How scientists’ narratives on AI signal a new era for science education

Journal article

Artificial intelligence (AI) has become a central topic in scientific and public discourse, raising important questions about its impact on science, society, and education. The paper reports an empirical study on 151 scientist commentaries published from 2021-2024 in the top scientific outlets of Nature and Science to examine evolving trends in AI-related communication in science. Using epistemic network analysis, we explored how scientists’ narratives frame the nature of science (NOS), a central area of science education research since at least the 1960s. The current impact of AI on science needs to be understood in order to ensure that contemporary depictions of NOS in science education are consistent with the fast-changing landscape of scientific research. Findings show that expert commentaries address multiple dimensions of NOS, with scientific practices as the central concept. Surrounding these practices, scientific and social values emerge as key considerations for utilising AI as a tool for scientific research and social contributions. While epistemic aspects of NOS remain a focus, the findings suggest that the launch of ChatGPT in late 2022 shifted attention toward socialinstitutional dimensions, emphasising more the social and political structures that underpin NOS. These trends highlight the evolving relationship between epistemic and societal aspects of AI, reflecting broader debates on ethics, governance, and applications. Expert commentaries serve as valuable resources for aligning science education with contemporary accounts of scientific research, and as such they signal a new era for science education in the age of AI where the social and institutional aspects of science play a major role. The implications for science education are discussed highlighting how socially embedded NOS can potentially prepare future scientists and citizens to critically engage with AI and other emerging technologies.

Authors: Chan, H-Y; Erduran, S

• Publication status: Accepted
• Peer review status: Peer reviewed
• Publisher: Springer
• Journal: Research in Science Education
• Acceptance date: 2025-10-28
• EISSN: 1573-1898
• ISSN: 0157-244X
• Language: English
• Keywords: science communication; artificial intelligence; Nature of Science; Family Resemblance Approach; epistemic network analysis