Environmental AI Research Priorities: What They Reveal About Optimism and Misalignment

Elli Danae Vartziotis, Tina Vartziotis, George Dasoulas, Ippolyti Dellatolas, Francesca Dominici, Sotirios Kotsopoulos, Aristotle Tympas

Track: Main Papers Track (6 to 9 pages)

Keywords: AI for Climate Change, Environmental AI, Research Priorities, Alignment with Environmental Needs, Technological Optimism, Meta-Research

Abstract: While artificial intelligence (AI) is increasingly integrated into environmental research, a comprehensive evaluation of this integration remains limited. In response to this, we analyzed a focused sample of 106 publications from Nature and Science (2017–2024) to characterize how AI's role in addressing the environmental crisis is represented within these journals. Our mapping reveals a substantial imbalance across all publication types in our dataset, with 73.6\% of the publications focusing on Forecasting, 19.8\% on Monitoring \& Assessment, and only 6.6\% on Mitigation. Notably, 81.1\% reference prior non-AI approaches, indicating that AI is often used for already-addressed environmental challenges. Most studies rely on standard machine learning techniques and remain at early development stages. Optimism about AI’s potential has increased over time; however, high novelty AI uses remain exploratory and rarely operational. These findings highlight trends in how AI is portrayed, deployed, and aligned with environmental priorities, and the importance of reflecting on their implications.

Submission Number: 20