Session 12. 14.

This research examines whether LLMs demonstrate environmental awareness, with particular focus on how they reflect and potentially bias human attitudes toward environmental issues as a result of their training data and processes.

With the widespread adoption of AI technologies, particularly large language models (LLMs) such as ChatGPT, questions arise about their training processes and inherent biases. While these models have become integral to daily life, the opacity of their training data and processes—protected as corporate trade secrets—makes it difficult to clarify their pre-training biases and potential toxicity. Even after developers implement fine-tuning measures aimed at neutrality, initial biases may persist in these systems.

We developed a comprehensive evaluation framework to assess major LLMs from leading AI developers worldwide. The framework examined environmental behavior, knowledge, attitudes, social norms regarding environmental issues, and the ability to predict human behavioral shifts toward pro-environmental behaviors. To establish a comparative baseline, we conducted a cross-sectional survey using the same evaluation framework with a stratified random sample of 385 participants representative of the U.S. population.

The results show that while LLMs possess extensive knowledge, this does not always lead to positive predictions about human environmental behavior. Significant gaps exist in environmental awareness between human participants and LLMs, with these gaps varying based on AI developer, national origin, and training language. Our study highlights the ongoing challenges in developing AI systems that can accurately interpret and represent complex human and societal issues. These findings have important implications for developing more culturally and environmentally aware AI systems, suggesting the need for more sophisticated approaches to AI training and development in the context of environmental understanding and human behaviors.

Antarctica is a significant component of the global commons (Chan et al., 2019; Rockström et al., 2024). Information on ecological conditions is an essential component of commons governance (Berkes and Folke, 1998; Dietz et al., 2003; Ostrom, 1990). Information loss has been identified as a barrier to sustainability (Crabtree et al., 2022), so there is a need to better understand its dynamics. Acquiring field data from Antarctica is financially and logistically costly (Lynch et al., 2016), suggesting the need for institutional infrastructure (Janssen and Anderies, 2023). Additional challenges are a lack of environmental monitoring and increasing human footprint in the region (Summerson and Tin, 2018; Tin et al., 2014), and preferences for a low cost political system (Liggett et al., 2017). This interdisciplinary research explores the relationship between researchers' environmental attitudes and the perceived balance between scientific fieldwork activity and environmental protection across the greater McMurdo Dry Valleys region in Antarctica. Specifically, the influence of researchers' perception limitations (Crabtree et al., 2022), and sensitivity to group norms is explored. An agent-based model is used to simulate researchers conducting seasonal fieldwork across an empirically informed landscape. Decision-making for fieldwork planning is based on the theory of planned behaviour and attitudes are updated as field observations are made. I explore the impact of information flows, heterogeneity in researchers' environmental attitudes, and variations in the weight placed on group norms. Antarctica is a remote and contested environmental commons (Collis, 2017). While governance is undertaken for 'mankind' (Rabitz, 2023; Secretariat of the Antarctic Treaty, 1991), Antarctica is often "out of sight, out of mind” (Chan et al., 2019). This research contributes to understanding the influence of information flow on the environmental attitudes of the primary visitor to this commons and prompts reflection on the environmental cost of scientific research (Dupont et al., 2024).

Improving the resilience and adaptive capacity of farmers to disasters is a major challenge for rural areas around the world in the context of climate change. The ability of farmers to participate in collective action in response to disasters (CARD) is directly related to the adaptive capacity of individual farmers to climate change. Existing studies have discussed the impacts of disaster risk governance and information and communication technology (ICT) respectively, but few have introduced the factors of ICT application into the analyses related to disaster governance. Based on the collective action theory, this paper builds a link between the application of ICT and farmers' participation in CARD, and proposes a mechanism path through which the application of ICT can enhance CARD by enhancing social learning. Taking 987 farmers in the border areas of China as the study sample, the analysis results show that, on the one hand, ICT has a significant positive effect on CARD, but this positive effect will be weakened with the increase of the users of ICT; on the other hand, ICT will increase the human capital, financial capital and social capital of farmers by promoting the social learning pathway of the farmers, which will enhance the farmers' willingness to participate in CARD. This paper increases awareness of the new public goods represented by digital space, and provides practical pathways for rural villages in developing countries to cope with the impacts of climate change.