|
@ -144,7 +122,7 @@
|
|||
},
|
||||
{
|
||||
"title": "Operator-Developer Overlap:",
|
||||
"description": "Framework developer also operates Village (conflict of interest)."
|
||||
"description": "Framework developer also operates the Village (conflict of interest)."
|
||||
},
|
||||
{
|
||||
"title": "Limited Adversarial Testing:",
|
||||
|
|
@ -161,27 +139,29 @@
|
|||
"supports": {
|
||||
"title": "Evidence Supports",
|
||||
"items": [
|
||||
"Architectural governance can operate in production without prohibitive overhead",
|
||||
"Six-service coordination is technically feasible",
|
||||
"Governance violations are detectable and auditable",
|
||||
"The framework learns from failures (documented incident responses)"
|
||||
"• Sovereign AI deployment is technically feasible for small community organisations",
|
||||
"• Polycentric governance can operate in production without prohibitive overhead",
|
||||
"• Multi-tenant isolation with per-community governance is achievable",
|
||||
"• Governance violations are detectable and auditable",
|
||||
"• The framework learns from failures (documented incident responses)"
|
||||
]
|
||||
},
|
||||
"does_not_support": {
|
||||
"title": "Evidence Does NOT Support",
|
||||
"items": [
|
||||
"Framework effectiveness at scale (thousands of concurrent users)",
|
||||
"Generalisability across different AI systems",
|
||||
"Resistance to sophisticated adversarial attacks",
|
||||
"Regulatory sufficiency (EU AI Act compliance untested)"
|
||||
"• Framework effectiveness at scale (thousands of concurrent users)",
|
||||
"• Generalisability across different AI systems or model architectures",
|
||||
"• Resistance to sophisticated adversarial attacks",
|
||||
"• Regulatory sufficiency (EU AI Act compliance untested)"
|
||||
]
|
||||
}
|
||||
},
|
||||
"cta": {
|
||||
"heading": "Explore Further",
|
||||
"description": "See the Village platform in action, or dive deeper into the technical architecture.",
|
||||
"description": "Dive deeper into the technical architecture, read the research, or see the Village platform in action.",
|
||||
"visit_village": "Visit the Village →",
|
||||
"system_architecture": "System Architecture →",
|
||||
"home_ai": "Sovereign Language Model →",
|
||||
"research_paper": "Research Paper →",
|
||||
"research_details": "Research Details →"
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -1,154 +1,132 @@
|
|||
{
|
||||
"breadcrumb": "Étude de cas Village",
|
||||
"breadcrumb": "Étude de cas d'un village",
|
||||
"hero": {
|
||||
"badge": "ÉTUDE DE CAS",
|
||||
"title": "Plateforme Village : Tractatus en production",
|
||||
"subtitle": "Comment Home AI applique les six services de gouvernance Tractatus à chaque interaction utilisateur dans une plateforme communautaire en direct.",
|
||||
"title": "Le Village : L'IA de la Communauté souveraine en production",
|
||||
"subtitle": "Une plateforme communautaire multi-locataires utilisant son propre modèle linguistique sur sa propre infrastructure, régie par une architecture polycentrique où les communautés détiennent une autorité égale.",
|
||||
"disclaimer": {
|
||||
"label": "Important :",
|
||||
"text": "Il s'agit d'un déploiement fédéré à un stade précoce entre quatre locataires par le développeur du cadre de travail. Les mesures sont rapportées par l'utilisateur lui-même. Un audit indépendant et une validation plus large sont prévus mais n'ont pas encore été réalisés."
|
||||
"text": "Il s'agit d'une première phase de déploiement multi-locataires réalisée par le développeur du cadre. Les mesures sont rapportées par l'utilisateur lui-même. Un audit indépendant et une validation plus large sont prévus mais n'ont pas encore été réalisés."
|
||||
}
|
||||
},
|
||||
"overview": {
|
||||
"heading": "Vue d'ensemble",
|
||||
"platform_purpose": {
|
||||
"title": "Objectif de la plateforme",
|
||||
"description": "Village est une plateforme communautaire détenue par ses membres, offrant un stockage souverain des données, des fonctionnalités assistées par l'IA et une architecture respectueuse de la vie privée. Les communautés fonctionnent avec une propriété totale des données et une assistance IA protégée par la gouvernance."
|
||||
},
|
||||
"deployment_facts": {
|
||||
"title": "Faits de déploiement",
|
||||
"duration_label": "Durée :",
|
||||
"duration_value": "11+ mois en production",
|
||||
"tenant_label": "Modèle locataire :",
|
||||
"tenant_value": "Mono-locataire (multi-locataire prévu)",
|
||||
"features_label": "Fonctionnalités IA :",
|
||||
"features_value": "4 fonctionnalités gouvernées en production",
|
||||
"services_label": "Services/Réponse :",
|
||||
"services_value": "6 contrôles de gouvernance"
|
||||
"what_is": {
|
||||
"heading": "Qu'est-ce que le village ?",
|
||||
"description": "The Village est une plateforme appartenant aux membres pour les whānau, les marae, les clubs et les organisations communautaires. Chaque communauté obtient son propre locataire isolé avec un stockage de données souverain, des fonctionnalités assistées par l'IA et une confidentialité protégée par la gouvernance. La plateforme prend en charge le te reo Māori dans son ensemble.",
|
||||
"description_2": "Tout le traitement de l'IA s'effectue sur la propre infrastructure de la plateforme - un modèle Llama ajusté localement, sans que les données ne soient envoyées à des fournisseurs d'IA externes. Les communautés sont pleinement propriétaires de leurs données et peuvent retirer leur consentement à tout moment.",
|
||||
"facts_title": "Faits concernant le déploiement",
|
||||
"facts": {
|
||||
"duration_label": "Durée de l'enquête :",
|
||||
"duration_value": "18+ mois en production",
|
||||
"tenant_label": "Modèle de locataire :",
|
||||
"tenant_value": "Multi-locataires (plusieurs communautés)",
|
||||
"ai_label": "Modèle AI :",
|
||||
"ai_value": "Lama souverain (QLoRA affiné)",
|
||||
"features_label": "Caractéristiques de l'IA :",
|
||||
"features_value": "4 fonctions gouvernées en direct",
|
||||
"infra_label": "L'infrastructure :",
|
||||
"infra_value": "NZ + UE (pas de dépendance à l'égard des États-Unis)"
|
||||
}
|
||||
},
|
||||
"architecture_mapping": {
|
||||
"heading": "Cartographie de l'architecture",
|
||||
"description": "Chaque fonctionnalité IA de Village correspond à des services de gouvernance Tractatus spécifiques. Le tableau ci-dessous montre comment les six services se coordonnent pour chaque fonctionnalité.",
|
||||
"table": {
|
||||
"col_feature": "Fonctionnalité Village",
|
||||
"col_service": "Service principal",
|
||||
"col_function": "Fonction",
|
||||
"rows": [
|
||||
{
|
||||
"feature": "Réponses Home AI",
|
||||
"service": "BoundaryEnforcer",
|
||||
"function": "Bloque les jugements de valeur, s'en remet à l'humain"
|
||||
"sovereign": {
|
||||
"heading": "Architecture souveraine de l'IA",
|
||||
"intro": "Le village exploite son propre modèle linguistique - il ne s'agit pas d'un appel API à un hyperscaler américain, mais d'un modèle affiné localement où les données d'apprentissage, les poids du modèle et le pipeline d'inférence restent sous le contrôle de la communauté.",
|
||||
"model_title": "Modèle de langue locale",
|
||||
"model_desc": "Llama 3.1 8B et Llama 3.2 3B, ajustées avec QLoRA sur des données spécifiques à la communauté. Toutes les inférences sont exécutées sur l'infrastructure GPU de la plateforme.",
|
||||
"infra_title": "Infrastructure souveraine",
|
||||
"infra_desc": "Serveurs de production en Nouvelle-Zélande et dans l'UE. Aucune donnée ne transite par la juridiction américaine. Les données communautaires ne quittent jamais le déploiement auquel elles appartiennent.",
|
||||
"training_title": "Formation sous contrôle communautaire",
|
||||
"training_desc": "Mise au point de QLoRA sur des données spécifiques à un domaine avec suivi du consentement et de la provenance. Les communautés peuvent retirer des données d'entraînement et déclencher le recyclage du modèle.",
|
||||
"link_note": "Pour une présentation détaillée de l'architecture du modèle, de l'approche de la formation et de l'intégration de la gouvernance, voir <a href=\"/home-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">Home AI / SLL: Sovereign Locally-Trained Language Model</a>."
|
||||
},
|
||||
{
|
||||
"feature": "Traitement des requêtes utilisateur",
|
||||
"service": "CrossReferenceValidator",
|
||||
"function": "Empêche l'injection de prompts, valide l'intention"
|
||||
"polycentric": {
|
||||
"heading": "Gouvernance polycentrique",
|
||||
"intro": "La contribution distinctive du Village est son architecture de gouvernance. Plutôt qu'un opérateur unique prenant toutes les décisions, la plateforme met en œuvre une gouvernance polycentrique, c'est-à-dire plusieurs autorités égales qui partagent le contrôle structurel de l'utilisation de l'IA.",
|
||||
"coequal_title": "L'autorité conjointe",
|
||||
"coequal_desc": "Les communautés conservent une co-gouvernance architecturale - non seulement des droits de consultation, mais aussi une autorité structurelle sur la manière dont leurs données sont utilisées. Tiré des concepts te ao Māori de rangatiratanga (autodétermination) et kaitiakitanga (tutelle).",
|
||||
"nonparticipation_title": "Droit de non-participation",
|
||||
"nonparticipation_desc": "Les membres peuvent se retirer de toute fonction de l'IA sans perdre l'accès à la plateforme. La gouvernance de l'IA s'en remet au jugement humain sur les questions de valeurs et n'annule jamais les décisions de la communauté.",
|
||||
"taonga_title": "Conception centrée sur le Taonga",
|
||||
"taonga_desc": "Les trésors culturels (taonga) sont considérés comme des objets de première classe, avec un suivi de leur provenance, des droits de retrait et l'autorité de la communauté sur la manière dont ils apparaissent dans les contextes de l'IA.",
|
||||
"tenant_title": "Isolation des locataires",
|
||||
"tenant_desc": "Chaque communauté fonctionne dans un isolement complet des données. Il n'y a pas de partage de données entre locataires. Les décisions de gouvernance de chaque locataire ne s'appliquent qu'à l'intérieur de son propre périmètre.",
|
||||
"paper_note": "La base de la recherche est décrite dans le document <a href=\"/downloads/taonga-centred-steering-governance-polycentric-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">Taonga-Centred Steering Governance : Polycentric AI for Indigenous Data Sovereignty (IA polycentrique pour la souveraineté des données autochtones)</a>."
|
||||
},
|
||||
{
|
||||
"feature": "Gestion de session",
|
||||
"service": "ContextPressureMonitor",
|
||||
"function": "Suit la santé de la session, déclenche les transferts"
|
||||
"flow": {
|
||||
"heading": "Comment fonctionne la gouvernance dans la pratique",
|
||||
"intro": "Lorsqu'un membre utilise une fonction d'IA, la demande passe par six contrôles de gouvernance avant qu'une réponse ne soit générée. Chaque contrôle est indépendant et peut bloquer ou modifier la demande.",
|
||||
"steps": {
|
||||
"1": {
|
||||
"title": "Demande d'adhésion reçue",
|
||||
"desc": "Un membre demande de l'aide, sollicite l'OCR ou fait appel à l'aide à la rédaction d'un article."
|
||||
},
|
||||
{
|
||||
"feature": "Opérations multi-étapes",
|
||||
"service": "MetacognitiveVerifier",
|
||||
"function": "Détecte le dépassement de portée, nécessite une révision"
|
||||
"2": {
|
||||
"title": "Contrôle des limites des valeurs",
|
||||
"desc": "S'agit-il d'une question de valeurs qui nécessite un jugement humain ? Dans l'affirmative, l'IA s'abstient de répondre."
|
||||
},
|
||||
{
|
||||
"feature": "Flags de fonctionnalités",
|
||||
"service": "InstructionPersistenceClassifier",
|
||||
"function": "Classification de persistance pour les paramètres"
|
||||
"3": {
|
||||
"title": "Validation des intentions",
|
||||
"desc": "La demande est-elle en conflit avec les règles de gouvernance stockées ou tente-t-elle une injection rapide ? Références croisées avec des jeux d'instructions connus."
|
||||
},
|
||||
{
|
||||
"feature": "Gestion du consentement",
|
||||
"service": "PluralisticDeliberationOrchestrator",
|
||||
"function": "Décisions multi-parties prenantes"
|
||||
"4": {
|
||||
"title": "Contexte et santé de la session",
|
||||
"desc": "La session se situe-t-elle dans des limites acceptables ? Surveille la pression du contexte et déclenche un transfert gracieux si nécessaire."
|
||||
},
|
||||
"5": {
|
||||
"title": "Recherche et réponse avec filtrage des autorisations",
|
||||
"desc": "Le modèle du lama souverain génère une réponse en utilisant le contexte RAG filtré par les autorisations du membre. Tout le traitement reste sur l'infrastructure."
|
||||
},
|
||||
"6": {
|
||||
"title": "Vérification du champ d'application",
|
||||
"desc": "La réponse est-elle adaptée à ce qui a été demandé ? Détecte l'élargissement du champ d'application et bloque les réponses qui dépassent la demande initiale."
|
||||
},
|
||||
"7": {
|
||||
"title": "Livraison avec attribution",
|
||||
"desc": "La réponse est envoyée au membre avec l'indication de la source. Chaque étape est enregistrée à des fins d'audit."
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
"home_ai_flow": {
|
||||
"heading": "Le flux Home AI",
|
||||
"description": "Lorsqu'un utilisateur soumet une requête à Home AI, elle passe par six étapes de vérification avant qu'une réponse ne soit générée. Ce flux opère dans le chemin d'exécution critique.",
|
||||
"steps": [
|
||||
{
|
||||
"title": "Requête utilisateur reçue",
|
||||
"description": "L'utilisateur soumet une requête via le widget Help Chat ou l'aide à la rédaction"
|
||||
"features": {
|
||||
"heading": "Ce que la plateforme offre",
|
||||
"help": {
|
||||
"title": "Centre d'aide",
|
||||
"benefit": "Les membres posent des questions en langage naturel et obtiennent des réponses tirées du contenu de l'aide, des histoires et de la documentation, filtrées en fonction de leurs autorisations.",
|
||||
"governance": "Gouvernance : Le contrôle des limites des valeurs empêche l'IA de porter des jugements ; la validation de l'intention bloque les tentatives d'injection rapide."
|
||||
},
|
||||
{
|
||||
"title": "Vérification BoundaryEnforcer",
|
||||
"description": "S'agit-il d'une question de valeurs nécessitant un jugement humain ?"
|
||||
},
|
||||
{
|
||||
"title": "Vérification CrossReferenceValidator",
|
||||
"description": "Cela entre-t-il en conflit avec les instructions stockées ou tente-t-il une injection ?"
|
||||
},
|
||||
{
|
||||
"title": "Vérification ContextPressureMonitor",
|
||||
"description": "La santé de la session est-elle dans des limites acceptables ?"
|
||||
},
|
||||
{
|
||||
"title": "Traitement de la requête",
|
||||
"description": "Le système RAG récupère le contexte avec filtrage des permissions"
|
||||
},
|
||||
{
|
||||
"title": "Génération de la réponse",
|
||||
"description": "L'IA génère la réponse (Claude Haiku pour non-EN, Llama local pour EN)"
|
||||
},
|
||||
{
|
||||
"title": "Vérification MetacognitiveVerifier",
|
||||
"description": "La réponse est-elle appropriée à la portée de la requête ?"
|
||||
},
|
||||
{
|
||||
"title": "Livraison",
|
||||
"description": "Réponse envoyée à l'utilisateur avec attribution de source"
|
||||
}
|
||||
]
|
||||
},
|
||||
"governed_features": {
|
||||
"heading": "Fonctionnalités gouvernées en détail",
|
||||
"rag_help": {
|
||||
"title": "Centre d'aide basé sur RAG",
|
||||
"description": "Recherche vectorielle sur le contenu d'aide indexé, les histoires et la documentation. Les résultats sont filtrés par permissions utilisateur avant inclusion dans le contexte.",
|
||||
"services": "CrossReferenceValidator, BoundaryEnforcer"
|
||||
},
|
||||
"document_ocr": {
|
||||
"ocr": {
|
||||
"title": "OCR de documents",
|
||||
"description": "Extraction automatisée de texte à partir de documents téléchargés. Fonctionne avec contrôles de consentement explicites et journalisation d'audit.",
|
||||
"services": "PluralisticDeliberationOrchestrator (consentement)"
|
||||
"benefit": "Téléchargez un document et faites extraire le texte automatiquement. Utile pour numériser des lettres, des certificats et des documents historiques.",
|
||||
"governance": "Gouvernance : Le consentement explicite est requis avant le traitement. Toutes les opérations font l'objet d'un enregistrement d'audit avec une provenance complète."
|
||||
},
|
||||
"story_assistance": {
|
||||
"stories": {
|
||||
"title": "Aide à la rédaction",
|
||||
"description": "Suggestions de rédaction assistée par l'IA pour les histoires familiales. Les suggestions de contenu sont filtrées par BoundaryEnforcer pour éviter les recommandations inappropriées.",
|
||||
"services": "BoundaryEnforcer, MetacognitiveVerifier"
|
||||
"benefit": "Suggestions d'écriture assistée par l'IA pour les histoires de communautés et de familles. Aide à la structuration, aux messages-guides et à la révision en douceur.",
|
||||
"governance": "Gouvernance : La vérification des limites des valeurs permet d'éviter les suggestions de contenu inappropriées ; la vérification du champ d'application permet de s'assurer que l'IA reste dans les limites de ce qui a été demandé."
|
||||
},
|
||||
"ai_memory": {
|
||||
"title": "Transparence de la mémoire IA",
|
||||
"description": "Synthèse contrôlée par l'utilisateur avec tableau de bord d'audit complet. Les membres peuvent voir, modifier et supprimer ce que l'IA retient à leur sujet.",
|
||||
"services": "InstructionPersistenceClassifier, PluralisticDeliberationOrchestrator"
|
||||
},
|
||||
"services_label": "Services :"
|
||||
"memory": {
|
||||
"title": "Transparence de la mémoire de l'IA",
|
||||
"benefit": "Les membres peuvent voir, modifier et supprimer ce que l'IA \"retient\" d'eux. Un tableau de bord d'audit complet montre chaque interaction de l'IA.",
|
||||
"governance": "Gouvernance : Le consentement de plusieurs parties prenantes est requis. Les décisions relatives à la persistance sont classifiées et vérifiables. Les membres contrôlent leurs propres données."
|
||||
}
|
||||
},
|
||||
"limitations": {
|
||||
"heading": "Limites honnêtes",
|
||||
"intro": "Cette étude de cas documente les preuves préliminaires d'un déploiement fédéré à un stade précoce entre quatre locataires. Nous sommes transparents quant aux limites suivantes :",
|
||||
"heading": "Des limites honnêtes",
|
||||
"intro": "Cette étude de cas présente des preuves préliminaires d'un déploiement de production multi-locataire. Nous sommes transparents quant aux limites suivantes :",
|
||||
"items": [
|
||||
{
|
||||
"title": "Déploiement limité :",
|
||||
"description": "Village fonctionne avec quatre locataires fédérés au sein d'une même plateforme. La généralisation à d'autres contextes est inconnue."
|
||||
"title": "Petite échelle :",
|
||||
"description": "Le Village dessert actuellement un petit nombre de locataires communautaires. La généralisation à des déploiements plus importants ou à différents types de communautés n'est pas connue."
|
||||
},
|
||||
{
|
||||
"title": "Métriques auto-déclarées :",
|
||||
"title": "Mesures auto-rapportées :",
|
||||
"description": "Aucune vérification indépendante des données enregistrées n'a été effectuée."
|
||||
},
|
||||
{
|
||||
"title": "Chevauchement opérateur-développeur :",
|
||||
"description": "Le développeur du framework exploite également Village (conflit d'intérêts)."
|
||||
"description": "Le promoteur du cadre exploite également le village (conflit d'intérêts)."
|
||||
},
|
||||
{
|
||||
"title": "Tests adverses limités :",
|
||||
"description": "Aucune évaluation formelle red-team n'a été réalisée."
|
||||
"title": "Tests contradictoires limités :",
|
||||
"description": "Aucune évaluation formelle de l'équipe rouge n'a été réalisée."
|
||||
},
|
||||
{
|
||||
"title": "Invocation volontaire :",
|
||||
|
|
@ -159,29 +137,31 @@
|
|||
"demonstrates": {
|
||||
"heading": "Ce que cela démontre",
|
||||
"supports": {
|
||||
"title": "Les preuves soutiennent",
|
||||
"title": "Preuves à l'appui",
|
||||
"items": [
|
||||
"La gouvernance architecturale peut fonctionner en production sans surcharge prohibitive",
|
||||
"La coordination de six services est techniquement faisable",
|
||||
"Les violations de gouvernance sont détectables et auditables",
|
||||
"Le framework apprend des échecs (réponses aux incidents documentées)"
|
||||
"• Le déploiement de l'IA souveraine est techniquement possible pour les petites organisations communautaires",
|
||||
"• La gouvernance polycentrique peut fonctionner en production sans frais généraux prohibitifs",
|
||||
"• Il est possible d'isoler plusieurs locataires avec une gouvernance par communauté",
|
||||
"• Les violations de la gouvernance sont détectables et peuvent faire l'objet d'un audit",
|
||||
"• Le cadre tire les leçons des échecs (réponses documentées aux incidents)"
|
||||
]
|
||||
},
|
||||
"does_not_support": {
|
||||
"title": "Les preuves NE soutiennent PAS",
|
||||
"title": "Les preuves ne sont pas suffisantes",
|
||||
"items": [
|
||||
"L'efficacité du framework à grande échelle (milliers d'utilisateurs simultanés)",
|
||||
"La généralisabilité à différents systèmes d'IA",
|
||||
"La résistance aux attaques adverses sophistiquées",
|
||||
"La suffisance réglementaire (conformité EU AI Act non testée)"
|
||||
"• Efficacité du cadre à grande échelle (milliers d'utilisateurs simultanés)",
|
||||
"• Généralisabilité à différents systèmes d'intelligence artificielle ou architectures de modèles",
|
||||
"• Résistance aux attaques adverses sophistiquées",
|
||||
"• Suffisance de la réglementation (la conformité à la loi européenne sur l'IA n'a pas été testée)"
|
||||
]
|
||||
}
|
||||
},
|
||||
"cta": {
|
||||
"heading": "Explorer plus loin",
|
||||
"description": "Découvrez la plateforme Village en action ou approfondissez l'architecture technique.",
|
||||
"visit_village": "Visiter Village →",
|
||||
"system_architecture": "Architecture système →",
|
||||
"research_details": "Détails de recherche →"
|
||||
"heading": "En savoir plus",
|
||||
"description": "Plongez dans l'architecture technique, lisez les études ou voyez la plateforme Village en action.",
|
||||
"visit_village": "Visiter le village →",
|
||||
"home_ai": "Modèle de langue souveraine →",
|
||||
"research_paper": "Document de recherche →",
|
||||
"research_details": "Détails de la recherche →"
|
||||
}
|
||||
}
|
||||
|
|
@ -2,149 +2,127 @@
|
|||
"breadcrumb": "Tātaritanga Take o te Kāinga",
|
||||
"hero": {
|
||||
"badge": "Tātaritanga Take",
|
||||
"title": "Tūāpapa o te Kāinga: Tractatus kei te whakaputanga",
|
||||
"subtitle": "Me pēhea e whakamahi ai a Home AI i ngā ratonga whakahaere e ono katoa o Tractatus ki ia whakawhitinga a te kaiwhakamahi i runga i tētahi tūāpapa hapori ora.",
|
||||
"title": "Te Kāinga: AI Hapori Motuhake kei te Whakaputanga",
|
||||
"subtitle": "He tūāpapa hapori maha-teneti e whakahaere ana i tōna ake tauira reo i runga i tōna ake hanganga, e whakahaerehia ana e tētahi hanganga pokapū maha, ā, kei ngā hapori te mana ōrite.",
|
||||
"disclaimer": {
|
||||
"label": "Hira:",
|
||||
"text": "E whakaatu ana tēnei tuhinga i tētahi whakaurunga honohono ā-tīneti e whā i te wāhanga tuatahi, nā te kaiwhakawhanake anga. Ko ngā ine i pūrongo ā-ringa. Kei te whakamaherehia he arotake motuhake me te whakamana whānui, engari kāore anō kia whakahaerehia."
|
||||
"text": "E tuhi ana tēnei i tētahi whakaurunga maha-kaiwhakamahi i te wā tuatahi e whakahaerehia ana e te kaiwhakawhanake anga. Ko ngā ine e pūrongo ana te kaiwhakawhanake anga. Kua whakamaheretia he arotake motuhake me te whakamana whānui, engari kāore anō kia whakahaerehia."
|
||||
}
|
||||
},
|
||||
"overview": {
|
||||
"heading": "Whakarāpopototanga",
|
||||
"platform_purpose": {
|
||||
"title": "Te take o te tūāpapa",
|
||||
"description": "Ko The Village he tūāpapa hapori nō ngā mema, e whakarato ana i te rokiroki raraunga rangatira, ngā āhuatanga āwhina atamai (AI), me te hanganga tūmataitinga i te hoahoa. Ka whakahaere ngā hapori me te rangatiratanga katoa o ngā raraunga me te āwhina atamai (AI) e tiakina ana e te mana whakahaere."
|
||||
},
|
||||
"deployment_facts": {
|
||||
"title": "Ngā Mōhiohio Whakawhiwhinga",
|
||||
"what_is": {
|
||||
"heading": "He aha te kāinga",
|
||||
"description": "Ko The Village he papanga rangahaua e ngā mema mō ngā whānau, ngā marae, ngā karapu, me ngā rōpū hapori. Ka whiwhi ia hapori i tōna ake wāhi noho motuhake me te rokiroki raraunga rangatira, ngā āhuatanga āwhina AI, me te tūmataitinga e tiakina ana e te mana whakahaere. Ka tautoko te papanga i te reo Māori puta noa.",
|
||||
"description_2": "Ka whakahaerehia ngā tukanga AI katoa i runga i te hanganga ake o te papa — he tauira Llama kua whakatikatikahia ā-rohe, kāore he raraunga e tukuna ki ngā kaiwhakarato AI o waho. Ka whakahaere ngā hapori me te rangatiratanga katoa o ā rātou raraunga, ā, ka taea e rātou te tango i tō rātou whakaaetanga i ngā wā katoa.",
|
||||
"facts_title": "Ngā Mōhiohio Whakawhiwhinga",
|
||||
"facts": {
|
||||
"duration_label": "Te roa:",
|
||||
"duration_value": "11 marama neke atu i te wāhanga whakaputa",
|
||||
"duration_value": "18 marama neke atu i te wāhanga whakaputa",
|
||||
"tenant_label": "Mōhiohio mō te Kaihiri:",
|
||||
"tenant_value": "Kai-rēti kotahi (kei te whakamahere kia maha ngā kai-rēti)",
|
||||
"tenant_value": "Maha-teneti (hāpori maha)",
|
||||
"ai_label": "Mōdēra AI:",
|
||||
"ai_value": "Sovereign Llama (i whakangāwarihia mō QLoRA)",
|
||||
"features_label": "Ngā āhuatanga AI:",
|
||||
"features_value": "4 āhuatanga whakahaere ora",
|
||||
"services_label": "Ngā Ratonga/Whakautu:",
|
||||
"services_value": "6 tirohanga whakahaere"
|
||||
"infra_label": "Tūāpapa:",
|
||||
"infra_value": "Aotearoa + Uniana o Europi (kāore he whakawhirinaki ki te US)"
|
||||
}
|
||||
},
|
||||
"architecture_mapping": {
|
||||
"heading": "Mapi hanganga",
|
||||
"description": "Ia āhuatanga o Village AI e hono ana ki ngā ratonga whakahaere motuhake o Tractatus. E whakaatu ana te ripanga i raro nei me pēhea ngā ratonga e ono e mahi ngātahi ana mō ia āhuatanga.",
|
||||
"table": {
|
||||
"col_feature": "Āhuatanga o te Kāinga",
|
||||
"col_service": "Ratonga Matua",
|
||||
"col_function": "Tānga",
|
||||
"rows": [
|
||||
{
|
||||
"feature": "Ngā Whakautu AI o te Kāinga",
|
||||
"service": "Kaitiaki Rohe",
|
||||
"function": "Ka aukati i ngā uara me ngā whakataunga, ka whakawhirinaki ki te tangata"
|
||||
"sovereign": {
|
||||
"heading": "Hoahoanga AI Rangatira",
|
||||
"intro": "Ka whakahaere a The Village i tōna ake tauira reo — ehara i te karanga API ki tētahi kaiwhakarato hyperscale o Amerika, engari he tauira kua whakangāwarihia ā-rohe, ā, kei raro tonu i te mana whakahaere a te hapori ngā raraunga whakangungu, ngā taumaha tauira, me te ara tukatuka whakatau.",
|
||||
"model_title": "Mōdela Reo ā-Rohe",
|
||||
"model_desc": "Llama 3.1 8B me Llama 3.2 3B, kua whakangāwarihia mā QLoRA ki ngā raraunga motuhake o te hapori. Ka whakahaerehia ngā aromatawai katoa i runga i te hanganga GPU motuhake o te tūāpapa.",
|
||||
"infra_title": "Tūāpapa Rangatira",
|
||||
"infra_desc": "Ngā tūmau whakaputa kei Aotearoa me te Uniana o Europi. Kāore ngā raraunga e whakawhiti i raro i te mana whakahaere o Amerika. Kāore ngā raraunga hapori e puta atu i te whakatakotoranga e hāngai ana ki a rātou.",
|
||||
"training_title": "Whakangungu e whakahaerehia ana e te hapori",
|
||||
"training_desc": "Te whakangāwari QLoRA ki ngā raraunga motuhake o te rohe, me te whai i ngā whakaaetanga me te takenga mai. Ka taea e ngā hapori te tango i ngā raraunga whakangungu, ā, ka whakaoho i te whakahou anō i te tauira.",
|
||||
"link_note": "Mō tētahi whakamārama taipitopito mō te hanganga tauira, te huarahi whakangungu, me te whakaurunga whakahaere, tirohia <a href=\"/home-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">Home AI / SLL: Sovereign Locally-Trained Language Model</a>."
|
||||
},
|
||||
{
|
||||
"feature": "Te tukatuka pātai kaiwhakamahi",
|
||||
"service": "Kaitātari Whakataurite",
|
||||
"function": "Ka aukati i te werohanga ohorere, ka whakamana i te aronga"
|
||||
"polycentric": {
|
||||
"heading": "Kaitiakitanga ā-pokapū maha",
|
||||
"intro": "Ko te āhuatanga motuhake o te Hapori ko tōna hanganga whakahaere. Ehara i te mea kotahi te kaiwhakahaere e whakatau ana i ngā whakataunga katoa; engari, mā te tūāpapa e whakatinana te whakahaere maha-pū – arā, he maha ngā mana ōrite e tiritiri ana i te mana hanganga mō te whakamahinga o te AI.",
|
||||
"coequal_title": "Tūmatua ōrite",
|
||||
"coequal_desc": "Ka pupuri ngā hapori i te rangatōpū whakahaere hoahoanga — ehara i te mea he mana uiui anake, engari he mana hanganga hoki ki te whakahaere me pēhea te whakamahinga o ā rātou raraunga. Nā ngā ariā o te ao Māori o rangatiratanga (te mana motuhake) me kaitiakitanga (te tiakitanga) i ahu mai.",
|
||||
"nonparticipation_title": "Te Tika ki te Kore Whai Wāhanga",
|
||||
"nonparticipation_desc": "Ka taea e ngā mema te whakakore i tētahi āhuatanga AI ahakoa he aha, me te kore e ngaro tō rātou uru ki te papanga. Ka whakawhirinaki te whakahaere AI ki ngā whakataunga a te tangata mō ngā pātai uara, ā, kāore rawa e whakakore i ngā whakataunga a te hapori.",
|
||||
"taonga_title": "Hoahoa e aro ana ki ngā taonga",
|
||||
"taonga_desc": "Ka whakahaerehia ngā taonga ahurea (taonga) hei taonga tuatahi, me te whai i te takenga mai, me ngā mana tango, me te mana ā-hapori ki te whakatau me pēhea rātou e puta ai i ngā horopaki AI.",
|
||||
"tenant_title": "Wehewehenga ā-Rohe o te Kiritaki",
|
||||
"tenant_desc": "Ka whakahaere ia hapori i roto i te wehewehe raraunga katoa. Kāore he tiritiri raraunga i waenga i ngā kaiwhakamahi. Ko ngā whakataunga whakahaere a ia kaiwhakamahi e pā ana anake ki tōna ake rohe.",
|
||||
"paper_note": "Ka whakamārama te tūāpapa rangahau i roto i <a href=\"/downloads/taonga-centred-steering-governance-polycentric-ai.html\" class=\"text-blue-600 hover:text-blue-700 font-medium underline\">te Whakahaere Arataki e Arotahi ana ki ngā Taonga: AI maha-pokapū mō te rangatiratanga raraunga taketake</a>."
|
||||
},
|
||||
{
|
||||
"feature": "Whakahaere Wāhanga",
|
||||
"service": "Te Kaiwhakahaere Pēhanga Horopaki",
|
||||
"function": "E aroturuki ana i te hauora o te wāhanga, ā, ka whakaoho i ngā whakawhitinga mahi."
|
||||
"flow": {
|
||||
"heading": "Te āhua o te whakahaere i roto i te mahi",
|
||||
"intro": "Ka whakamahi tētahi mema i tētahi āhuatanga AI, ka puta te tono i ngā tirohanga whakahaere e ono i mua i te whakaputa i tētahi whakautu. He motuhake ia tirohanga, ā, ka taea e ia te aukati, te whakarerekē rānei i te tono.",
|
||||
"steps": {
|
||||
"1": {
|
||||
"title": "Kua whiwhi i te tono a te mema",
|
||||
"desc": "Ka tono tētahi mema mō te āwhina, ka tono OCR, ka whakamahi rānei i te āwhina kōrero."
|
||||
},
|
||||
{
|
||||
"feature": "Ngā Mahi-ā-Hipanga",
|
||||
"service": "Kaitātari Whakaaroaro",
|
||||
"function": "Ka kitea te pikinga o te whānuitanga o te kaupapa, me arotake"
|
||||
"2": {
|
||||
"title": "Tātari rohe uara",
|
||||
"desc": "He pātai uara tēnei e hiahiatia ana te whakataunga a te tangata? Mēnā āe, ka whakaroa te AI, kāore e whakautu."
|
||||
},
|
||||
{
|
||||
"feature": "Ngā haki āhuatanga",
|
||||
"service": "Akoranga Manawaroa Kaiwhakarōpū",
|
||||
"function": "Te whakarōpūtanga mau tonu mō ngā tautuhinga"
|
||||
"3": {
|
||||
"title": "Whakamana i te aronga",
|
||||
"desc": "Kei te taupatupatu te tono ki ngā ture whakahaere kua rokiroki, ki te ngana rānei ki te werohia o ngā whakahau ohorere? Ka whakataurite ki ngā huinga whakahau kua mōhiotia."
|
||||
},
|
||||
{
|
||||
"feature": "Te Whakahaere Whakaaetanga",
|
||||
"service": "Kaiwhakahaere Whiriwhiringa Maha",
|
||||
"function": "Ngā whakataunga ā ngā kaiwhai pānga maha"
|
||||
"4": {
|
||||
"title": "Te horopaki me te hauora o te wāhanga",
|
||||
"desc": "Kei roto i ngā pae e whakaaetia ana te wāhanga? Ka aroturuki i te pēhanga horopaki, ā, ka whakaoho i te tuku whakawhiti ātaahua ina hiahiatia."
|
||||
},
|
||||
"5": {
|
||||
"title": "Tiki me te whakautu kua tātaritia e ngā whakaaetanga",
|
||||
"desc": "Ka whakaputa urupare te tauira Llama rangatira mā te whakamahi i te horopaki RAG kua tātarihia e ngā whakaaetanga a te mema. Ka noho katoa ngā tukanga ki runga i te hanganga."
|
||||
},
|
||||
"6": {
|
||||
"title": "Whakamana i te whānuitanga",
|
||||
"desc": "He tika te whakautu ki tā i pātaihia? Ka kitea te whakawhānuitanga o te kaupapa, ā, ka aukati i ngā whakautu e neke atu ana i te tono taketake."
|
||||
},
|
||||
"7": {
|
||||
"title": "Tuku me te tohu kaituhi",
|
||||
"desc": "Ka tukuna te whakautu ki te mema me te tohu i te puna. Ka tuhia ia hikoinga mō te arotake."
|
||||
}
|
||||
]
|
||||
}
|
||||
},
|
||||
"home_ai_flow": {
|
||||
"heading": "Te rerenga o te AI kāinga",
|
||||
"description": "Ka tukuna e te kaiwhakamahi he pātai ki Home AI, ka haere mā ngā taumata whakamana e ono i mua i te whakaputa i te whakautu. Ka mahi tēnei rerenga i roto i te ara whakahaere matua.",
|
||||
"steps": [
|
||||
{
|
||||
"title": "Kua whiwhi i te pātai a te kaiwhakamahi",
|
||||
"description": "Ka tuku pātai te kaiwhakamahi mā te taputapu Kōrerorero Āwhina, mā te āwhina kōrero rānei."
|
||||
"features": {
|
||||
"heading": "Ngā mea ka tukuna e te tūāpapa",
|
||||
"help": {
|
||||
"title": "Pūnaha Awhina",
|
||||
"benefit": "Ka pātai ngā mema i roto i te reo ā-taiao, ā, ka whiwhi whakautu i tangohia mai i ngā ihirangi āwhina, i ngā kōrero, me ngā tuhinga — kua tātarihia e ā rātou mana whakaahei.",
|
||||
"governance": "Whakahaere: Ka ārai te tirotiro rohe o ngā uara i te AI kia kaua e whakatau; ka aukati te whakamana whāinga i ngā whakamātau whakauru tono."
|
||||
},
|
||||
{
|
||||
"title": "Tātari Kaiwhakahaere Rohe",
|
||||
"description": "He pātai mō ngā uara tēnei e hiahiatia ana te whakataunga a te tangata?"
|
||||
},
|
||||
{
|
||||
"title": "Tātari Whakau Whakawhitiwhiti",
|
||||
"description": "Kei te taupatupatu tēnei ki ngā tohutohu kua rokiroki, ki te ngana rānei ki te werohanga?"
|
||||
},
|
||||
{
|
||||
"title": "Tātari i te Pēhanga Horopaki",
|
||||
"description": "Kei roto i ngā pae e whakaaetia ana te hauora o te wāhanga?"
|
||||
},
|
||||
{
|
||||
"title": "Te tukatuka pātai",
|
||||
"description": "Ka tiki e te pūnaha RAG te horopaki mā te tātari whakaaetanga"
|
||||
},
|
||||
{
|
||||
"title": "Te Whakaputa Whakautu",
|
||||
"description": "Ka whakaputa a AI i te whakautu (Claude Haiku mō ngā reo kāore i te Ingarihi, Llama ā-rohe mō te reo Ingarihi)"
|
||||
},
|
||||
{
|
||||
"title": "Kaitātari Whakaaroaro",
|
||||
"description": "He tika te whakautu ki te whānuitanga o te pātai?"
|
||||
},
|
||||
{
|
||||
"title": "Tuku",
|
||||
"description": "Ka tukuna te whakautu ki te kaiwhakamahi me te tohu i te puna."
|
||||
}
|
||||
]
|
||||
},
|
||||
"governed_features": {
|
||||
"heading": "Ngā āhuatanga whakahaere i te taipitopito",
|
||||
"rag_help": {
|
||||
"title": "Pūnaha Awhina RAG",
|
||||
"description": "Rapu ā-vector i ngā ihirangi āwhina kua tāruatia, ngā kōrero, me ngā tuhinga. Ka tātarihia ngā hua e ngā mana whakamahi a te kaiwhakamahi i mua i te whakauru ki te horopaki.",
|
||||
"services": "Kaitātari Whakawhāititanga, Kaitiaki Rohe"
|
||||
},
|
||||
"document_ocr": {
|
||||
"ocr": {
|
||||
"title": "Tuhipuka OCR",
|
||||
"description": "Te tangohanga aunoa o ngā kuputuhi mai i ngā tuhinga kua tukuna. Ka whakahaerehia i raro i ngā mana whakaaetanga mārama me te rēhitatanga arotake.",
|
||||
"services": "Kaiwhakahaere Whiriwhiringa Maha (whakaaetanga)"
|
||||
"benefit": "Tukuna he tuhinga, ka tangohia aunoa te kuputuhi. He whaihua mō te huri i ngā reta, ngā tiwhikete, me ngā rekoata hītori ki te āhua matihiko.",
|
||||
"governance": "Whakahaere: Me whakaaetanga mārama i mua i te tukatuka. Ka tuhia ki ngā rangitaki arotake ngā mahi katoa me te takenga mai katoa."
|
||||
},
|
||||
"story_assistance": {
|
||||
"stories": {
|
||||
"title": "Āwhina kōrero",
|
||||
"description": "Ngā tūtohunga tuhi āwhina AI mō ngā kōrero whānau. Ngā tūtohunga ihirangi kua tātarihia mā BoundaryEnforcer hei aukati i ngā tūtohunga kāore e tika ana.",
|
||||
"services": "Kaitiaki Rohe, Kaitātari Mātauranga Hinengaro"
|
||||
"benefit": "Ngā tūtohutanga tuhi āwhina AI mō ngā kōrero hapori me ngā hītori whānau. Ka āwhina ki te hanganga, ki ngā whakahau, me te whakatika māmā.",
|
||||
"governance": "Whakahaere: Ka ārai te tirotiro rohe o ngā uara i ngā tūtohutanga ihirangi hē; mā te whakamana i te whānuitanga ka noho te AI ki roto i ngā mea i tonoa."
|
||||
},
|
||||
"ai_memory": {
|
||||
"memory": {
|
||||
"title": "Mārama o te Mahara AI",
|
||||
"description": "Whakarapopototanga e whakahaerehia ana e te kaiwhakamahi, me te papamātaki arotake katoa. Ka taea e ngā mema te tiro, te whakatika, me te muku i ngā mea e 'maumahara ana' te AI mō rātou.",
|
||||
"services": "Kaiwhakarato Whakahau Manawanui, Kaiwhakahaere Whiriwhiringa Maha"
|
||||
},
|
||||
"services_label": "Ngā ratonga:"
|
||||
"benefit": "Ka taea e ngā mema te kite, te whakatika, me te muku i ngā mea e 'maumahara ana' te AI mō rātou. Ka whakaatu te papamātātai arotake katoa i ia whakawhitinga katoa ki te AI.",
|
||||
"governance": "Whakahaere: Me whakaae ngā kaiwhai pānga maha. Ka whakarōpūhia ngā whakataunga mō te manawanui, ā, ka taea te arotake. Ka whakahaere ngā mema i ā rātou ake raraunga."
|
||||
}
|
||||
},
|
||||
"limitations": {
|
||||
"heading": "Ngā here pono",
|
||||
"intro": "E tuhi ana tēnei rangahau take i ngā taunakitanga tuatahi mai i tētahi whakaurunga honohono ā-tīmatanga puta noa i ngā kaipā e whā. E mārama ana mātou ki ngā here e whai ake nei:",
|
||||
"intro": "E tuhi ana tēnei rangahau take i ngā taunakitanga tuatahi mai i tētahi whakaurunga maha-teneti i te taumata whakaputa. E mārama ana mātou ki ngā here e whai ake nei:",
|
||||
"items": [
|
||||
{
|
||||
"title": "Whakaurunga Herenga:",
|
||||
"description": "Kei te whakahaere a Village i ngā kaipānga whakawhanaunga e whā i roto i tētahi tūāpapa kotahi. Kāore i te mōhiotia te whānuitanga ki ētahi atu horopaki."
|
||||
"title": "Āhua iti:",
|
||||
"description": "Kei te whakarato ināianei te Village i ngā ratonga ki tētahi tokoiti o ngā kaipāmu hapori. Kāore i te mōhiotia mēnā ka taea te whakamahia whānuitia ki ngā whakaurunga nui ake, ki ngā momo hapori rerekē rānei."
|
||||
},
|
||||
{
|
||||
"title": "Ngā Tohu i Rīpoatatia e Te Tangata Anō:",
|
||||
"description": "Kāore anō kia whakahaerehia he whakamana motuhake mō ngā raraunga kua tuhia."
|
||||
"description": "Kāore anō kia whakahaeretia he whakamana motuhake mō ngā raraunga kua tuhia."
|
||||
},
|
||||
{
|
||||
"title": "Te whakawhānuitanga o ngā kaiwhakahaere me ngā kaiwhakawhanake:",
|
||||
"description": "Ka whakawhanake hoki te kaiwhakawhanake anga i te Village (tāraru painga)."
|
||||
"description": "Ka whakahaere hoki e te kaiwhakawhanake anga te Hāpori (taupatupatu painga)."
|
||||
},
|
||||
{
|
||||
"title": "Whakamātautau Whakataetae Herenga:",
|
||||
|
|
@ -152,7 +130,7 @@
|
|||
},
|
||||
{
|
||||
"title": "Te karanga ā-tūao:",
|
||||
"description": "I te ariā, ka taea e te AI te karo i te whakahaere mēnā kāore i whirihorahia kia whakamahia."
|
||||
"description": "I te ariā, ka taea e te AI te karo i te whakahaere mēnā kāore i whakaritea kia whakamahia."
|
||||
}
|
||||
]
|
||||
},
|
||||
|
|
@ -161,27 +139,29 @@
|
|||
"supports": {
|
||||
"title": "E tautoko ana ngā taunakitanga",
|
||||
"items": [
|
||||
"Ka taea e te whakahaere hanganga te whakahaere i te taiao whakaputa me te kore he utu taumaha.",
|
||||
"He mea hangarau e taea ana te whakakotahitanga o ngā ratonga e ono.",
|
||||
"Ka taea te kitea, ka taea hoki te arotake i ngā hē whakahaere",
|
||||
"Ka ako te anga i ngā hapa (ngā urupare ki ngā aitua kua tuhia)"
|
||||
"• He mea ā-hangarau e taea ana te whakaurunga o Sovereign AI mō ngā whakahaere hapori iti",
|
||||
"• Ka taea e te whakahaere ā-pokapū maha te whakahaere i roto i te whakaputanga me te kore he utu taumata teitei",
|
||||
"• Ka taea te wehewehe maha-teneti me te whakahaere ā-hapori",
|
||||
"• Ka taea te kitea, ka taea hoki te arotake i ngā hē whakahaere",
|
||||
"• Ka ako te anga i ngā hapa (ngā urupare aitua kua tuhia)"
|
||||
]
|
||||
},
|
||||
"does_not_support": {
|
||||
"title": "Ehara i te mea e tautoko ana ngā taunakitanga",
|
||||
"items": [
|
||||
"Te whaihua o te anga i te taumata nui (ngā mano kaiwhakamahi e mahi ana i te wā kotahi)",
|
||||
"Te āhei ki te whakamahia whānuitia puta noa i ngā pūnaha AI rerekē",
|
||||
"Te ātete ki ngā whakaeke whakataetae matatau",
|
||||
"Te whai rawa o ngā whakaritenga (kāore anō kia whakamātauria te ū ki te Ture AI o te EU)"
|
||||
"• Te whai hua o te anga i te taumata whānui (ngā mano kaiwhakamahi e mahi ana i te wā kotahi)",
|
||||
"• Te whānuitanga ki ngā pūnaha AI rerekē, ki ngā hanganga tauira rānei",
|
||||
"• Ārai ki ngā whakaeke whakataetae matatau",
|
||||
"• Te whai kiko o ngā whakaritenga (kāore anō kia whakamātauria te ū ki te Ture AI o te EU)"
|
||||
]
|
||||
}
|
||||
},
|
||||
"cta": {
|
||||
"heading": "Tūhura anō",
|
||||
"description": "Tirohia te papaanga Village e mahi ana, rānei toro hohonu atu ki te hanganga hangarau.",
|
||||
"description": "Ruku hohonu ki te hanganga hangarau, pānuitia ngā rangahau, kia kite rānei i te papaanga Village e mahi ana.",
|
||||
"visit_village": "Haere ki te kāinga →",
|
||||
"system_architecture": "Hoahoa Pūnaha →",
|
||||
"home_ai": "Mōdela Reo Rangatira →",
|
||||
"research_paper": "Pepa Rangahau →",
|
||||
"research_details": "Ngā taipitopito rangahau →"
|
||||
}
|
||||
}
|
||||
|
|
@ -6,14 +6,14 @@
|
|||
<meta http-equiv="Cache-Control" content="no-cache, no-store, must-revalidate">
|
||||
<meta http-equiv="Pragma" content="no-cache">
|
||||
<meta http-equiv="Expires" content="0">
|
||||
<title>Village Case Study | Tractatus AI Safety Framework</title>
|
||||
<meta name="description" content="Technical case study of Tractatus governance in the Village platform. How Home AI applies six governance services to every user interaction.">
|
||||
<title>The Village: Sovereign Community AI in Production | Tractatus</title>
|
||||
<meta name="description" content="How the Village platform delivers governed AI to whānau, marae, and community organisations — sovereign infrastructure, polycentric governance, and a locally-trained language model.">
|
||||
|
||||
<!-- Open Graph / Facebook -->
|
||||
<meta property="og:type" content="website">
|
||||
<meta property="og:url" content="https://agenticgovernance.digital/village-case-study.html">
|
||||
<meta property="og:title" content="Village Case Study - Tractatus in Production">
|
||||
<meta property="og:description" content="Technical case study of Tractatus governance in the Village platform. How Home AI applies six governance services to every user interaction.">
|
||||
<meta property="og:title" content="The Village: Sovereign Community AI in Production">
|
||||
<meta property="og:description" content="How the Village platform delivers governed AI to whānau, marae, and community organisations — sovereign infrastructure, polycentric governance, and a locally-trained language model.">
|
||||
<meta property="og:image" content="https://agenticgovernance.digital/images/social-preview.png">
|
||||
<meta property="og:image:width" content="1200">
|
||||
<meta property="og:image:height" content="630">
|
||||
|
|
@ -22,8 +22,8 @@
|
|||
<!-- Twitter -->
|
||||
<meta name="twitter:card" content="summary_large_image">
|
||||
<meta name="twitter:url" content="https://agenticgovernance.digital/village-case-study.html">
|
||||
<meta name="twitter:title" content="Village Case Study - Tractatus in Production">
|
||||
<meta name="twitter:description" content="Technical case study of how Home AI applies six governance services to every user interaction.">
|
||||
<meta name="twitter:title" content="The Village: Sovereign Community AI in Production">
|
||||
<meta name="twitter:description" content="Sovereign community AI with polycentric governance, locally-trained language model, and no US hyperscaler dependency.">
|
||||
<meta name="twitter:image" content="https://agenticgovernance.digital/images/social-preview.png">
|
||||
|
||||
<link rel="icon" type="image/svg+xml" href="/favicon-new.svg">
|
||||
|
|
@ -76,17 +76,16 @@
|
|||
CASE STUDY
|
||||
</div>
|
||||
<h1 class="text-4xl font-bold mb-6 text-black" data-i18n="hero.title">
|
||||
Village Platform: Tractatus in Production
|
||||
The Village: Sovereign Community AI in Production
|
||||
</h1>
|
||||
<p class="text-xl mb-8 max-w-3xl mx-auto text-black" data-i18n="hero.subtitle">
|
||||
How Home AI applies all six Tractatus governance services to every user interaction in a live community platform.
|
||||
A multi-tenant community platform running its own language model on its own infrastructure, governed by a polycentric architecture where communities hold co-equal authority.
|
||||
</p>
|
||||
|
||||
<!-- Caveat Box - HIGH VISIBILITY DISCLAIMER -->
|
||||
<!-- Caveat Box -->
|
||||
<div class="bg-amber-100 disclaimer-border border-amber-500 rounded-lg p-6 max-w-3xl mx-auto">
|
||||
<p class="text-amber-900 text-lg">
|
||||
<strong class="text-amber-800">⚠️ <span data-i18n="hero.disclaimer.label">Important:</span></strong> <span data-i18n="hero.disclaimer.text">This documents an early-stage federated deployment across four tenants by the framework developer.
|
||||
Metrics are self-reported. Independent audit and broader validation are planned but not yet conducted.</span>
|
||||
<strong class="text-amber-800">⚠️ <span data-i18n="hero.disclaimer.label">Important:</span></strong> <span data-i18n="hero.disclaimer.text">This documents an early-stage multi-tenant deployment by the framework developer. Metrics are self-reported. Independent audit and broader validation are planned but not yet conducted.</span>
|
||||
</p>
|
||||
</div>
|
||||
</div>
|
||||
|
|
@ -97,38 +96,42 @@
|
|||
<!-- Main Content -->
|
||||
<main id="main-content" role="main" class="max-w-5xl mx-auto px-4 sm:px-6 lg:px-8 py-12">
|
||||
|
||||
<!-- Overview Section -->
|
||||
<!-- What Is the Village -->
|
||||
<section class="mb-16">
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="overview.heading">Overview</h2>
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="what_is.heading">What Is the Village</h2>
|
||||
|
||||
<div class="bg-white rounded-xl shadow-md p-8 mb-8">
|
||||
<div class="grid grid-cols-1 md:grid-cols-2 gap-8">
|
||||
<div>
|
||||
<h3 class="text-lg font-semibold text-gray-900 mb-4" data-i18n="overview.platform_purpose.title">Platform Purpose</h3>
|
||||
<p class="text-gray-700 leading-relaxed" data-i18n="overview.platform_purpose.description">
|
||||
The Village is a member-owned community platform providing sovereign data storage,
|
||||
AI-assisted features, and privacy-by-design architecture. Communities operate
|
||||
with full data ownership and governance-protected AI assistance.
|
||||
<p class="text-gray-700 leading-relaxed mb-4" data-i18n="what_is.description">
|
||||
The Village is a member-owned platform for whānau, marae, clubs, and community organisations. Each community gets its own isolated tenant with sovereign data storage, AI-assisted features, and governance-protected privacy. The platform supports te reo Māori throughout.
|
||||
</p>
|
||||
<p class="text-gray-700 leading-relaxed" data-i18n="what_is.description_2">
|
||||
All AI processing runs on the platform's own infrastructure — a locally fine-tuned Llama model with no data sent to external AI providers. Communities operate with full data ownership and can withdraw consent at any time.
|
||||
</p>
|
||||
</div>
|
||||
<div>
|
||||
<h3 class="text-lg font-semibold text-gray-900 mb-4" data-i18n="overview.deployment_facts.title">Deployment Facts</h3>
|
||||
<h3 class="text-lg font-semibold text-gray-900 mb-4" data-i18n="what_is.facts_title">Deployment Facts</h3>
|
||||
<ul class="space-y-2 text-gray-700">
|
||||
<li class="flex items-center">
|
||||
<span class="w-32 font-medium" data-i18n="overview.deployment_facts.duration_label">Duration:</span>
|
||||
<span data-i18n="overview.deployment_facts.duration_value">11+ months in production</span>
|
||||
<span class="w-36 font-medium" data-i18n="what_is.facts.duration_label">Duration:</span>
|
||||
<span data-i18n="what_is.facts.duration_value">18+ months in production</span>
|
||||
</li>
|
||||
<li class="flex items-center">
|
||||
<span class="w-32 font-medium" data-i18n="overview.deployment_facts.tenant_label">Tenant Model:</span>
|
||||
<span data-i18n="overview.deployment_facts.tenant_value">Single-tenant (multi-tenant planned)</span>
|
||||
<span class="w-36 font-medium" data-i18n="what_is.facts.tenant_label">Tenant Model:</span>
|
||||
<span data-i18n="what_is.facts.tenant_value">Multi-tenant (multiple communities)</span>
|
||||
</li>
|
||||
<li class="flex items-center">
|
||||
<span class="w-32 font-medium" data-i18n="overview.deployment_facts.features_label">AI Features:</span>
|
||||
<span data-i18n="overview.deployment_facts.features_value">4 governed features live</span>
|
||||
<span class="w-36 font-medium" data-i18n="what_is.facts.ai_label">AI Model:</span>
|
||||
<span data-i18n="what_is.facts.ai_value">Sovereign Llama (QLoRA fine-tuned)</span>
|
||||
</li>
|
||||
<li class="flex items-center">
|
||||
<span class="w-32 font-medium" data-i18n="overview.deployment_facts.services_label">Services/Response:</span>
|
||||
<span data-i18n="overview.deployment_facts.services_value">6 governance checks</span>
|
||||
<span class="w-36 font-medium" data-i18n="what_is.facts.features_label">AI Features:</span>
|
||||
<span data-i18n="what_is.facts.features_value">4 governed features live</span>
|
||||
</li>
|
||||
<li class="flex items-center">
|
||||
<span class="w-36 font-medium" data-i18n="what_is.facts.infra_label">Infrastructure:</span>
|
||||
<span data-i18n="what_is.facts.infra_value">NZ + EU (no US dependency)</span>
|
||||
</li>
|
||||
</ul>
|
||||
</div>
|
||||
|
|
@ -136,68 +139,75 @@
|
|||
</div>
|
||||
</section>
|
||||
|
||||
<!-- Architecture Mapping -->
|
||||
<!-- Sovereign AI Architecture -->
|
||||
<section class="mb-16">
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6">Architecture Mapping</h2>
|
||||
<p class="text-gray-700 mb-8">
|
||||
Each Village AI feature maps to specific Tractatus governance services.
|
||||
The table below shows how the six services coordinate for each feature.
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="sovereign.heading">Sovereign AI Architecture</h2>
|
||||
|
||||
<div class="bg-white rounded-xl shadow-md p-8">
|
||||
<p class="text-gray-700 leading-relaxed mb-6" data-i18n="sovereign.intro">
|
||||
The Village runs its own language model — not an API call to a US hyperscaler, but a locally fine-tuned model where the training data, model weights, and inference pipeline all remain under community control.
|
||||
</p>
|
||||
|
||||
<div class="bg-white rounded-xl shadow-md overflow-hidden">
|
||||
<table class="min-w-full divide-y divide-gray-200">
|
||||
<thead class="bg-gray-50">
|
||||
<tr>
|
||||
<th class="px-6 py-4 text-left text-sm font-semibold text-gray-900">Village Feature</th>
|
||||
<th class="px-6 py-4 text-left text-sm font-semibold text-gray-900">Primary Service</th>
|
||||
<th class="px-6 py-4 text-left text-sm font-semibold text-gray-900">Function</th>
|
||||
</tr>
|
||||
</thead>
|
||||
<tbody class="divide-y divide-gray-200">
|
||||
<tr>
|
||||
<td class="px-6 py-4 text-gray-900 font-medium">Home AI Responses</td>
|
||||
<td class="px-6 py-4 text-gray-700">BoundaryEnforcer</td>
|
||||
<td class="px-6 py-4 text-gray-600">Blocks values judgments, defers to human</td>
|
||||
</tr>
|
||||
<tr class="bg-gray-50">
|
||||
<td class="px-6 py-4 text-gray-900 font-medium">User Query Processing</td>
|
||||
<td class="px-6 py-4 text-gray-700">CrossReferenceValidator</td>
|
||||
<td class="px-6 py-4 text-gray-600">Prevents prompt injection, validates intent</td>
|
||||
</tr>
|
||||
<tr>
|
||||
<td class="px-6 py-4 text-gray-900 font-medium">Session Management</td>
|
||||
<td class="px-6 py-4 text-gray-700">ContextPressureMonitor</td>
|
||||
<td class="px-6 py-4 text-gray-600">Tracks session health, triggers handoffs</td>
|
||||
</tr>
|
||||
<tr class="bg-gray-50">
|
||||
<td class="px-6 py-4 text-gray-900 font-medium">Multi-Step Operations</td>
|
||||
<td class="px-6 py-4 text-gray-700">MetacognitiveVerifier</td>
|
||||
<td class="px-6 py-4 text-gray-600">Detects scope creep, requires review</td>
|
||||
</tr>
|
||||
<tr>
|
||||
<td class="px-6 py-4 text-gray-900 font-medium">Feature Flags</td>
|
||||
<td class="px-6 py-4 text-gray-700">InstructionPersistenceClassifier</td>
|
||||
<td class="px-6 py-4 text-gray-600">Persistence classification for settings</td>
|
||||
</tr>
|
||||
<tr class="bg-gray-50">
|
||||
<td class="px-6 py-4 text-gray-900 font-medium">Consent Handling</td>
|
||||
<td class="px-6 py-4 text-gray-700">PluralisticDeliberationOrchestrator</td>
|
||||
<td class="px-6 py-4 text-gray-600">Multi-stakeholder decisions</td>
|
||||
</tr>
|
||||
</tbody>
|
||||
</table>
|
||||
<div class="grid grid-cols-1 md:grid-cols-3 gap-6 mb-6">
|
||||
<div class="bg-teal-50 rounded-lg p-5">
|
||||
<h3 class="font-semibold text-teal-900 mb-2" data-i18n="sovereign.model_title">Local Language Model</h3>
|
||||
<p class="text-sm text-teal-800" data-i18n="sovereign.model_desc">Llama 3.1 8B and Llama 3.2 3B, fine-tuned with QLoRA on community-specific data. All inference runs on the platform's own GPU infrastructure.</p>
|
||||
</div>
|
||||
<div class="bg-teal-50 rounded-lg p-5">
|
||||
<h3 class="font-semibold text-teal-900 mb-2" data-i18n="sovereign.infra_title">Sovereign Infrastructure</h3>
|
||||
<p class="text-sm text-teal-800" data-i18n="sovereign.infra_desc">Production servers in New Zealand and the EU. No data transits US jurisdiction. Community data never leaves the deployment it belongs to.</p>
|
||||
</div>
|
||||
<div class="bg-teal-50 rounded-lg p-5">
|
||||
<h3 class="font-semibold text-teal-900 mb-2" data-i18n="sovereign.training_title">Community-Controlled Training</h3>
|
||||
<p class="text-sm text-teal-800" data-i18n="sovereign.training_desc">QLoRA fine-tuning on domain-specific data with consent tracking and provenance. Communities can withdraw training data and trigger model retraining.</p>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<p class="text-sm text-gray-600" data-i18n="sovereign.link_note">
|
||||
For a detailed account of the model architecture, training approach, and governance integration, see <a href="/home-ai.html" class="text-blue-600 hover:text-blue-700 font-medium underline">Home AI / SLL: Sovereign Locally-Trained Language Model</a>.
|
||||
</p>
|
||||
</div>
|
||||
</section>
|
||||
|
||||
<!-- The Home AI Flow -->
|
||||
<!-- Polycentric Governance -->
|
||||
<section class="mb-16">
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6">The Home AI Flow</h2>
|
||||
<p class="text-gray-700 mb-4">
|
||||
When a user submits a query to Home AI, it passes through six verification stages
|
||||
before a response is generated. This flow operates in the critical execution path.
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="polycentric.heading">Polycentric Governance</h2>
|
||||
|
||||
<div class="bg-white rounded-xl shadow-md p-8">
|
||||
<p class="text-gray-700 leading-relaxed mb-6" data-i18n="polycentric.intro">
|
||||
The distinctive contribution of the Village is its governance architecture. Rather than a single operator making all decisions, the platform implements polycentric governance — multiple co-equal authorities that share structural control over how AI is used.
|
||||
</p>
|
||||
<p class="text-sm text-gray-600 mb-8">
|
||||
For a detailed account of Home AI's sovereign architecture, training approach, and governance integration, see <a href="/home-ai.html" class="text-blue-600 hover:text-blue-700 font-medium underline">Home AI / SLL: Sovereign Locally-Trained Language Model</a>.
|
||||
|
||||
<div class="grid grid-cols-1 md:grid-cols-2 gap-6 mb-6">
|
||||
<div class="border-l-4 border-teal-500 pl-5">
|
||||
<h3 class="font-semibold text-gray-900 mb-2" data-i18n="polycentric.coequal_title">Co-Equal Authority</h3>
|
||||
<p class="text-gray-600 text-sm" data-i18n="polycentric.coequal_desc">Communities maintain architectural co-governance — not just consultation rights, but structural authority over how their data is used. Drawn from te ao Māori concepts of rangatiratanga (self-determination) and kaitiakitanga (guardianship).</p>
|
||||
</div>
|
||||
<div class="border-l-4 border-teal-500 pl-5">
|
||||
<h3 class="font-semibold text-gray-900 mb-2" data-i18n="polycentric.nonparticipation_title">Right of Non-Participation</h3>
|
||||
<p class="text-gray-600 text-sm" data-i18n="polycentric.nonparticipation_desc">Members can opt out of any AI feature without losing access to the platform. AI governance defers to human judgment on values questions and never overrides community decisions.</p>
|
||||
</div>
|
||||
<div class="border-l-4 border-teal-500 pl-5">
|
||||
<h3 class="font-semibold text-gray-900 mb-2" data-i18n="polycentric.taonga_title">Taonga-Centred Design</h3>
|
||||
<p class="text-gray-600 text-sm" data-i18n="polycentric.taonga_desc">Cultural treasures (taonga) are governed as first-class objects with provenance tracking, withdrawal rights, and community authority over how they appear in AI contexts.</p>
|
||||
</div>
|
||||
<div class="border-l-4 border-teal-500 pl-5">
|
||||
<h3 class="font-semibold text-gray-900 mb-2" data-i18n="polycentric.tenant_title">Tenant-Scoped Isolation</h3>
|
||||
<p class="text-gray-600 text-sm" data-i18n="polycentric.tenant_desc">Each community operates in complete data isolation. No cross-tenant data sharing. Each tenant's governance decisions apply only within their own boundary.</p>
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<p class="text-sm text-gray-600" data-i18n="polycentric.paper_note">
|
||||
The research foundation is described in <a href="/downloads/taonga-centred-steering-governance-polycentric-ai.html" class="text-blue-600 hover:text-blue-700 font-medium underline">Taonga-Centred Steering Governance: Polycentric AI for Indigenous Data Sovereignty</a>.
|
||||
</p>
|
||||
</div>
|
||||
</section>
|
||||
|
||||
<!-- How Governance Works in Practice -->
|
||||
<section class="mb-16">
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="flow.heading">How Governance Works in Practice</h2>
|
||||
<p class="text-gray-700 mb-4" data-i18n="flow.intro">
|
||||
When a member uses any AI feature, the request passes through six governance checks before a response is generated. Each check is independent and can block or modify the request.
|
||||
</p>
|
||||
|
||||
<div class="bg-white rounded-xl shadow-md p-8">
|
||||
|
|
@ -205,69 +215,62 @@
|
|||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-teal-600 text-white flex items-center justify-center font-bold mr-4">1</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">User Query Received</h4>
|
||||
<p class="text-gray-600">User submits query via Help Chat widget or story assistance</p>
|
||||
<h4 class="font-semibold text-gray-900" data-i18n="flow.steps.1.title">Member request received</h4>
|
||||
<p class="text-gray-600" data-i18n="flow.steps.1.desc">A member asks for help, requests OCR, or uses story assistance.</p>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-teal-600 text-white flex items-center justify-center font-bold mr-4">2</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">BoundaryEnforcer Check</h4>
|
||||
<p class="text-gray-600">Is this a values question requiring human judgment?</p>
|
||||
<h4 class="font-semibold text-gray-900" data-i18n="flow.steps.2.title">Values boundary check</h4>
|
||||
<p class="text-gray-600" data-i18n="flow.steps.2.desc">Is this a values question that requires human judgment? If so, the AI defers rather than answering.</p>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-teal-600 text-white flex items-center justify-center font-bold mr-4">3</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">CrossReferenceValidator Check</h4>
|
||||
<p class="text-gray-600">Does this conflict with stored instructions or attempt injection?</p>
|
||||
<h4 class="font-semibold text-gray-900" data-i18n="flow.steps.3.title">Intent validation</h4>
|
||||
<p class="text-gray-600" data-i18n="flow.steps.3.desc">Does the request conflict with stored governance rules or attempt prompt injection? Cross-references against known instruction sets.</p>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-teal-600 text-white flex items-center justify-center font-bold mr-4">4</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">ContextPressureMonitor Check</h4>
|
||||
<p class="text-gray-600">Is session health within acceptable bounds?</p>
|
||||
<h4 class="font-semibold text-gray-900" data-i18n="flow.steps.4.title">Context and session health</h4>
|
||||
<p class="text-gray-600" data-i18n="flow.steps.4.desc">Is the session within acceptable bounds? Monitors for context pressure and triggers graceful handoff when needed.</p>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-teal-600 text-white flex items-center justify-center font-bold mr-4">5</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">Query Processing</h4>
|
||||
<p class="text-gray-600">RAG system retrieves context with permission filtering</p>
|
||||
<h4 class="font-semibold text-gray-900" data-i18n="flow.steps.5.title">Permission-filtered retrieval and response</h4>
|
||||
<p class="text-gray-600" data-i18n="flow.steps.5.desc">The sovereign Llama model generates a response using RAG context filtered by the member's permissions. All processing stays on-infrastructure.</p>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-teal-600 text-white flex items-center justify-center font-bold mr-4">6</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">Response Generation</h4>
|
||||
<p class="text-gray-600">AI generates response (Claude Haiku for non-EN, local Llama for EN)</p>
|
||||
<h4 class="font-semibold text-gray-900" data-i18n="flow.steps.6.title">Scope verification</h4>
|
||||
<p class="text-gray-600" data-i18n="flow.steps.6.desc">Is the response appropriate to what was asked? Detects scope creep and blocks responses that exceed the original request.</p>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-teal-600 text-white flex items-center justify-center font-bold mr-4">7</div>
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-emerald-600 text-white flex items-center justify-center font-bold mr-4">7</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">MetacognitiveVerifier Check</h4>
|
||||
<p class="text-gray-600">Is response appropriate to query scope?</p>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
<div class="flex-shrink-0 w-8 h-8 rounded-full bg-emerald-600 text-white flex items-center justify-center font-bold mr-4">8</div>
|
||||
<div>
|
||||
<h4 class="font-semibold text-gray-900">Delivery</h4>
|
||||
<p class="text-gray-600">Response delivered to user with source attribution</p>
|
||||
<h4 class="font-semibold text-gray-900" data-i18n="flow.steps.7.title">Delivery with attribution</h4>
|
||||
<p class="text-gray-600" data-i18n="flow.steps.7.desc">Response delivered to the member with source attribution. Every step is logged for audit.</p>
|
||||
</div>
|
||||
</li>
|
||||
</ol>
|
||||
</div>
|
||||
</section>
|
||||
|
||||
<!-- Governed Features -->
|
||||
<!-- What the Platform Delivers -->
|
||||
<section class="mb-16">
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6">Governed Features in Detail</h2>
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="features.heading">What the Platform Delivers</h2>
|
||||
|
||||
<div class="grid grid-cols-1 md:grid-cols-2 gap-6">
|
||||
<!-- RAG Help Centre -->
|
||||
<!-- Help Centre -->
|
||||
<div class="bg-white rounded-xl shadow-md p-6">
|
||||
<div class="flex items-center mb-4">
|
||||
<div class="w-12 h-12 rounded-lg bg-teal-100 flex items-center justify-center mr-4">
|
||||
|
|
@ -275,15 +278,14 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M8.228 9c.549-1.165 2.03-2 3.772-2 2.21 0 4 1.343 4 3 0 1.4-1.278 2.575-3.006 2.907-.542.104-.994.54-.994 1.093m0 3h.01M21 12a9 9 0 11-18 0 9 9 0 0118 0z"/>
|
||||
</svg>
|
||||
</div>
|
||||
<h3 class="text-xl font-semibold text-gray-900">RAG-Based Help Centre</h3>
|
||||
<h3 class="text-xl font-semibold text-gray-900" data-i18n="features.help.title">Help Centre</h3>
|
||||
</div>
|
||||
<p class="text-gray-600 mb-4">
|
||||
Vector search over indexed help content, stories, and documentation.
|
||||
Results filtered by user permissions before inclusion in context.
|
||||
<p class="text-gray-600 mb-3" data-i18n="features.help.benefit">
|
||||
Members ask questions in natural language and get answers drawn from help content, stories, and documentation — filtered by their permissions.
|
||||
</p>
|
||||
<p class="text-sm text-gray-500" data-i18n="features.help.governance">
|
||||
Governance: Values boundary check prevents AI from making judgments; intent validation blocks prompt injection attempts.
|
||||
</p>
|
||||
<div class="text-sm text-gray-500">
|
||||
<strong>Services:</strong> CrossReferenceValidator, BoundaryEnforcer
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<!-- Document OCR -->
|
||||
|
|
@ -294,15 +296,14 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M9 12h6m-6 4h6m2 5H7a2 2 0 01-2-2V5a2 2 0 012-2h5.586a1 1 0 01.707.293l5.414 5.414a1 1 0 01.293.707V19a2 2 0 01-2 2z"/>
|
||||
</svg>
|
||||
</div>
|
||||
<h3 class="text-xl font-semibold text-gray-900">Document OCR</h3>
|
||||
<h3 class="text-xl font-semibold text-gray-900" data-i18n="features.ocr.title">Document OCR</h3>
|
||||
</div>
|
||||
<p class="text-gray-600 mb-4">
|
||||
Automated text extraction from uploaded documents.
|
||||
Operates under explicit consent controls with audit logging.
|
||||
<p class="text-gray-600 mb-3" data-i18n="features.ocr.benefit">
|
||||
Upload a document and get the text extracted automatically. Useful for digitising letters, certificates, and historical records.
|
||||
</p>
|
||||
<p class="text-sm text-gray-500" data-i18n="features.ocr.governance">
|
||||
Governance: Requires explicit consent before processing. All operations are audit-logged with full provenance.
|
||||
</p>
|
||||
<div class="text-sm text-gray-500">
|
||||
<strong>Services:</strong> PluralisticDeliberationOrchestrator (consent)
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<!-- Story Assistance -->
|
||||
|
|
@ -313,16 +314,14 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M11 5H6a2 2 0 00-2 2v11a2 2 0 002 2h11a2 2 0 002-2v-5m-1.414-9.414a2 2 0 112.828 2.828L11.828 15H9v-2.828l8.586-8.586z"/>
|
||||
</svg>
|
||||
</div>
|
||||
<h3 class="text-xl font-semibold text-gray-900">Story Assistance</h3>
|
||||
<h3 class="text-xl font-semibold text-gray-900" data-i18n="features.stories.title">Story Assistance</h3>
|
||||
</div>
|
||||
<p class="text-gray-600 mb-4">
|
||||
AI-assisted writing suggestions for family stories.
|
||||
Content suggestions filtered through BoundaryEnforcer to prevent
|
||||
inappropriate recommendations.
|
||||
<p class="text-gray-600 mb-3" data-i18n="features.stories.benefit">
|
||||
AI-assisted writing suggestions for community stories and family histories. Helps with structure, prompts, and gentle editing.
|
||||
</p>
|
||||
<p class="text-sm text-gray-500" data-i18n="features.stories.governance">
|
||||
Governance: Values boundary check prevents inappropriate content suggestions; scope verification ensures the AI stays within what was asked.
|
||||
</p>
|
||||
<div class="text-sm text-gray-500">
|
||||
<strong>Services:</strong> BoundaryEnforcer, MetacognitiveVerifier
|
||||
</div>
|
||||
</div>
|
||||
|
||||
<!-- AI Memory Transparency -->
|
||||
|
|
@ -333,27 +332,25 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M9 5H7a2 2 0 00-2 2v12a2 2 0 002 2h10a2 2 0 002-2V7a2 2 0 00-2-2h-2M9 5a2 2 0 002 2h2a2 2 0 002-2M9 5a2 2 0 012-2h2a2 2 0 012 2"/>
|
||||
</svg>
|
||||
</div>
|
||||
<h3 class="text-xl font-semibold text-gray-900">AI Memory Transparency</h3>
|
||||
<h3 class="text-xl font-semibold text-gray-900" data-i18n="features.memory.title">AI Memory Transparency</h3>
|
||||
</div>
|
||||
<p class="text-gray-600 mb-4">
|
||||
User-controlled summarisation with full audit dashboard.
|
||||
Members can view, edit, and delete what AI "remembers" about them.
|
||||
<p class="text-gray-600 mb-3" data-i18n="features.memory.benefit">
|
||||
Members can see, edit, and delete what the AI "remembers" about them. Full audit dashboard shows every AI interaction.
|
||||
</p>
|
||||
<p class="text-sm text-gray-500" data-i18n="features.memory.governance">
|
||||
Governance: Multi-stakeholder consent required. Persistence decisions classified and auditable. Members control their own data.
|
||||
</p>
|
||||
<div class="text-sm text-gray-500">
|
||||
<strong>Services:</strong> InstructionPersistenceClassifier, PluralisticDeliberationOrchestrator
|
||||
</div>
|
||||
</div>
|
||||
</div>
|
||||
</section>
|
||||
|
||||
<!-- Honest Limitations -->
|
||||
<section class="mb-16">
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6">Honest Limitations</h2>
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="limitations.heading">Honest Limitations</h2>
|
||||
|
||||
<div class="bg-amber-50 border-l-4 border-amber-500 rounded-r-xl p-8">
|
||||
<p class="text-amber-900 mb-6">
|
||||
This case study documents preliminary evidence from an early-stage federated deployment across four tenants.
|
||||
We are transparent about the following limitations:
|
||||
<p class="text-amber-900 mb-6" data-i18n="limitations.intro">
|
||||
This case study documents preliminary evidence from a production multi-tenant deployment. We are transparent about the following limitations:
|
||||
</p>
|
||||
|
||||
<ul class="space-y-4">
|
||||
|
|
@ -362,8 +359,8 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M12 9v2m0 4h.01m-6.938 4h13.856c1.54 0 2.502-1.667 1.732-3L13.732 4c-.77-1.333-2.694-1.333-3.464 0L3.34 16c-.77 1.333.192 3 1.732 3z"/>
|
||||
</svg>
|
||||
<div>
|
||||
<strong class="text-amber-900">Limited Deployment:</strong>
|
||||
<span class="text-amber-800"> Village operates across four federated tenants within one platform. Generalisability to other contexts is unknown.</span>
|
||||
<strong class="text-amber-900" data-i18n="limitations.items.0.title">Small Scale:</strong>
|
||||
<span class="text-amber-800" data-i18n="limitations.items.0.description"> The Village currently serves a small number of community tenants. Generalisability to larger deployments or different community types is unknown.</span>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
|
|
@ -371,8 +368,8 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M12 9v2m0 4h.01m-6.938 4h13.856c1.54 0 2.502-1.667 1.732-3L13.732 4c-.77-1.333-2.694-1.333-3.464 0L3.34 16c-.77 1.333.192 3 1.732 3z"/>
|
||||
</svg>
|
||||
<div>
|
||||
<strong class="text-amber-900">Self-Reported Metrics:</strong>
|
||||
<span class="text-amber-800"> No independent verification of logged data has been conducted.</span>
|
||||
<strong class="text-amber-900" data-i18n="limitations.items.1.title">Self-Reported Metrics:</strong>
|
||||
<span class="text-amber-800" data-i18n="limitations.items.1.description"> No independent verification of logged data has been conducted.</span>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
|
|
@ -380,8 +377,8 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M12 9v2m0 4h.01m-6.938 4h13.856c1.54 0 2.502-1.667 1.732-3L13.732 4c-.77-1.333-2.694-1.333-3.464 0L3.34 16c-.77 1.333.192 3 1.732 3z"/>
|
||||
</svg>
|
||||
<div>
|
||||
<strong class="text-amber-900">Operator-Developer Overlap:</strong>
|
||||
<span class="text-amber-800"> Framework developer also operates Village (conflict of interest).</span>
|
||||
<strong class="text-amber-900" data-i18n="limitations.items.2.title">Operator-Developer Overlap:</strong>
|
||||
<span class="text-amber-800" data-i18n="limitations.items.2.description"> Framework developer also operates the Village (conflict of interest).</span>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
|
|
@ -389,8 +386,8 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M12 9v2m0 4h.01m-6.938 4h13.856c1.54 0 2.502-1.667 1.732-3L13.732 4c-.77-1.333-2.694-1.333-3.464 0L3.34 16c-.77 1.333.192 3 1.732 3z"/>
|
||||
</svg>
|
||||
<div>
|
||||
<strong class="text-amber-900">Limited Adversarial Testing:</strong>
|
||||
<span class="text-amber-800"> No formal red-team evaluation has been conducted.</span>
|
||||
<strong class="text-amber-900" data-i18n="limitations.items.3.title">Limited Adversarial Testing:</strong>
|
||||
<span class="text-amber-800" data-i18n="limitations.items.3.description"> No formal red-team evaluation has been conducted.</span>
|
||||
</div>
|
||||
</li>
|
||||
<li class="flex items-start">
|
||||
|
|
@ -398,8 +395,8 @@
|
|||
<path stroke-linecap="round" stroke-linejoin="round" stroke-width="2" d="M12 9v2m0 4h.01m-6.938 4h13.856c1.54 0 2.502-1.667 1.732-3L13.732 4c-.77-1.333-2.694-1.333-3.464 0L3.34 16c-.77 1.333.192 3 1.732 3z"/>
|
||||
</svg>
|
||||
<div>
|
||||
<strong class="text-amber-900">Voluntary Invocation:</strong>
|
||||
<span class="text-amber-800"> AI could theoretically bypass governance if not configured to use it.</span>
|
||||
<strong class="text-amber-900" data-i18n="limitations.items.4.title">Voluntary Invocation:</strong>
|
||||
<span class="text-amber-800" data-i18n="limitations.items.4.description"> AI could theoretically bypass governance if not configured to use it.</span>
|
||||
</div>
|
||||
</li>
|
||||
</ul>
|
||||
|
|
@ -408,49 +405,55 @@
|
|||
|
||||
<!-- What This Demonstrates -->
|
||||
<section class="mb-16">
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6">What This Demonstrates</h2>
|
||||
<h2 class="text-3xl font-bold text-gray-900 mb-6" data-i18n="demonstrates.heading">What This Demonstrates</h2>
|
||||
|
||||
<div class="grid grid-cols-1 md:grid-cols-2 gap-8">
|
||||
<div class="bg-green-50 border-l-4 border-green-500 rounded-r-xl p-6">
|
||||
<h3 class="text-lg font-semibold text-green-900 mb-4">Evidence Supports</h3>
|
||||
<h3 class="text-lg font-semibold text-green-900 mb-4" data-i18n="demonstrates.supports.title">Evidence Supports</h3>
|
||||
<ul class="space-y-2 text-green-800">
|
||||
<li>• Architectural governance can operate in production without prohibitive overhead</li>
|
||||
<li>• Six-service coordination is technically feasible</li>
|
||||
<li>• Governance violations are detectable and auditable</li>
|
||||
<li>• The framework learns from failures (documented incident responses)</li>
|
||||
<li data-i18n="demonstrates.supports.items.0">• Sovereign AI deployment is technically feasible for small community organisations</li>
|
||||
<li data-i18n="demonstrates.supports.items.1">• Polycentric governance can operate in production without prohibitive overhead</li>
|
||||
<li data-i18n="demonstrates.supports.items.2">• Multi-tenant isolation with per-community governance is achievable</li>
|
||||
<li data-i18n="demonstrates.supports.items.3">• Governance violations are detectable and auditable</li>
|
||||
<li data-i18n="demonstrates.supports.items.4">• The framework learns from failures (documented incident responses)</li>
|
||||
</ul>
|
||||
</div>
|
||||
|
||||
<div class="bg-gray-100 border-l-4 border-gray-400 rounded-r-xl p-6">
|
||||
<h3 class="text-lg font-semibold text-gray-900 mb-4">Evidence Does NOT Support</h3>
|
||||
<h3 class="text-lg font-semibold text-gray-900 mb-4" data-i18n="demonstrates.does_not_support.title">Evidence Does NOT Support</h3>
|
||||
<ul class="space-y-2 text-gray-700">
|
||||
<li>• Framework effectiveness at scale (thousands of concurrent users)</li>
|
||||
<li>• Generalisability across different AI systems</li>
|
||||
<li>• Resistance to sophisticated adversarial attacks</li>
|
||||
<li>• Regulatory sufficiency (EU AI Act compliance untested)</li>
|
||||
<li data-i18n="demonstrates.does_not_support.items.0">• Framework effectiveness at scale (thousands of concurrent users)</li>
|
||||
<li data-i18n="demonstrates.does_not_support.items.1">• Generalisability across different AI systems or model architectures</li>
|
||||
<li data-i18n="demonstrates.does_not_support.items.2">• Resistance to sophisticated adversarial attacks</li>
|
||||
<li data-i18n="demonstrates.does_not_support.items.3">• Regulatory sufficiency (EU AI Act compliance untested)</li>
|
||||
</ul>
|
||||
</div>
|
||||
</div>
|
||||
</section>
|
||||
|
||||
<!-- CTAs -->
|
||||
<!-- Explore Further -->
|
||||
<section class="cta-gradient-vcs rounded-xl p-8 text-center">
|
||||
<h2 class="text-2xl font-bold mb-4 text-black" data-i18n="cta.heading">Explore Further</h2>
|
||||
<p class="text-black mb-6 max-w-2xl mx-auto" data-i18n="cta.description">
|
||||
See the Village platform in action, or dive deeper into the technical architecture.
|
||||
Dive deeper into the technical architecture, read the research, or see the Village platform in action.
|
||||
</p>
|
||||
<div class="flex flex-wrap gap-4 justify-center">
|
||||
<a href="https://mysovereignty.digital"
|
||||
<a href="https://community.myfamilyhistory.digital"
|
||||
target="_blank"
|
||||
rel="noopener noreferrer"
|
||||
class="inline-block bg-white text-teal-700 px-8 py-3 rounded-lg font-semibold no-underline shadow-md hover:shadow-lg transition"
|
||||
data-i18n="cta.visit_village">
|
||||
Visit the Village →
|
||||
</a>
|
||||
<a href="/architecture.html"
|
||||
<a href="/home-ai.html"
|
||||
class="inline-block bg-teal-900 text-white px-8 py-3 rounded-lg font-semibold no-underline border-2 border-teal-400 hover:bg-teal-800 transition"
|
||||
data-i18n="cta.system_architecture">
|
||||
System Architecture →
|
||||
data-i18n="cta.home_ai">
|
||||
Sovereign Language Model →
|
||||
</a>
|
||||
<a href="/downloads/taonga-centred-steering-governance-polycentric-ai.html"
|
||||
class="inline-block bg-teal-900 text-white px-8 py-3 rounded-lg font-semibold no-underline border-2 border-teal-400 hover:bg-teal-800 transition"
|
||||
data-i18n="cta.research_paper">
|
||||
Research Paper →
|
||||
</a>
|
||||
<a href="/researcher.html"
|
||||
class="inline-block bg-teal-900 text-white px-8 py-3 rounded-lg font-semibold no-underline border-2 border-teal-400 hover:bg-teal-800 transition"
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue