Meta任命赵晟佳为超级智能实验室首席科学家；我国自主研发“磐石·科学基础大模型”发布丨AIGC日报

Group 1 - The future of AI will be characterized by multiple competing models, with a trend towards Multi-Agent systems where various models and tools collaborate [1] - The cost of training single models is becoming fixed, while inference costs are expected to decrease significantly in the next one to two years, potentially by an order of magnitude [1] - The demand for computational power will continue to grow due to an increase in token volume, despite tokens becoming cheaper [1] Group 2 - The "Panshi·Scientific Foundation Model," developed by a team from the Chinese Academy of Sciences, has been released, focusing on scientific tasks and capable of deep understanding of various scientific data modalities [2] - Meta has appointed Shengjia Zhao, a core developer of ChatGPT, as the Chief Scientist of its Super Intelligence Lab [3] - The "Qiyuan General Embodied Model" from Zhiyuan Robotics won the SAIL Star Award at the 2025 World Artificial Intelligence Conference for its outstanding performance across multiple dimensions [4] - Alibaba launched its first self-developed AI glasses, Quark AI glasses, which integrate various functionalities and capabilities within the Alibaba ecosystem [5]