Beyond replacement anxiety: a psychological framework for understanding AI in natural science research

Artificial intelligence (AI) is moving from a peripheral aid to a structuring force in natural science research. Large language models, multimodal systems, autonomous laboratories, and agentic tools now assist with literature analysis, coding, image interpretation, experimental optimization, manuscript preparation, and elements of peer review. Public discussion has largely emphasized acceleration, automation, and discovery potential. Yet the psychological implications of this transition for researchers remain underdeveloped. This perspective argues that AI in natural science should be understood not only as a technical shift, but also as a reorganization of the psychological environment of scientific work. Drawing on recent work on autonomous experimentation, AI-assisted writing and review, technostress, AI workplace anxiety, attitudes toward AI at work, and human-AI delegation, we propose a four-part framework organized around labor visibility, identity stability, accountability under delegated cognition, and institutional climate. We treat this framework as a theoretically grounded and partly inferential lens whose expression is expected to vary across disciplines, tasks, career stages, and institutional settings. We further argue that responsible AI adoption in science should be judged not only by speed or scale, but also by whether research systems preserve human agency, interpretive responsibility, developmental learning, and sustainable scientific careers. A psychological perspective does not oppose scientific modernization; rather, it clarifies the human conditions under which AI-enabled science can remain rigorous, trustworthy, and professionally livable.