Hoda Heidari and Aydin Mohseni published a framework for science in an AI era. It argues science must preserve durable human understanding and direction. The framework sets a two-part test for delegating research tasks to AI. The first part requires the scientific community to explain and contest claims. The second requires humans to judge which questions matter and what findings mean. It applies this test to every phase of research in the Contributor Roles Taxonomy. The framework differs from existing guidance by focusing on science’s purpose rather than integrity norms. It holds even if AI outperforms humans. It closes with reforms to keep science aligned with human understanding and direction.