Google Unveils Genie 3: New ‘World Model’ to Train AI Robots in Realistic Virtual Environments

Google has introduced Genie 3, an advanced AI “world model,” marking a significant milestone towards artificial general intelligence (AGI). Developed by its DeepMind division, Genie 3 enables AI systems to operate and learn within sophisticated simulations that closely mirror real-world environments, like warehouses. The model accelerates the training of intelligent robots and autonomous vehicles by letting them interact with nearly authentic digital spaces. Although Genie 3 is not yet ready for public release, its capabilities were demonstrated to journalists, revealing detailed and adaptable scenarios created instantly from text prompts. Google’s move intensifies competition within the AI sector, as rivals like OpenAI also unveil new models. While hailed as an essential step towards AGI and more autonomous agents, experts note that Genie 3—and AGI at large—raises complex questions about job automation and the future role of AI in both virtual and physical realms.

Read more