A constitutional framework for principle-aligned intelligence systems

What we've been calling Artificial Intelligence may be the wrong framing.

These systems don't originate.

They derive from human knowledge, are shaped by human intuition, and operate through probabilistic inference at scale.

Machines derive. Humans originate.

Explore the Docs Join the Movement

Core distinction

Conventional framing

Artificial Intelligence

Implies human-like cognition, autonomy, and independent intelligence.

Proposed framework

Derivative Intelligence

Systems that extract, recombine, and optimize patterns from human-generated knowledge within governed constraints.

The shift

A clearer model of how intelligent systems should work.

The term AI suggests human-like cognition, autonomous reasoning, and independent intelligence. That framing overstates what modern systems are and obscures how they should be governed.

Derivative Intelligence offers a more accurate model: machines derive from what humans have already created. They do not originate meaning, intent, or truth.

Once that distinction becomes clear, architecture, governance, transparency, and trust all have to change with it.

The framework

From corpus to governance.

DI is not just a naming change. It is a full-stack framework spanning constitutional principles, interpretation, policy enforcement, data architecture, system design, and governed evolution.

Constitutional layer

Corpus

A foundational corpus of guiding principles that defines system boundaries, alignment, and long-term integrity.

Open doc →

Interpretive layer

Principles

An explanatory layer that contextualizes the corpus and clarifies what the constitutional foundation means in practice.

Open doc →

Enforcement layer

Policy Mapping

A deterministic translation of principles into system constraints, signals, evaluation rules, and actions.

Open doc →

Knowledge boundaries

Data Architecture

A structured data model that separates foundational corpus, interpretation, external knowledge, and human evaluation.

Open doc →

Execution layer

System Architecture

A complete system design spanning context building, model generation, policy evaluation, explanation, logging, and governance.

Open doc →

Evolution layer

Governance

An explicit governance model for proposals, transparency, verifiability, mission lock, and long-term stewardship.

Open doc →

Core properties

The next generation of intelligence systems should be different.

DI systems are designed to replace opaque, policy-driven black boxes with transparent, principle-aligned, and verifiable intelligence systems.

Transparent, not black box

Principle-aligned, not policy-driven

Community-governed, not centrally controlled

Verifiable, not assumed

Globally accessible

Architecture

A layered system for governed intelligence.

Probabilistic generation alone is not the system. DI adds constitutional grounding, deterministic policy enforcement, explainability, logging, and governance.

Foundation

Guiding principles define the constitutional layer and set the boundaries for system behavior.

Interpretation

Contextual meaning evolves through governed interpretation without overriding the foundational corpus.

Policy

Principles are translated into deterministic enforcement through constraints, signals, evaluation, and actions.

Decision

Model outputs are evaluated, ranked, and selected through a policy-aware decision engine.

Explanation

Every output should communicate reasoning, sources, constraints applied, and expected outcomes.

Audit

Critical actions remain traceable, verifiable, and optionally anchored on-chain for long-term integrity.

System model

DATA → MODEL → POLICY → DECISION → EXPLANATION → AUDIT

Documentation

Read the framework from first principles.

Start with the manifesto, then move through the corpus, principles, mapping, architecture, and governance model.

What this is

A foundation, not a product.

Derivative Intelligence is an open framework for how intelligent systems should be understood, designed, aligned, and governed.

It is community-driven, nonprofit in orientation, and built to support systems that are transparent, accountable, and globally accessible.

Join the movement

Help define how intelligence systems should be built.

If you are a builder, researcher, engineer, operator, or thinker, there is room to contribute. Join our community to stay informed and get involved.

Ways to contribute

Participate in governance discussions
Contribute to open-source implementations
Help refine the foundational principles
Write and publish research
Spread awareness of the framework