Toward a 21st Century Turing Test: Games, Authority, and Interpretive Intelligence in AI

We propose a 21st century Turing Test, one concerned not with whether machines can imitate humans, but whether they can exercise genuine interpretive intelligence: the capacity to read context, navigate ambiguity, and exercise moral judgement. Turing's original test was itself a game—a structured environment in which judgement becomes visible through play—and we follow this insight, introducing the Game within a Game: an experimental framework in which an AI games master oversees an ensemble of players, with varying degrees of authority to set rules, intervene in outcomes, and evaluate performance. The games master's decisions reveal its implicit values and judgment across four distinct governance regimes, from pure evaluation to full environmental control. We instantiate this framework in The Machine, a collaborative game in which players construct an entity of unknown purpose, delegating tasks and accumulating status under epistemic asymmetry controlled by the games master. Drawing on humanistic traditions of meaning-making, narrative theory, and the philosophy of judgement, we argue that AI-human games constitute a genuinely new benchmarking paradigm, and that AI systems capable of exercising interpretive intelligence could become genuine partners in human flourishing, helping us navigate the complex social worlds we increasingly share with them.