Year: 2026
Type: article
Abstract: Geographic Information Systems (GIS) have long been positioned along Wright, Goodchild and Proctor's influential continuum of GIS as a tool, GIS as toolmaking, and GIScience, an arrangement that presumes GIS is epistemically passive and primarily executes human authored intent. This paper argues that the integration of generative AI into GIS introduces functional autonomy in goal interpretation, workflow and code generation, execution monitoring, and iterative revision, which increasingly breaks this boundary condition. Through an integrative narrative review of AI-enabled, agentic, and autonomous GIS, we trace a migration of intent from expert driven command execution toward partial delegation of planning and procedural assembly to systems operating in iterative loops. Building on this analysis, we propose the Agentic Role Entanglement (ARE) model, which replaces the tool-toolmaking-science continuum with an epistemic action loop in which tool use, toolmaking, and epistemic inquiry become dynamically coupled within a single episode. ARE model yields falsifiable implications for evaluation and peer review, motivating a shift from output checking to episode level validation. To support this shift, we outline a layered validation stack centered on provenance, reproducibility, workflow and code verification, geospatial sanity checks, and uncertainty bounded claims. We extend these requirements into a lifecycle view of distributed responsibility and ethical governance. We conclude by advocating a post instrumental GIScience that treats human-agent systems, rather than software alone, as the unit of analysis for scientific defensibility, reproducibility, and accountability.
Authors: Ali Khosravi Kazazi, Zhenlong Li
Institution: Pennsylvania State University
Language: English
Subfield: Information Systems and Management
Field: Decision Sciences
Domain: Social Sciences
Open Access status: gold
Estimated APC paid: $2,390
OpenAlex ID: W7167359874