A new AI model demonstrates significant improvements in processing and understanding complex visual data, according to researchers. The system, developed by a team at a leading university, uses a novel architecture that allows it to analyze images with greater context and nuance than previous approaches. Early benchmarks show it outperforms existing models on several standardized tests for visual reasoning. The researchers emphasize that this is a step toward more general-purpose visual AI that can interpret scenes in a way closer to human perception. The full details of the model’s architecture and training methodology are available in the research paper published in a scientific journal. Read the full article at https://technologyreview.com/2024/05/15/ai-model-visual-reasoning-advance/.