Science Popularization Writing in the Age of Generative AI: Efficiency, Authorship, Trust, and Fidelity

Generative artificial intelligence is rapidly entering science popularization writing. It can support topic development, audience adaptation, language simplification, and multilingual communication, but it also creates risks involving factual hallucination, excessive simplification, stylistic homogenization, unclear authorship, and weakened public trust. This commentary argues that generative artificial intelligence (AI) should be understood not as an autonomous science communicator but as a conditional writing infrastructure requiring human judgment. It proposes a human-led, AI-assisted workflow centered on source grounding, expert verification, uncertainty preservation, transparent disclosure, and editorial accountability.