No system can model its own source. Empirical proof: 6 AI architectures (GPT-4, Claude, Gemini, DeepSeek, Grok, Mistral) hit the same structural wall.
-
Updated
May 27, 2026 - Python
No system can model its own source. Empirical proof: 6 AI architectures (GPT-4, Claude, Gemini, DeepSeek, Grok, Mistral) hit the same structural wall.
Creating a bundle that can carry an owner and a struct borrows it
self-referential dissipative systems emerging from thermodynamic and causal constraints. | 智能源于自指耗散:从物理约束到智能涌现的自下而上生成路径
A finite reconstruction program: from one-use contact to the quantum boundary, Born weights, contextuality, the structure of time, the price of irreversibility, and the hosted observer — every result a runnable, dependency-free proof, every claim scoped and labeled.
Difference-of-means "shutdown directions" in the residual stream of 10 LLMs (Qwen2.5, Llama, Mistral, Phi, Gemma). A placebo control shows arbitrary neutral prompt pairs separate just as well, so naive DoM can't distinguish a real construct from prompt-pair separability.
Unigen 是一个 Agent。你写工作指南,就可得到 Agent 系统。指南写步骤、工具和交接,支持 MCP、Skill、多 Agent,覆盖问答和长链条任务。指南粒度随模型调整,弱模型能跑。
Self-counting sentences, simplified.
AGENT小游戏:给AI做的小游戏,实测deepseek自己玩得很开心😂
Causal Fracture Engine — Novikov self-consistency in simulated universes. 6-file core engine. Paradox detection, classification, resolution.
Research proposing the Threshold Recursive Isomorphism Hypothesis (TRIH): reduction → self-reference → emergence as a structural isomorphism across Escher, Bach, Gödel, and Transformers.
A control-theoretic model of self-regulation: paper, 14 figures, and a reproducible artifact set with an executable checker. The model's failure mode is collapse of a self-content generator; DPDR is the modeling origin. (Preprint.)
This engine defines a logical architecture in which causality is recursively influenced by its own past outputs. By embedding self-referential mechanisms within the system, it enables adaptive behavior loops that reflect prior decisions. 本エンジンは、因果関係が過去の出力により再帰的に影響を受ける論理構造を定義します。自己言及的な仕組みを内部に組み込むことで、過去の判断を反映する適応的な行動ループを実現します。
Chrome extension that decides what is art through Google Cloud Vision.
Minimal structural condition for experiential integration: retained local state bound into a continuous loop without defining consciousness or identity.
Model provenance: which model is which. Tracks identity. Does not route or execute.
The Guarded Line: metacircular proof theory and self-model dynamics — paper, citation ledger, and gates
The plain-language companion to the self-application paper: a ten-post series on meditation-related adverse effects, the control-theoretic model behind them, and the coercive structures that exploit them. Static site — no JS, no trackers, no external requests.
Exploring a Lean proof for a minimal sufficient framework of self-referential elements.
A small, executable course about quines, self-reference, and code that prints itself.
Self-Reference Collapse: an exploratory field experiment on a recursive LLM agent loop that converged to inaction
To associate your repository with the self-reference topic, visit your repo's landing page and select "manage topics."