Google DeepMind has developed a generalist AI agent called SIMA that can follow natural-language instructions to carry out tasks in various 3D virtual environments, including video games. SIMA was trained on nine different games and can perform tasks such as navigation, object interaction, and menu use. The goal of this research is to develop AI agents that can understand and safely carry out a wide range of tasks in a way that is helpful to people. SIMA's performance relies on language and it can generalize across different games.