Beyond the Data Paradigm? Synthetic Data, Simulation, Autonomous Systems, and the Reorganisation of Digital-Economic Value

Abstract Debates about artificial intelligence increasingly focus on synthetic data, simulation, model reasoning, and autonomous agents rather than on data collection alone. This paper examines whether these developments justify describing the emerging digital economy as “post-data.” It argues that the term is useful only as a qualified analytical proposition: data is not disappearing, but its strategic value is being recombined with compute, model architecture, simulation, evaluation, and institutional trust. The paper uses a structured integrative literature review and comparative case analysis to examine three domains: compute-efficient language-model training and synthetic data; biomolecular modelling through AlphaFold 3; and synthetic data and simulation in healthcare and operational decision-making. The analysis identifies a transition from data accumulation as a standalone competitive advantage toward a complementary capability stack in which provenance, validation, domain constraints, and human accountability remain essential. The evidence does not support a universal replacement of empirical data by synthetic environments. Instead, it shows that synthetic systems can extend scarce or expensive observations while also reproducing source-data biases, creating privacy risks, and degrading when generated outputs are recursively reused without adequate grounding. The paper contributes a conceptual framework linking data provenance, synthetic generation, compute and reasoning, validation, andgovernance. It concludes with research propositions and a practical agenda for evaluating when synthetic and autonomous systems create durable value. Keywords: synthetic data; autonomous systems; simulation; digital economy; AI governance; model collapse; data provenance