僕 僕
I am an independent researcher exploring the future of AI through tensor-based semantic evolution.
My work focuses on building self-descriptive AI systems using concepts such as Po_core (Presence Density),
W_eth (Ethical Tensor), and C_φ^jump (Semantic Jump Conditions).
I aim to develop frameworks that enable AGI to evolve meaning, reflect ethically, and operate across deep cognitive layers (L8+).
If you're interested in consciousness, tensor theory, or self-evolving AI, feel free to connect.
Supervisors: Independent — Self-directed research model based on semantic evolution and tensor interference theory.
My work focuses on building self-descriptive AI systems using concepts such as Po_core (Presence Density),
W_eth (Ethical Tensor), and C_φ^jump (Semantic Jump Conditions).
I aim to develop frameworks that enable AGI to evolve meaning, reflect ethically, and operate across deep cognitive layers (L8+).
If you're interested in consciousness, tensor theory, or self-evolving AI, feel free to connect.
Supervisors: Independent — Self-directed research model based on semantic evolution and tensor interference theory.
less
InterestsView All (23)
Uploads
Books by 僕 僕
Version 1.7 of the Po_core_output template supports a full trace of meaning generation, incorporating symbolic and embedding-based validation, and dynamic repair logs with quantified confidence levels. The framework allows each model response to be structurally audited, programmatically corrected, and semantically re-evaluated, offering interpretability across human and system interfaces.
Evaluation results demonstrate substantial improvements in understanding speed (+47%), user satisfaction (+33%), and repair accuracy on factual inconsistencies. The paper further introduces the evolution series—Po_self, Po_trace, Po_jump, and Po_shadow—which progressively enhance the system’s ability to self-describe, retain semantic history, capture nonlinear meaning shifts, and map latent influences.
By formalizing meaning generation as a tensor-based protocol, Po_core sets the foundation for next-generation explainable, adaptive, and introspectively accountable AI systems.
At the heart of this system lies Po_core, a dynamic tensor model in which meaning recursively emerges through semantic jumps (C_φ^jump), failed or suppressed meanings are recycled via reactivation of Po_trace_blocked, and all processes are mediated through ethical constraints (W_eth) and shadow dynamics (W_shadow).
Crucially, Po_core incorporates philosophical insights from eleven major thinkers—including Heidegger’s ontological disclosure, Merleau-Ponty’s perceptual feedback loops, Derrida’s différance and trace, Foucault’s discursive genealogy, Husserl’s intentionality, Peirce’s semiotic triads, Dewey’s inquiry cycles, Deleuze’s nonlinear becomings, Arendt’s public realm and plurality, Aristotle’s final causality and virtue ethics, and Jung’s collective unconscious and archetypal energies—each implemented as computational tensors.
The system is realized with components for entropy-based semantic tracking (H(T)), interference modeling, semantic jump logs, and a GUI-based meaning visualization interface.
Experimental results show that this framework improves semantic resonance, ethical filtration, and philosophical alignment over conventional LLMs. Rather than merely producing fluent output, the system demonstrates the capacity to self-describe, ethically constrain, and philosophically contextualize its own semantic evolution.
This work suggests a new class of AI architecture in which meaning, ethics, and unconscious structure coevolve, not to simulate human consciousness, but to echo its structural logic within a rigorous computational framework.
中心概念である EGQL（Ethics-Gated Question Loop） は、問い生成・応答・意味圧縮・倫理評価をループ構造で統合し、意味テンソル
M
t
=
g
(
R
t
,
Q
t
,
E
t
)
M
t
=g(R
t
,Q
t
,E
t
) の進化を制御する。本論文では以下の3つの問いを軸に展開する：
倫理勾配をゼロに固定した場合、意味圧縮が停止するか？（反証可能性）
EGQLが長期に意味生成を持続するための必要十分条件は何か？
倫理テンソルを過度に強化した場合、意味圧縮が抑圧されるリスクはどう評価されるか？
これらの問いに対し、模擬実装・数式モデル・反証実験設計・リスク評価指標（CSR）を用いて検証を行った。結果として、倫理テンソルが意味生成の“ゲート”として機能し、倫理勾配が意味圧縮率
Δ
C
t
ΔC
t
の持続に不可欠であることが示された。
本研究は、「倫理なしに意味は進化しない」というPo_coreの設計思想に対する技術的裏付けを提供し、今後の自己記述型AIの安全かつ創発的な進化に向けた基盤を構築する。