AI coding agents can modernize research software but can't judge if the science is right

A field report from OpenAI and academic partners indicates that coding agents can modernize neglected research software, achieving speedups of up to 60 times. The focus is shifting from writing code to the labor-intensive task of verifying scientific correctness. Participants noted that these systems can be "eloquent and convincing, yet confidently wrong in ways that are easy to overlook." This highlights the necessity for additional effort to ensure the accuracy of scientific results, emphasizing the need for a balance between speed and precision in scientific research.