Akademik Veri Yönetim Sistemi
Science of the Total Environment, cilt.1012, 2026 (SCI-Expanded, Scopus)
This study examines the paradoxical relationship between artificial intelligence (AI) adoption in fisheries and CO2 emissions across twelve heterogeneous countries from 1995 to 2024. Employing Cross-sectionally Augmented Autoregressive Distributed Lag (CS-ARDL) methodology and wavelet coherence analysis, we investigate whether AI technologies facilitate environmental sustainability in marine resource management. AI adoption is measured through the number of AI-related scientific publications (OECD.AI database), serving as a proxy for research intensity and technological development in the sector, though this metric captures knowledge production rather than direct operational deployment. Our findings reveal that AI adoption (proxied by the total number of AI-related scientific publications, hereafter ‘AI metric’) is associated with increased rather than decreased emissions in both short-run (0.110) and long-run (0.172) periods, challenging conventional assumptions about technology-driven environmental improvements. Granger causality tests indicate unidirectional causality from CO2 emissions to AI development, suggesting that environmental concerns drive technological innovation rather than AI proactively addressing sustainability challenges. Wavelet coherence analysis demonstrates strong synchronization between AI publications and emissions in the 64–128 period band during 2005–2015, with AI publications leading emissions, reflecting contemporaneous positive correlation consistent with Granger causality evidence that emissions drive AI development. These results highlight the need for explicit environmental objectives in AI development for fisheries and integration of carbon pricing mechanisms in AI-driven decision-making algorithms to realize the potential environmental benefits of technological innovation in marine resource management.