📄 中文摘要
神经科学与人工智能在过去几年中取得了显著进展，但两者之间的联系尚不紧密。基于2025年8月举办的一次研讨会，本文旨在识别并探讨这两个领域当前及未来的协同潜力。具体而言，文章深入分析了具身智能、语言与交流、机器人技术、人类与机器的学习机制以及神经形态工程等子领域。在具身智能方面，探讨了如何将生物智能的物理交互和环境感知能力融入AI系统，以期实现更高级别的智能行为，例如通过模仿生物运动控制机制来提升机器人操作的精细度和适应性。在语言与交流领域，研究了大脑如何处理和生成语言，并将其机制应用于改进自然语言处理模型，例如通过模拟人脑的上下文理解和语义推理能力来提升AI对话系统的流畅性和准确性。
📄 English Summary
NeuroAI and Beyond
Neuroscience and Artificial Intelligence (AI) have achieved significant progress in recent years, yet their interconnections remain relatively loose. Drawing insights from a workshop held in August 2025, this work aims to identify and explore the current and future synergistic potential between these two fields. Specifically, the paper delves into subareas such as embodiment, language and communication, robotics, learning in humans and machines, and neuromorphic engineering. In embodiment, the discussion centers on integrating the physical interaction and environmental perception capabilities of biological intelligence into AI systems to achieve higher levels of intelligent behavior, for instance, by mimicking biological motor control mechanisms to enhance robot manipulation precision and adaptability. For language and communication, the research investigates how the brain processes and generates language, applying these mechanisms to improve natural language processing models, such as enhancing the fluency and accuracy of AI conversational systems by simulating human brain's contextual understanding and semantic reasoning capabilities. In robotics, the paper explores deriving inspiration from neuroscience to design more autonomous and adaptive robots, encompassing biomimetic structural design and neural feedback-based learning algorithms. Regarding learning in humans and machines, a comparison is made between the efficiency and generality of biological learning, and how these principles, such as meta-learning and lifelong learning, can be applied to deep learning models to overcome current AI limitations in data dependency and generalization.