The evolution of information ecosystems: a macro-exploratory analysis of digital intelligence empowerment

Generative AI and agent-based systems are reshaping how information is produced, circulated, and validated, making synthetic content and model hallucinations increasingly routine features of contemporary information environments. While Web 2.0 perspectives explain connectivity, participation, and diffusion well, they are less able to account for ecosystems in which model-mediated generation transforms informational objects and shifts value toward verification and provenance. The article’s central claim is that information ecosystems are undergoing a macro-level transition from networked connectivity to generative symbiosis, a condition in which human actors and generative systems recursively co-produce informational outputs, and in which verification, rather than attention, becomes the ecosystem’s central organizing constraint. To theorize this transition, the article conceptualizes digital intelligence as the coupling of ubiquitous digital sensing and generative agency, and links information ecology with algorithmic epistemology. Empirically, the study adopts an exploratory qualitative and abductive design based on two analytically complementary Chinese cases spanning public governance and market production: Hangzhou’s City Brain 3.0 and Haier’s COSMOPlat/HIMOS, which jointly capture top-down administrative and bottom-up market logics and thus support a macro-exploratory analysis of systemic evolution. Drawing on triangulated technical, policy, and public-discourse materials, and with explicit reflexive attention to promotional bias in self-disclosed sources, the analysis traces how this transition operates through two coupling mechanisms. Semantic alignment translates unstructured inputs into machine-actionable representations, in a statistical rather than intentional sense, and agent-based coordination delegates matching, scheduling, and negotiation tasks to bounded autonomous agents. The same mechanisms generate three interrelated paradoxes concerning power, materiality, and verification, read here as dialectical contradictions rooted in the dual constitution of digital intelligence as technological morphology and intelligent function. As a theory-building, macro-exploratory study grounded in two cases and largely public materials, the article advances analytical rather than causal claims. It concludes by proposing a reverse-reconstruction governance approach that shifts attention from downstream content harms alone to the upstream epistemic and institutional conditions under which outputs become intelligible, credible, and actionable, and operationalizes this approach through a baseline epistemic audit covering provenance integrity, uncertainty disclosure, human-fallback rules, model-change traceability, and drift and flattening monitoring, alongside green-computing constraints and human oversight along the information-production chain. Overall, the article offers a framework for explaining how digital intelligence is reconfiguring information ecosystems and for informing more accountable, transparent, and adaptive governance responses to AI-mediated information environments.