Chinese Academy of Sciences academician Yao Qizhi: Embodied intelligence must move from imitation to reasoning.

At the opening ceremony of the GDPS 2025 Global Developers Pioneer Conference, academician Yao Qizhi of the Chinese Academy of Sciences stated that the breakthrough points for the future of embodiment intelligence are as follows: firstly, we need to move from imitation to reasoning, as there is currently a lack of explainable world models and physical causal reasoning; secondly, we need to move from data scarcity to data leaps, as the cost of collecting embodied data is very high and there is a need to establish a large-scale data factory to verify the Scaling Law of embodied intelligence; thirdly, we need to move from local skills to full-body collaboration, integrating full-body control and fine hand operation planning to support multi-step general skills; fourthly, we need to move from different languages to unified evaluation, establishing open benchmarks and safety standards, encouraging open source and reproducibility, so that excellent algorithms can be reused, verified, and industrialized.