专题：2025科技风云榜 新浪科技讯 1月15日下午消息， "2025科技风云榜"年度盛典于2026年1月15日在北京举办，今年活动主 题为"启新智，赴新程"。 欧洲科学院院士、西湖大学可信与通用人工智能学院创始人金耀初发表《探路AGI：走向类脑具身智 能》主题演讲，他表示，具身智能的安全变得更为重要，大模型已经有幻觉，有很多安全隐患。如果把 新浪科技讯 1月15日下午消息， "2025科技风云榜"年度盛典于2026年1月15日在北京举办，今年活动主 题为"启新智，赴新程"。 欧洲科学院院士、西湖大学可信与通用人工智能学院创始人金耀初发表《探路AGI：走向类脑具身智 能》主题演讲，他表示，具身智能的安全变得更为重要，大模型已经有幻觉，有很多安全隐患。如果把 它放到具身系统里面，安全变得更重要。感知、推理以及和不同的机器人与人之间交互的安全，具身安 全更为重要。 金耀初指出，回到类脑具身智能，如何把脉冲这样的信息处理方式，不同信息处理的路径、机制等等都 放到具身智能系统上去，这是目前正在做的一些工作。如何适应不同任务，如何很灵活改变它的功能性 结构等等。这是具身导航方面的工作，把人的海马体里面最基本的机制，拿来 ...

Group 1 - Oracle denies reports of a delay in the completion of the data center for OpenAI, stating that all contractual milestones are on schedule and agreed upon by both parties [2] - Google introduces Gemini's translation feature into its text translation service and launches a beta version for real-time voice-to-voice translation through headphones [2] - Yuzhu Technology launches a humanoid robot App Store, allowing users to download action presets for complex operations, currently featuring presets like funny actions and dance moves [2] - Tianqiao Brain Science Research Institute establishes the Spiking Intelligence Lab, focusing on brain-like models and spiking neural network research, led by Professor Li Guoqi from the Chinese Academy of Sciences [2]

Core Insights - The article discusses the limitations of Spiking Neural Networks (SNNs) and introduces a new architecture called Max-Former that addresses these limitations by enhancing high-frequency information processing [5][24]. Group 1: Performance Limitations of SNNs - SNNs have been traditionally viewed as inferior to Artificial Neural Networks (ANNs) due to their binary pulse transmission, which was believed to cause significant information loss [5][6]. - The research indicates that the real issue lies in the frequency bias of SNNs, where spiking neurons act as low-pass filters, suppressing high-frequency components and favoring low-frequency information [4][8][19]. - This frequency imbalance leads to a degradation in the feature representation capabilities of SNNs, limiting their performance [10][23]. Group 2: Introduction of Max-Former - The Max-Former architecture is designed to counteract the inherent low-frequency preference of SNNs by incorporating two lightweight "frequency-enhancing lenses" [24][28]. - The architecture includes an additional Max-Pool operation in the Patch Embedding stage to actively inject high-frequency signals at the input source [28]. - It also replaces early-stage self-attention with deep convolution (DWC), which retains local high-frequency details while being computationally efficient [28]. Group 3: Performance Metrics and Results - Max-Former achieved a Top-1 accuracy of 82.39% on ImageNet with fewer parameters compared to Spikformer, demonstrating a significant performance improvement [27]. - The architecture also reduced energy consumption by over 30% while achieving performance breakthroughs [30]. - The findings suggest that optimizing SNNs with high-pass operators can lead to improvements in both performance and energy efficiency [31]. Group 4: Broader Implications - The insights gained from the Max-Former architecture are applicable beyond Transformer models, as demonstrated by the Max-ResNet architecture, which also benefited from the addition of high-frequency operations [33]. - The research provides a new perspective on the performance bottlenecks of SNNs, suggesting that their optimization should not merely mimic successful designs from ANNs [35].

Group 1: Gold Market Insights - Spot gold prices surpassed $3,300 per ounce for the first time since May 9, driven by rising geopolitical tensions and negative GDP growth in the U.S., which increased safe-haven demand [1] - Domestic gold consumption remains strong, with retail sales of gold and silver jewelry in April up 25.3% year-on-year and 14.7% month-on-month, indicating that domestic demand is independent of international gold price fluctuations [1] - There is a divergence in institutional views on gold; bullish arguments include inflation risks and a potential Fed rate cut, while cautious signals highlight the current high price levels and the possibility of profit-taking due to eased trade tensions [1] Group 2: Solar Industry Impact from Tariffs - The U.S. plans to impose extreme tariffs on Southeast Asian solar equipment, with Cambodia facing a 3,521% tariff due to non-cooperation in investigations, while Malaysia faces only 34% [2] - The U.S. heavily relies on Southeast Asia for solar imports, with 80% of imports coming from four countries, leading to a potential shift in procurement to domestic or third-party manufacturers [2] - U.S. solar project developers are facing increased costs due to these tariffs, which may delay installation progress and create cash flow pressures for EPC companies [2] Group 3: Humanoid Robots Development - The commercialization of humanoid robots depends on their ability to create actual value by addressing real-life challenges, with a long-term development cycle similar to that of autonomous driving, estimated at 10-20 years [3] - The industry is entering an accelerated phase due to supportive policies and the presence of a significant talent pool in the field of embodied intelligence, with a focus on practical applications [3] - Early application scenarios have been validated in sectors like power and chemical inspections, indicating a potential for successful technology-commercialization loops [3] Group 4: AI Agent Development - The AI agent market is rapidly evolving, with diverse technical paths and a focus on expanding application scenarios, although a unified standard has yet to be established [4] - There are significant differences between the North American and Chinese markets, with both targeting enterprise-level markets as a core breakthrough point [4] - Current challenges include high token consumption during interactions and the need for robust computational infrastructure, which remains a key limiting factor for commercial scalability [4] Group 5: Public Fund Regulation Changes - New regulations for public funds are driving a shift in strategy, with a focus on core asset pricing and a potential systemic adjustment in strategy paradigms [5] - The easing of U.S.-China tariffs has improved market risk appetite, with a focus on opportunities in the export chain [5] - Social financing growth is supported by low base effects and monetary policy, although potential impacts from tariff shocks should be monitored [5]