AILP — Protokolu Permisaun Aprendizajen AI
Protokolu Permisaun Aprendizajen AI
AI bele aprende husi ida ne'e ka lae? Iha profundidade saida, ba uzu saida, ho obrigasaun saida?
01 · Definisaun
AILP operationalizes the AIRS spectrum for the specific question of learning. It distinguishes acts that binary crawler rules collapse together: being read is not being learned from; being retrieved is not being trained on; fine-tuning is not distillation; verbatim memorization is different from statistical internalization. AILP lets a publisher declare, in a single machine-readable file, per-dimension permissions — access, indexing, inference input, embedding, training, fine-tuning, distillation, memory, output, attribution and compensation — each as allowed, denied, or license-required. An AILP grant answers "may AI learn from this content?" only — it does not answer "may this Agent invoke a tool, write to a database, or publish?" (AARS), nor "who authorized this Agent?" (AADP).
02 · Objetivu
- Hataan pergunta "AI bele aprende husi ida ne'e ka lae?" ho presizaun iha nivel dimensaun, la'ós bit crawl ida de'it.
- Fahe permisaun aprendizajen (internaliza ba modelu) husi permisaun asesu (buka pájina).
- Suporta modelu kompensasaun: grátis, la-komersiál, presiza-lisensa, fahe-rendimentu.
- Garante katak abertura bele exekuta liuhusi máquina — atu boa-vontade la sai deskarta hanesan insertesa legál.
03 · Ámbitu
asesu / indeksasaun — buka no indeksasaun semántiku
inference_input — uza hanesan kontestu iha tempu inferénsia (RAG)
embedding — vetorizasaun no indise rekuperasaun
training / fine_tuning / distillation — nivel internalizasaun modelu
verbatim_memory — se reprodusaun exatu mak permitidu
atribuisaun / kompensasaun — deveres sitasaun no kondisaun pagamentu
04 · Ezemplu ne'ebé máquina bele lee
Deklarasaun permisaun aprendizajen — /ai/rights-spectrum.json (perfil AILP)
{
"version": "0.1",
"protocol": "AILP",
"publisher": "example.org",
"default": {
"access": "allowed",
"indexing": "allowed",
"inference_input": "allowed",
"embedding": "allowed",
"training": "license_required",
"fine_tuning": "license_required",
"distillation": "denied",
"verbatim_memory": "denied",
"attribution": "required",
"compensation": "contact"
},
"contact": "licensing@example.org"
} 05 · Limitasaun
- Semántika permisaun-aprendizajen mak difísil liu atu verifika teknikamente — AILP deklara intensaun, seidauk bele prova kumprimentu.
- Lista dimensaun mak esbosu; limite entre treinamentu, ajustamentu finu, no destilasaun sei debate hela.
- Rekoñesimentu legál ba permisaun aprendizajen varia tuir jurisdisaun no seidauk rezolve.
- v0.1.1: AILP(Resource,ContentUse)=Allow does not imply AARS(Actor,Action)=Allow, and does not imply AADP(Principal,Actor,Authority)=Valid — a system exposing both content and tools must evaluate them separately.