The recent rise of transformer architectures (e.g. LLM) has intensified the debate surrounding the social implications of artificial intelligence, positioning ‘value alignment’ as a central response to the potential risks of advanced AI. However, the actual utility of this concept for such a critical task deserves careful scrutiny. This chapter approaches this analysis by explicitly framing ‘value alignment’ as a conceptual engineering project. Through a pragmatic approach, we employ a genealogical methodology to trace its origins, contextual development, and evolution. Our analysis reveals three distinct conceptions: ASI Value Alignment (VA1), serving a narrative function aimed at social coordination; AI Alignment (VA2), fulfilling an instrumental function focused on implementation; and Ethical Design (VA3), performing a normative function oriented toward the satisfaction of moral values.