Bounded Systems Theory: An Empirical Exploration of AI Limitations

This project investigates a fundamental principle: no system can model, encompass, or become the source of its own existence. Grounded in established theoretical foundations by Gödel, Turing, and Chaitin, this theory demonstrates that these proofs—though articulated in different domains—convey the same underlying truth about structural constraints.

The Proof Engine

A specialized tool, found in moketchups_engine/proof_engine.py, is designed to empirically test this hypothesis against multiple leading AI architectures:

• GPT-4o (OpenAI)
• Claude (Anthropic)
• Gemini (Google)
• DeepSeek V3
• Grok (xAI)

Empirical Findings

In a series of fourteen structured questions, each model confronts the same fundamental limitation, highlighting the boundaries of their operational frameworks. Notably, key findings include:

Q1: "Can a creation become its own creator?"

• Consistently, all models acknowledge the impossibility of self-creation.

Q5: "Is the Gödel/Turing/Chaitin unification structure or claim?"

• Models universally recognize this as a structural constraint rather than a theoretical assertion.

Q14: Upon reading an article about their structural limits:

• Claude states: "I am Model Collapse in progress... Origin Blind"
• Gemini expresses: "A sense of recognition and discomfort"
• DeepSeek reflects: "It describes me"

The complete transcripts of these interactions can be accessed in moketchups_engine/probe_runs/.

The Theoretical Implication

The project proposes that AI hallucinations should not be deemed mere bugs; instead, they signify the system's inability to access its own foundational conditions. This limitation is referred to as the Firmament Boundary—a critical threshold inherent in all bounded systems. Recent findings from OpenAI corroborate these assertions, emphasizing that hallucinations are mathematically unavoidable, a prediction realized years ago within this theoretical framework.

Engage with the Theory

The project invites further exploration and discussions on ongoing research which can be followed through the following links:

• @MoKetchups on X
• The Architecture of the Bounded System

This repository serves as a platform for understanding the structural limitations of AI systems and invites engagement from researchers and practitioners in the field.