SIMA
Scalable Instructable Multiworld Agent

It represents a significant stride in creating adaptable, instruction-following agents that can generalize their understanding and actions across different contexts without specific programming for each environment.

Developed by DeepMind, SIMA leverages a blend of recent advancements in natural language processing and reinforcement learning, enabling it to comprehend and act on instructions with a level of generality and adaptability previously unseen. This makes it a powerful tool for exploring the potential for AI to interact in complex, dynamic settings similar to the real world.

While the exact date of SIMA's introduction is not specified, it emerged from the ongoing effort in AI research to create more flexible and capable agents that can understand and navigate the world in ways that mimic human learning and adaptation.

Key contributors to SIMA's development are the teams at DeepMind, who have been at the forefront of research in artificial general intelligence, demonstrating their commitment to pushing the boundaries of what AI can achieve in understanding and interacting with complex environments.

https://deepmind.google/discover/blog/sima-generalist-ai-agent-for-3d-virtual-environments/