Stanford and Meta inch towards AI that acts human with new ‘CHOIS’ interaction model

Researchers from Stanford University and Meta’s Facebook AI Research lab have developed a groundbreaking AI system, CHOIS (Controllable Human-Object Interaction Synthesis), that generates natural and synchronized motions between virtual humans and objects based solely on text descriptions. This system allows virtual beings to understand and respond to language commands as fluidly as humans, representing a significant advancement in AI. CHOIS utilizes a conditional diffusion model to simulate detailed sequences of motion, interpreting language instructions to generate realistic human-object interactions. The implications of this system are profound, particularly in the fields of computer graphics, animation, virtual reality, and AI and robotics, with potential impacts on reducing animation production time, improving immersive VR experiences, and enabling more autonomous and context-aware robots. The researchers’ work represents significant progress in creating advanced AI systems that can simulate continuous human behaviors in diverse 3D environments and has implications for the future development of more sophisticated AI systems.