Generative AI and the Fracturing of Global Science: Implications for Ibero‑America

Abstract

This article examines how generative AI (GenAI) is reshaping scholarly communication, with Ibero-America, Spain and Latin America, as a critical test case. Using a world-system perspective, we map global publishing trends, AI governance regimes (EU, U.S., China), and regional benchmarks in output, collaboration, and open access. Our analysis identifies both structural assets, multilingual capacity, community-owned Diamond OA platforms, and vulnerabilities such as APC-driven stratification, English-centric LLMs, and dependence on proprietary editorial tools. We develop three scenarios for 2025–2030: inclusive convergence, where mature Spanish/Portuguese public LLMs, well-funded Diamond OA, and interoperable repositories close visibility and collaboration gaps; parallel systems, where divergent AI ecosystems and regulatory regimes raise compliance costs and fragment workflows; and cost capture, where rising APCs and paywalled tools concentrate AI benefits among well-resourced actors, undermining bibliodiversity. Across all scenarios, we propose a “no-regret” strategy: pair GenAI investments with open infrastructures, multilingual datasets, and Diamond journals; standardize AI-use disclosure; and strengthen regional bargaining power through consortial funding, capacity-building, and equitable API access. These measures can align GenAI adoption with open science principles, language equity, and sustainable scholarly ecosystems in Ibero-America.