The paper studies how to make and judge scientific images that are not just pretty but scientifically correct.
ATLAS is a system that picks the best mix of AI models and helper tools for each question, instead of using just one model or a fixed tool plan.
Real people often ask vague questions with pictures, and today’s vision-language models (VLMs) struggle with them.