MIT Researchers Unveil Gen, a Flexible AI Programming System for All Skill Levels
MIT introduces Gen, a new probabilistic-programming system that simplifies artificial intelligence for beginners while offering advanced capabilities for experts. The versatile platform handles tasks ranging from 3D body pose inference to automated data analytics.
MIT researchers introduce a new probabilistic-programming system called Gen that allows users to build AI models for computer vision, robotics, and statistics without dealing with complex equations or manual coding. The system handles various AI techniques behind the scenes, making it highly accessible for users with different levels of programming experience.
In a demonstration of Gen's capabilities, the researchers show that a short program successfully infers 3-D body poses by combining graphics rendering, deep learning, and probability simulations. This specific approach achieves better accuracy and speed on this difficult computer-vision task compared to earlier systems developed by the team.
Gen also simplifies data analytics by automatically generating sophisticated statistical models that experts typically use to find and predict patterns in data. The researchers emphasize that their goal is to make automated AI accessible to people with less computer science or math background while simultaneously increasing productivity for experts who need to rapidly prototype AI systems.