The silicon dilemma: a scientific mapping of the relationships between AI and SDG 13

Abstract

This study analyzes the paradoxical relationship between Artificial Intelligence (AI) and Sustainable Development Goal 13 (SDG 13 – Climate Action), a phenomenon referred to as the “Silicon Dilemma.” The main objective was to conduct a scientific mapping of publications addressing the intersection between these two themes, classifying their perspectives (positive, negative, or neutral) and identifying trends across different fields of knowledge. The methodology adopted was qualitative-quantitative in nature, with an exploratory and descriptive approach, employing bibliometric analysis combined with text mining techniques. Data were collected from the Web of Science database, covering the period from September 2015 to December 2024, resulting in a final corpus of 331 articles. For classification purposes, a Python-based algorithm using lexical dictionaries was applied. The results show a predominance of positively oriented studies, which present AI as an enabling tool for sustainability, particularly in areas such as climate forecasting, energy optimization, and precision agriculture. However, a significant number of studies also exhibit a negative perspective, especially within Computer Science, highlighting environmental costs such as high energy consumption and the carbon footprint associated with training large-scale models. It is concluded that, although AI offers unprecedented capabilities for mitigating climate change, its material infrastructure poses significant environmental challenges that have not yet reached a clearly defined quantitative balance in the literature.