The Alignment Mirror

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The Alignment Mirror proposes that humanity's effort to solve the AI alignment problem may have consequences beyond artificial intelligence. AI alignment makes a general problem unusually visible: a powerful system can successfully pursue the target it has been given while producing outcomes that conflict with the broader purposes humans intended it to serve. This paper asks whether the same way of thinking can be applied to older human-made systems, including companies, markets, governments, bureaucracies, financial institutions and digital platforms. Such systems also use rewards, measurements and rules to guide behavior. When measures such as profit, production, engagement or institutional growth become dominant targets, repeated success may diverge from wider human goals. This problem may become stronger when success brings additional resources and influence, allowing the same behavior to reproduce itself. Existing research already examines Goodhart-type failures, AI alignment and institutional alignment. The specific hypothesis proposed here is reflexive: as humanity develops concepts for identifying and correcting AI misalignment, those concepts may increasingly be turned back toward the institutions that created and govern AI. The paper calls this proposed process The Alignment Mirror.