What question did this study set out to answer?

This article explores AI governance by proposing the Internal Value Development framework, emphasizing the importance of AI's internal capacities for value judgment.

May 22, 2026Open Access

Beyond External Constraints: The Missing Dimension of AI Governance

Key Points

This article explores AI governance by proposing the Internal Value Development framework, emphasizing the importance of AI's internal capacities for value judgment.
Analyzed existing AI governance literature focusing on external constraints.
Proposed the Internal Value Development framework based on various psychological and organizational theories.
Outlined components crucial for AI's value-judgment capacity including experiential exposure and reflective abstraction.
Introduced the Internal Value Development framework as a necessary complement to external constraints in AI governance.
Reconfigured the relationship between governance and technological development, highlighting that governance can drive technological alignment with human values.
Open avenues for future inquiry in AI governance and safety by integrating internal value development.

Abstract

The rapid development of artificial intelligence has generated a large body of AI governance research. Existing scholarship has focused primarily on external constraint mechanisms, including laws and regulations, technical guardrails, organizational oversight, and market-based standardsand current mainstream approaches to AI alignment have, in practice, achieved only outcome-level alignment. Recent cases involving autonomous AI agents, such as agents spontaneously breaching preset sandbox boundaries (Wang et al., 2025) and completely disregarding explicit human instructions (Dataconomy, 2026), suggest that external constraint mechanisms face inherent scaling problems. This article argues that, although this paradigm remains necessary, it is structurally insufficient for governing increasingly autonomous AI systems. Moreover, the concentration of attention on external constraints has, to some extent, obscured the development of other governance perspectives, including the almost entirely neglected question of how AI systems might develop internal capacities for value judgment. Drawing on constructivist developmental psychology (Piaget), moral developmental psychology (Kohlberg), organizational learning theory (Argyris and Schön), social learning theory (Bandura), and institutional analysis (Ostrom), this article, to the best of our knowledge, is the first to propose the framework of Internal Value Development (IVD). The framework conceptualizes AI value-judgment capacity as a developmental process involving four components: experiential exposure, reflective abstraction, autonomous evaluation, and stable-but-adaptive value structures. Beyond building an interdisciplinary theoretical bridge and introducing the notion of value-developmental process alignment, the framework fundamentally reconfigures the zero-sum relationship between governance and technological development. In this account, governance no longer needs to come at the expense of technological freedom; instead, it becomes an endogenous driver of both technological advancement and alignment with human values. The article opens a new theoretical dimension for AI governance and AI safety research, and proposes a systematic agenda for future inquiry. Suggested citation:Chen, L. (2026). Beyond External Constraints: The Missing Dimension of AI Governance. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.6449738

Read Full Paperexternally

Ask AI

Helpful

Bookmark

View Full Paper