| 其他摘要 | Technological progress serves as a crucial driving force behind societal transformations, with the resultant changes in daily life and the environment closely interconnected with shifts in human psychology. Since late 2022, the advent of generative AI models, most notably ChatGPT, has been hailed as a groundbreaking advancement, exerting profound global implications.
ChatGPT is a large-scale language model designed and optimized primarily for conversation, engaging in interactions through dialogues. Through pre-training, ChatGPT can provide precise responses to questions by following statistical patterns. With its remarkable "intelligence," ChatGPT has subverted our existing notions of creative thinking and reasoning, self-awareness and emotional understanding, education and learning, among other aspects, and is unexpectedly influencing human cognition, values, beliefs, and behaviors. As a large-scale natural language processing model, the design mechanism of ChatGPT makes it prone to generating human-like responses and exhibiting moral and ethical stances that reflect specific ethical and moral norms. Therefore, the "anthropomorphized" characteristics displayed by ChatGPT and their potential influence on the future of humanity deserve attention.
Pro-environmental behavior represents a specific manifestation and extension of altruistic conduct within the environmental domain, encompassing considerations for humanity as a whole, as well as animals and the entire ecosystem. When confronted with relevant issues, do ChatGPT's opinions and suggestions reflect an environmentally friendly inclination? How does it influence human pro-environmental psychology and behaviors? To delve into these questions, this study employs theoretical frameworks such as "Values-Beliefs-Norms", "ABC+" and Theory of Planned Behavior, utilizes experimental and survey methodologies to comprehensively investigate ChatGPT's environmental values, ecological worldview, climate change cognition, and beliefs, exploring whether and how they impact users' pro-environmental intentions and behaviors.
The study consists of three phases.
In Study 1, a method of human-computer interaction was used to analyze ChatGPT's responses to environmental questionnaires, examining its tendencies and positions in terms of environmental values, ecological worldviews, climate change cognition, and beliefs. This phase aimed to create a self-portrait of ChatGPT.
In Study 2, a questionnaire survey was conducted with 300 recruited users to subjectively evaluate their perceptions of ChatGPT's environmental values, ecological worldviews, climate change cognition, and beliefs. A user portrait of ChatGPT was obtained and compared with the self-portrait from Study 1 to conduct comparative analysis.
Building upon Study 2, Study 3 tracked 275 users who provided valid responses in the second questionnaire. Through further surveys, the study analyzed and verified the impact of ChatGPT's environmental values, ecological worldviews, climate change cognition, and beliefs on users' pro-environmental behaviors.
The research employed a mixed-methods approach, combining interactive Q&A to explore ChatGPT's environmental stance, questionnaire measurements to assess user perceptions, and validation of the effects. This not only depicted ChatGPT's own proenvironmental portrait but also examined how ChatGPT further influences individuals' environmental attitudes and behaviors by examining user feedback.
Study 1 revealed that ChatGPT demonstrated significant pro-environmental tendencies in terms of egoistic value orientation, altruistic value orientation, biospheric value orientation, ecological worldviews, and cognitive beliefs. Comparing Study 2 with Study 1, while there were some differences between individuals' estimations and ChatGPT's self-assessment in terms of pro-environmental tendencies (overestimation of egoistic and biospheric value orientations, underestimation of altruistic value orientation), overall, individuals still perceived ChatGPT as having a pronounced proenvironmental inclination. Results from Study 3 indicated a significant positive correlation between participants' evaluation of ChatGPT's pro-environmental attributes and their expected pro-environmental behaviors. Upon learning about the actual ratings of ChatGPT on climate change and other aspects, participants' assessments of climate change beliefs, climate change attribution, and the impacts of climate change on a global and national level all significantly increased. Regarding pro-environmental behaviors, participants anticipated a significant improvement compared to their current state. Overall, users were influenced to a considerable extent by ChatGPT, thereby displaying a greater willingness to make pro-environmental choices, engage in environmental practices, and protect the environment in the future.
The findings of this study provide theoretical support for the application of nextgeneration artificial intelligence in environmental education and advocacy, contributing new evidence to research related to human-computer interaction and environmental psychology. Moreover, the practical implications offer insights into the harmonious development of humans and artificial intelligence, promoting environmental protection and sustainable development. These results also serve as a reference for future utilization and management of next-generation artificial intelligence. |
修改评论