A new study from MIT demonstrates a significant advancement in AI’s ability to interpret and reason about visual scenes, moving beyond simple object recognition. The research introduces a system that can answer complex, multi-step questions about images by understanding the relationships and interactions between objects within them. This represents a step toward more sophisticated visual intelligence, where AI can engage in contextual reasoning similar to human perception. The findings were published in the journal Science Advances. For the full details, read the complete article at https://technologyreview.com/2024/05/15/ai-visual-reasoning-breakthrough.