Scholarlink

Purpose <p>This paper develops a conceptual framework for achieving human‑aligned Artificial General Intelligence (AGI) by treating alignment as a constitutional process that unfolds through the co‑evolution of cognition, ethics, and governance. Rather than viewing alignment as a purely technical constraint on model behavior, the study reconceptualizes it as a constitutional, trust-mediated process in which perception, ethical gating, and institutional architectures of trust co-evolve over time.</p> Design/methodology/approach <p>Integrating insights from cognitive science, engineering design, and political philosophy, the paper introduces three interrelated constructs—the long loop of perception, ethical gating, and architectures of trust—to explain how intelligence evolves through feedback across moral and institutional domains.</p> Findings <p>The analysis reveals that sustainable alignment depends on reflexive feedback loops connecting perception, ethical reasoning, and institutional legitimacy. These loops enable adaptive and transparent governance of intelligent systems through shared moral and political accountability.</p> Originality/value <p>The paper proposes a constitutional model of co‑evolutionary governance for AGI that transforms alignment from a static objective into a living partnership between human judgment and machine cognition, emphasizing legitimacy, adaptability, and trust as foundations for responsible intelligence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Co-evolving intelligence: the long loop of perception, ethical gating, and architectures of trust on the path to human-aligned AGI

  • 期刊论文

  • 发表日期：

    2026年2月4日

查看全文
  • Kuo Ming Chu,
  • Hui Chun Chan

摘要

Purpose

This paper develops a conceptual framework for achieving human‑aligned Artificial General Intelligence (AGI) by treating alignment as a constitutional process that unfolds through the co‑evolution of cognition, ethics, and governance. Rather than viewing alignment as a purely technical constraint on model behavior, the study reconceptualizes it as a constitutional, trust-mediated process in which perception, ethical gating, and institutional architectures of trust co-evolve over time.

Design/methodology/approach

Integrating insights from cognitive science, engineering design, and political philosophy, the paper introduces three interrelated constructs—the long loop of perception, ethical gating, and architectures of trust—to explain how intelligence evolves through feedback across moral and institutional domains.

Findings

The analysis reveals that sustainable alignment depends on reflexive feedback loops connecting perception, ethical reasoning, and institutional legitimacy. These loops enable adaptive and transparent governance of intelligent systems through shared moral and political accountability.

Originality/value

The paper proposes a constitutional model of co‑evolutionary governance for AGI that transforms alignment from a static objective into a living partnership between human judgment and machine cognition, emphasizing legitimacy, adaptability, and trust as foundations for responsible intelligence.