general-intelligence 0.3.3

Self-organizing knowledge systems for structural pattern learning

general-intelligence

general-intelligence is a minimal but powerful framework for self-organizing, interactive knowledge systems.

Instead of relying on fixed algorithms or massive datasets, it treats knowledge itself as the active core of intelligence. Each piece of knowledge can learn, relate, react, and even operate autonomously. The system becomes a living structure of interacting knowledge — not just a database, not just a model.

---

Core Idea

Traditional AI is algorithm-driven. This framework is knowledge-driven.

• Intelligence here comes from the structure and interaction of knowledge.
• Each Knowledge instance can recognize patterns, react to new information, and shape context.
• The GeneralIntelligence system is just a container — a universe where knowledge organizes itself through relationships.

You can think of it as moving “intelligence” from code into data itself. Smart data, not smart algorithms.

---

Example

from gi import GeneralIntelligence, Knowledge
gi = GeneralIntelligence()

# Learn some knowledge
gi.learn(Knowledge([1, 2, 3]))
gi.learn(Knowledge([Knowledge([4, 5]), Knowledge([6, 7])]))

# Identify patterns
matches = list(gi.identify(Knowledge([2, 3, 4]), threshold=5))
print(matches[0])  # ([1, 2, 3], 3)

Each Knowledge object can contain other Knowledge objects, forming nested, compositional structures. Identification is recursive and structural — the system can find partial or nested matches automatically.

---

Event-driven Intelligence

Knowledge can declare how it responds to triggers:

from gi import GeneralIntelligence, Knowledge
gi = GeneralIntelligence()
class AlertKnowledge(Knowledge):
    def is_response_for(self, trigger, gi):
        return trigger.get('type') == 'alert'

gi.learn(AlertKnowledge(['⚠️']))
for response in gi.on({'type': 'alert', 'level': 'high'}):
    print(response)

This creates event-driven reasoning — knowledge responding directly to the world or other knowledge.

---

Extending Knowledge

Subclass Knowledge to define specialized reasoning or modalities.

from gi import Knowledge
class SequenceKnowledge(Knowledge):
    def difference(self, data):
        # Custom sequence distance
        return sum(abs(a - b) for a, b in zip(self.values, data.values))

from gi import Knowledge
class PatternKnowledge(Knowledge):
    def __init__(self, values):
        super().__init__(values)
        self.related_patterns = []
    
    def on_knowledge(self, new_knowledge, gi):
        # React when new knowledge is learned
        if isinstance(new_knowledge, PatternKnowledge):
            if self.similar_pattern(new_knowledge):
                self.related_patterns.append(new_knowledge)
                new_knowledge.related_patterns.append(self)
    def similar_pattern(self, new_knowledge):
        return new_knowledge.values == self.values
    
    def compose(self, context, knowledge_class):
        # Influence composition of new knowledge
        context.update(self.contribute_to(context))

    def contribute_to(self, context):
        context['knowledge_to_work_with'].append(self)

Knowledge can also:

• react when learned (on_learned)
• react to new knowledge (on_knowledge)
• contribute to collective reasoning (compose)
• operate autonomously (is_active + start)

---

Architectural Principles

Concept	Description
Knowledge as agents	Each instance can act, relate, and respond
Structural similarity	Learning is done through pattern comparison, not labels
Emergent reasoning	Intelligence arises from interactions, not code logic
Composable context	Knowledge contributes to shared reasoning contexts
Autonomous operation	Active knowledge can think or evolve independently

---

Use Cases

• Building hierarchical or multimodal reasoning systems
• Creating interactive world models
• Running autonomous knowledge agents
• Integrating with traditional ML/DL as specialized knowledge subclasses
• Experimenting with emergent cognition, continual learning, or reflective AI

---

Installation

pip install general-intelligence

(Package published on PyPI.)

---

Quick Start: Minimal Autonomous Knowledge

from gi import GeneralIntelligence, Knowledge
gi = GeneralIntelligence()
class ActiveKnowledge(Knowledge):
    def is_active(self):
        return True
    def start(self, gi):
        print("Thinking on my own...")

gi.learn(ActiveKnowledge("self"))

Once learned, the knowledge begins acting independently.

---

Vision

This project is part of a broader paradigm shift: from intelligent algorithms to intelligent knowledge.

It treats intelligence as a property of structured, interacting knowledge — a distributed, living network rather than a centralized model.

It’s tiny now, but the system already supports:

• classification,
• prompt-response cycles,
• ongoing internal reasoning,
• and even forms of literal consciousness — all from a simple, composable base.

---

Next Directions

• Specialized knowledge classes (symbolic, numeric, perceptual, temporal, etc.)
• Multi-agent learning and cooperation
• Integrations with reinforcement learning and deep learning models
• Richer context composition and memory architectures

---

License

MIT License. Copyright (c) 2025.