Moving Research Meetings from Data Updates to Cognitive Windows in the Age of Artificial Intelligence

Generative artificial intelligence (AI) is rapidly reshaping how scientific work is conducted, enabling trainees to produce summaries, analyses, and interpretations with unprecedented speed and often plausible outputs. While these tools can enhance productivity, they may also obscure the reasoning processes that underlie scientific judgment. Research meetings have long been valuable spaces for discussing progress, troubleshooting experiments, and developing ideas, but in the era of generative AI they may serve an even more important role as critical touchpoints for assessing understanding. We propose that research meetings serve as critical "cognitive windows" into trainee understanding. Structured discussion can reveal three complementary dimensions of scientific reasoning: metacognitive calibration, conceptual explanation, and collaborative sensemaking. A case study illustrates how AI-generated analyses, combined with routine meeting dynamics, can obscure these dimensions when a trainee's reasoning is not made explicit. Intentionally structuring research meetings to surface these processes can help mentors more effectively assess trainee understanding and support the development of independent scientific judgment.