The Formative Lens: A Fourth Dimension for AI Governance Beyond Risk, Safety, and Alignment

Contemporary AI governance is organized around three established dimensions: risk, safety, and alignment. Each treats the human user as a fixed reference point and examines the system's behaviour against that point. This paper argues that the assumption of a fixed user is both empirically wrong and increasingly consequential. Sustained interaction with AI systems cultivates some human capacities and atrophies others, establishes habits of attention and judgment, restructures dependencies, and reshapes users' self-understanding. Existing governance frameworks—including the NIST AI Risk Management Framework, ISO/IEC 42001, and the EU AI Act—have no operational vocabulary for these formative effects and no instruments to assess them. The paper introduces the Formative Lens, a triadic diagnostic framework that adds formativity as a fourth governance dimension and pairs it with two companion dimensions: intentionality (what the system was built to do to the world) and relationality (what posture it requires of its user). Together, the three dimensions constitute a single diagnostic act addressed to a question existing frameworks are structurally unable to ask: not whether a deployment is permitted, but whether it is worthy of the humans who will be shaped by it. The framework is positioned not as a replacement for existing compliance regimes but as a diagnostic instrument that sits above them, intended for ethics committees, board-level oversight bodies, and independent reviewers. The paper develops the Lens through four operational components: a worked case study applying the triadic structure to AI writing assistants in knowledge work; a five-step assessment protocol; a five-minute diagnostic checklist for procurement and design conversations; and a heuristic rubric for committee deliberation. The framework is located within the philosophy-of-technology tradition running from Aristotle through McLuhan, Illich, and Stiegler, and the paper closes with implications for regulators, deploying organizations, and researchers, alongside an honest accounting of open methodological questions. The contribution is conceptual rather than empirical: the paper introduces formativity as a distinct governance category, proposes the triadic Lens as a first articulation of how the category can be operationalized, and invites collaborative development of empirical instrumentation, taxonomic refinement, and domain case studies.