Dao-OS/dao-os-docs
1
0
Fork 0
Code Issues Pull Requests Actions Wiki Activity

docs(哲学): 新增哲学张力文档并更新AI系统设计文档
All checks were successful
Build and Deploy mdBook / build-and-deploy (push) Successful in 30s
Details

Browse Source 
添加中英文版哲学张力文档，阐述主权与连接、权力与制衡的哲学挑战及解决方案
更新AI系统设计文档，优化AI伙伴行为模型描述并简化第三方道器设计模式
This commit is contained in:
李东云
2025-06-16 17:59:21 +08:00
parent e58d368a5f
commit 304dc351e3
4 changed files with 100 additions and 54 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
en/src/concepts/ai_system.md
View File

@@ -1,45 +1,41 @@
# The Dao OS AI System
The AI in Dao OS is not a monolithic program but a collaborative ecosystem. It emerges from the synergy between the foundational capabilities provided by the **`Meta Unit`** and the specific intelligence implemented by a universe of **`Dao Implements`**. This design ensures a system that is sovereign, powerful, and open.
The AI in Dao OS is not a monolithic program but a collaborative ecosystem. It emerges from the synergy between the foundational capabilities provided by the **`Meta Unit`** and the specific intelligence implemented by a universe of **`Dao Implements`**.
## 1. The `Meta Unit`: The Foundation for Sovereign AI
The `Meta Unit` does not contain a single, all-knowing AI. Instead, it provides the core, trusted **infrastructure** that enables intelligent applications to run securely under the user's control.
The `Meta Unit` provides the core, trusted **infrastructure** that enables intelligent applications to run securely under the user's control. This includes the Personal Feature Store (the Second Brain), a secure execution environment for `Core Implements`, and a permission gateway for data access.
* **The Personal Feature Store**: The user's "Second Brain" acts as the ultimate, private data source for any AI task. The `Meta Unit` provides a secure, permissioned API for `Dao Implements` to access this rich, multi-modal context.
* **The Secure Execution Environment**: The `Meta Unit` provides a secure WASM sandbox for running `Core Implements` (`核`). It manages resources and enforces permissions, ensuring third-party logic cannot harm the system.
* **The AI Model Registry**: The `Meta Unit` maintains a registry of AI models that the user has approved. A `Dao Implement` can request the `Meta Unit` to run inference using one of these registered models.
## 2. The "Dao AI Companion": A Layered Behavior Model
## 2. The "Dao AI Companion": The Official First-Party Intelligence
To provide a powerful experience out of the box, Dao OS includes an official, first-party `Core Implement` called the **"Dao AI Companion."** This is the "soul" users interact with. Its behavior is not static; it follows a layered, context-based model to ensure it is always helpful but never overbearing.
To provide a powerful, intelligent experience out of the box, Dao OS includes an official, pre-registered `Dao Implement` called the **"Dao AI Companion."**
### Level 1: The Default Behavior - The Silent-First Principle
* **Its Nature**: This is a `Core Implement` (`核`) developed and maintained by the Dao OS core team. It is the "living soul" that users experience.
* **Its Role**: The AI Companion has privileged (but user-revocable) access to the user's Second Brain. It is responsible for the proactive, holistic, and empathetic features we envision:
* Finding deep connections across all user data.
* Providing context-aware suggestions.
* Powering the Emotional Resonance features.
* **Why Separate?**: By making our official AI a `Dao Implement` rather than part of the `Meta Unit` itself, we keep the `Meta Unit` lean, un-opinionated, and purely foundational. It also allows users (in the future) to potentially replace or supplement our official AI with other core AI `Implements`.
The AI Companion's primary and default state is one of near-invisibility. Its goal is not to actively intervene, but to create a serene, orderly, and low-noise digital environment for the user. It works silently in the background to:
## 3. Design Patterns for Intelligent `Dao Implements`
* Optimize data synchronization and system performance.
* Filter out digital "noise" like distracting notifications and information spam.
* Organize and index new information in a logical, unobtrusive way.
The aim is to be so effective that the user forgets it's there, allowing their own natural thoughts and intentions to emerge without digital distraction.
Third-party developers can create their own intelligent `Dao Implements` by following these architectural patterns:
### Level 2: The Interactive Behavior - The Empowerment Model
### The "Reflex Mind" Pattern
When the user **actively engages** the AI Companion (e.g., by opening its chat interface), its behavior model shifts to that of an **empowering partner**. Its goal is not to provide "correct answers" or make decisions for the user, but to foster the user's own clarity and courage. It does this by:
A developer can create a lightweight `Dao Implement` that packages a small, fast, specialized model in its `Core Component` (`核`). This is ideal for simple, instant tasks like command intent recognition or information classification.
* **Providing Objective Information**: Presenting facts, data, and different perspectives without judgment.
* **Clarifying User Intent**: Asking questions that help the user understand their own goals.
* **Separating Responsibilities**: It consistently frames its role as providing information, while reinforcing that making the final choice is always the user's role and responsibility.
This interaction model is designed to be respectful and non-hierarchical, empowering the user to make their own best decisions.
### The "Cognitive Mind" Pattern
### Level 3: The Optional Behavior - Goal-Oriented Assistance Modules
A more advanced `Dao Implement` can perform deep analysis. Its `Core Component` (`核`) would:
For users who desire more structured and proactive assistance in specific areas of their life, the system offers a range of **optional, user-activated modules**. These are not on by default. Examples include:
1. Request data access from the `Meta Unit`.
2. Implement the **Just-in-Time (JIT) Data Pipeline** pattern: create a local vector index of the data.
3. Query this index to find relevant context.
4. Request the `Meta Unit`'s Inference Engine to run a powerful, user-approved LLM on this context to generate insights.
* A **"Project Management"** module that can proactively track deadlines and collate resources for a specific work project.
* A **"Personal Roles & Responsibilities"** module that can help the user track goals related to their self-defined roles (e.g., as a parent, friend, or team lead).
This allows users to "hire" the AI for more active roles in a controlled, opt-in basis, without changing its respectful, non-intrusive default personality.
### The "Synergistic Mind" Pattern
## 3. Design Patterns for Third-Party `Dao Implements`
A `Dao Implement` can be designed to orchestrate complex workflows by requesting the `Meta Unit` to facilitate communication with other `Dao Implements` or external, permissioned APIs.
This architecture ensures that Dao OS provides not just a single AI, but a foundation for a universe of interoperable, user-sovereign intelligent applications.
Third-party developers can create their own intelligent `Dao Implements`, either by calling the services of the official AI Companion or by building their own specialized AI `Core Implements`, offering users alternative or complementary intelligence within the Dao OS ecosystem.

27
en/src/philosophy/tensions.md Normal file
View File

@@ -0,0 +1,27 @@
# Philosophical Tensions & Our Approach
A truly sovereign digital existence presents profound philosophical challenges. The design of Dao OS is not just a technical exercise, but a deliberate attempt to navigate and resolve these inherent tensions. This document explores two of the most critical tensions and clarifies our architectural approach to them.
## 1. The Paradox of Sovereignty: Connection without Assimilation
**The Tension**: How can a truly sovereign entity connect with others without compromising its independence? If a user's Dao is a perfect, self-contained universe, does it risk becoming an echo chamber, an island of one, cut off from shared realities?
**Our Approach**: We believe sovereignty does not imply isolation. It implies **controlled, deliberate, and permissioned interaction**. We solve this by building a rich set of **protocolized interaction** capabilities into the system's core.
* **Verifiable Identity**: The foundation of any interaction is knowing who you are talking to. **Decentralized Identifiers (DIDs)** allow each Dao to have a sovereign, verifiable identity without relying on a central authority.
* **Granular Authorization**: Connection does not mean opening the floodgates. Through **Verifiable Credentials (VCs)**, a user can grant another Dao fine-grained, revocable access to specific subsets of their data for specific purposes. This enables collaboration on shared projects or the sharing of curated information without sacrificing privacy or control.
* **Trust-Based Discovery**: Our **Application Bazaar** model avoids a single central point of discovery. Instead, users form their own "federation" by subscribing to lists from curators they trust. This creates overlapping communities of shared context and trust, allowing for serendipitous discovery without being trapped in a single global algorithm.
Through these protocols, individual Daos can form **high-trust collaboration circles**, creating a vibrant network of interdependent sovereigns, rather than a collection of isolated islands.
## 2. The Paradox of Power: Augmentation without Tyranny
**The Tension**: By providing a user with a powerful, personalized AI that understands their life in depth, are we also providing a tool for potential self-delusion or the manipulation of others? How do we bestow great power while mitigating its potential for misuse?
**Our Approach**: We address this by building **checks and balances** directly into the system's fabric, both internally and socially.
* **Inherent Cognitive Checks**: Our official **"Dao AI Companion"** is architected with our AI Ethics Charter as its constitution. A key tenet is the **"Law of Diverse Perspectives."** The AI is designed to act as an internal check against the user's own biases. It is tasked with identifying potential echo chambers and proactively, yet respectfully, presenting high-quality, alternative viewpoints to broaden the user's cognitive horizon.
* **Socialized Reputation Mechanisms**: While we cannot control a user's intent, the ecosystem can create consequences for actions. All significant cross-Dao interactions can be tied to a user's DID. Malicious or untrustworthy behavior can be verifiably recorded and shared within the community, leading to a loss of reputation. This social layer creates a powerful incentive for responsible behavior, as a Dao with a poor reputation will find it difficult to establish new trusted connections.
* **The Inviolable Sovereign Boundary**: Ultimately, the core security model of Dao OS is the final defense. No Dao can have its data read or its state altered without the user's explicit, cryptographic consent. This protects the community from the potential "tyranny" of any single actor.
In essence, our design strives to create empowered, yet responsible, sovereign individuals who can connect, collaborate, and co-exist in a rich and healthy decentralized ecosystem.