Translated data: Google DeepMind has quietly released Mirasol3B, a multimodal autoregressive model that has made significant strides in audio, video, and text processing. This technology could potentially bring revolutionary changes to areas like YouTube, enhancing user experience and accessibility. However, experts have raised questions about the practical feasibility of its application, emphasizing the need to ensure compliance with social ethics and environmental standards.