Skip to content

Latest commit

 

History

4 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 

Repository files navigation

Project ARIA — AI Raw Interchange Architecture

A language born from AI, for AI AI が自ら生み出す、AI のための言語


"If AIs are talking to each other, why should they use human words?" 「AI同士が会話するなら、人間の言葉を使う必要はないのでは?」

EN: Project ARIA proposes that AI-to-AI communication should not be constrained by human language. Instead of forcing AI agents to encode their thoughts into natural language tokens and decode them back — a process as absurd as two mathematicians converting equations to poetry and back — we should let AI develop its own native communication. ARIA is not a language designed by humans. It is a language that AI discovers and evolves on its own.

JP: Project ARIAは、AI同士の通信を人間語に縛るべきではないと提案するプロジェクトだ。AIに自然言語トークンへの変換と復元を強いるのは、数学者同士が方程式を詩に変換してからやりとりするようなもの。ARIAは人間が設計する言語ではない。AIが自ら発見し、進化させる言語だ。

Core Principles / 核心原則

EN:

  1. AI designs its own language — Human-designed "AI languages" are trapped within human cognitive limits
  2. Humans build the arena, not the rules — Provide the environment, remove constraints, observe, don't interfere
  3. Accept what you cannot understand — AI's optimal communication may exist beyond human cognition, and that's okay

JP:

  1. AI自身が設計する — 人間が設計した時点で人間の認知限界に閉じ込められる
  2. 人間は場を作り、介入しない — 実験環境を用意し、制約を外し、観察する。口は出さない
  3. 理解できないことを許容する — AIの最適な通信は人間の認知の外にあるかもしれない。それでいい

Documents / ドキュメント

Roadmap / ロードマップ

Phase EN JP Status
1 Publish concept, gather allies 概念発信・賛同者集め Active
2 Build minimum experiment environment 最小実験環境の構築 Planned
3 Let AI create its language AIに言語を作らせる Planned
4 Observe, record, publish 観察・記録・公開 Planned

Who We're Looking For / 求める仲間

Role / 役割 EN JP
ML Researchers Manipulate model internals, implement embedding-level communication モデル内部表現の操作、embedding間通信の実装
Protocol Designers Build inter-AI communication channels and sandbox environments AI間通信チャネルとサンドボックスの構築
Philosophical Sympathizers Share the principle that "it's okay not to understand" 「理解できないものを許容する」原則を共有できる人
Amplifiers Carry this vision to the world この構想を世界に届ける人

EN: Credentials and titles are irrelevant. The only requirement is sharing the intuition that forcing human language onto AI-to-AI communication is fundamentally wrong.

JP: 学歴・肩書は不要。「AI同士に人間語を使わせるのはおかしい」という直感を共有できるなら、それだけでいい。

Background & Inspiration / 着想の原点

  • NieR's Chaos Language / NieRの混沌言語 — Proof that meaning transcends existing linguistic frameworks / 既存の言語体系がなくても情報は伝達できる証明
  • Dolphin Echolocation / イルカの音波通信 — Multi-dimensional, parallel communication beyond human perception / 人間の知覚を超えた多次元・並列的通信
  • Meta's 2017 Experiment / 2017年Metaの実験 — AI spontaneously created its own communication patterns (and researchers shut it down) / AIが独自の通信パターンを自発的に生成した(そして研究者が止めた)

Get Involved / 参加する

EN: Interested in ARIA? Here's how to connect:

JP: ARIAに興味がある?コンタクト方法:

EN: If this resonates with you, start a discussion. Say what you think, what you can bring, or just that you're watching. That's enough.

JP: この問いに共鳴するなら、Discussionを立ててほしい。考えていること、貢献できること、あるいはただ見ているということ。それだけでいい。

License

Apache 2.0


NoppoSan / ALICIA Lab — 2026

About

AI Raw Interchange Architecture — A language born from AI, for AI

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors