AIR: Adaptive Interleaved Reasoning with Code in MLLMs
Teaching AI to switch between thinking and calculating when solving complex problems
Researchers trained AI systems that can see and understand images to seamlessly alternate between reasoning through a problem step-by-step and running code to do exact calculations. The trained models improved their accuracy by nearly 10 percentage points on math-heavy tasks and succeeded in using computational tools over 95% of the time.
Current AI systems struggle with problems that require both visual understanding and precise numerical work because they either guess at calculations or rely on hand-coded rules. This approach lets AI systems decide on their own when to stop reasoning and run code instead, which could unlock better performance on real-world tasks like engineering analysis, medical imaging with measurements, or financial analysis—where getting the numbers right matters as much as understanding what you're looking at.