Future structure of astrophysics in an LLM-proficient world

Determine how the study and practice of astrophysics would be organized in a future where large language models can perform data-science tasks with human-level proficiency, including implications for human roles, education, and the literature.

Background

Linking the rise of LLMs with the ongoing professionalization and data-scientification of astronomy, the author poses a direct question about the future configuration of astrophysics should LLMs reach parity with humans in data science.

The uncertainty is explicitly acknowledged, and the paper focuses instead on the foundational question of why we do astrophysics, leaving the structural question open.