尖峰智能实验室主张借鉴人脑这一自然界最精巧的智能载体，重点研发具有神经动力学特性的类脑大模型，构建一个既 具备强大感知力，又拥有深刻记忆与思考能力的全脑架构。 12月13日，在中国神经科学学会脑机接口与交互分会年会上，天桥脑科学研究院宣布成立尖峰智能实验室（Spiking Intelligence Lab, SIL），致力于类脑大模型和脉冲神经网络的研发，探索人工智能与人类智慧的深度融合。 第一财经记者从会议现场了解到，最新成立的尖峰智能实验室属于非营利研究机构，由中国科学院自动化研究所研究员 李国齐教授领衔，目标是为打造天桥脑科学研究院创始人陈天桥提出的"发现式智能"提供关键能力。 李国齐对第一财经记者表示："发现式智能的一个关键能力是神经动力学。不同于当前依赖规模法则堆叠参数的主流AI 模式，尖峰智能实验室主张借鉴人脑这一自然界最精巧的智能载体，重点研发具有神经动力学特性的类脑大模型，将脉 冲通信、时空动态编码等计算特性与树突神经元的精细结构深度耦合，构建一个既具备强大感知力，又拥有深刻记忆与 思考能力的全脑架构。" 这一路径也是陈天桥此前提出的实现通用人工智能的"结构路径"。人脑以仅约20瓦的功耗支撑起 ...

微软消费人工智能主管苏莱曼(Mustafa Suleyman)目前正致力于打造一款"符合人类利益"的超级智能。在本周最新的一场访谈栏目中他承 诺，若该技术对人类构成威胁，将立即停止相关研发工作。 苏莱曼在节目中表示："我们不会继续开发可能失控的系统。" 值得一提的是，在今年十月一份重塑微软与OpenAI关系的协议达成前，苏莱曼的工作其实一直受限于合同条款 ——该条款禁止微软研发通 用人工智能(通常指具备人类能力的系统)或超越人类能力的超级智能。 而根据新协议，OpenAI可与第三方联合开发部分产品。同时，微软也可独立或与第三方合作开发通用人工智能。 苏莱曼透露，微软实际上已放弃了限制这些权利以换取使用OpenAI最新产品的机会——这是双方此前合作的一部分，多年来微软一直为 OpenAI代建并配置数据中心。 谈及OpenAI时，苏莱曼表示，"他们如今与软银、甲骨文等多家企业达成了协议，建设的数据中心规模已超出微软原计划为其建造的数量。 因此相对应的，我们也获得了自主开发人工智能的权利。" 就目前而言，业内关于超级智能的讨论仍是理论性的。 虽然像ChatGPT这样的人工智能模型，能够以十年前计算机无法做到的方式与 ...

Core Insights - The release of the GPT-5.2 model by OpenAI shows a significant leap in cognitive abilities, particularly in abstract reasoning and generalization, as evidenced by its performance in the ARC-AGI-2 test, which increased from 17.6% to 52.9% [1] - The GDPval score, which measures the economic value of the model, rose dramatically from 38.8% to 70.9%, indicating a simultaneous breakthrough in both scalability and reasoning capabilities [1] Performance Comparison - In the SWE-Bench test, GPT-5.2 achieved a score of 55.6%, surpassing GPT-5.1's 50.8%, while Anthropic's Claude scored 52.0% and Google's Gemini scored 43.3% [2] - For the GPQA test, GPT-5.2 scored 92.4%, compared to GPT-5.1's 88.1%, with Claude at 87.0% and Gemini at 91.9% [2] - In the CharXiv reasoning test, GPT-5.2 scored 82.1%, significantly higher than GPT-5.1's 67.0%, while Gemini scored 81.4% [2] - The FrontierMath test results showed GPT-5.2 at 40.3%, GPT-5.1 at 31.0%, and Gemini at 37.6% [2] - In advanced mathematics, GPT-5.2 scored 14.6%, while Gemini scored 18.8% [2] Abstract Reasoning Metrics - The ARC-AGI 2 score for GPT-5.2 was 52.9%, a substantial increase from GPT-5.1's 17.6%, while Claude and Gemini scored 37.6% and 31.1% respectively [3] - The GDPval score for GPT-5.2 was reported at 70.9%, a significant rise from GPT-5.1's 38.8% [3]

Core Insights - The release of the GPT-5.2 model by OpenAI shows a significant leap in cognitive abilities, particularly in abstract reasoning and generalization, as indicated by its performance in the ARC-AGI-2 test, which increased from 17.6% to 52.9% [1] - The GDPval score, which measures the economic value of the model, rose dramatically from 38.8% to 70.9%, highlighting a breakthrough in both scaling and reasoning capabilities [1] Performance Metrics - In the SWE-Bench test, GPT-5.2 achieved a score of 55.6%, outperforming GPT-5.1 at 50.8% and other models like Claude and Gemini [2] - For GPQA, GPT-5.2 scored 92.4%, surpassing competitors such as Claude at 88.1% and Gemini at 91.9% [2] - In the CharXiv reasoning test, GPT-5.2 scored 82.1%, significantly higher than Claude's 67.0% [2] - In advanced mathematics, GPT-5.2 achieved a score of 40.3% in the FrontierMath test, compared to 31.0% for Claude and 37.6% for Gemini [2] - The ARC-AGI 1 test saw GPT-5.2 scoring 86.2%, while ARC-AGI 2 showed a notable increase to 52.9% from GPT-5.1's 17.6% [2] - The GDPval score of 70.9% for GPT-5.2 indicates a substantial improvement in knowledge work tasks compared to GPT-5's 38.8% [2]

Core Insights - Li Auto has launched its first AI glasses, Livis, marking it as the first automotive company to join the "Hundred Glasses War" [2] - The company aims to become a terminal enterprise in the era of general artificial intelligence, planning to develop more hardware products beyond AI glasses [2] - The competition in the electric vehicle market is intensifying, with companies like Tesla and Xpeng also venturing into robotics and AI technologies [4][10] Industry Trends - The market share of fuel vehicles in China has dropped below 50%, with electric vehicles and fuel vehicles now equally represented in new car sales [5] - Predictions indicate that the penetration rate of electric vehicles in China could exceed 85% in the next three years, with high-end electric vehicles expected to surpass 60% by 2026 [5] - The automotive market is experiencing a "zero-sum game," where the growth of electric vehicles comes at the expense of fuel vehicles, leading to a slowdown in overall sales growth [7] Competitive Landscape - New energy vehicle companies are increasingly resembling traditional automakers, focusing on cost control and profitability amid a competitive landscape [8][10] - Li Auto, Seres, and Leap Motor are among the few new energy vehicle companies that have achieved profitability, while others like Xpeng and NIO aim for quarterly profitability [7][10] - The narrative around "automobiles" is shifting, with companies like Li Auto and Xpeng adopting a more diversified approach similar to Tesla's model [8][11] R&D Investments - R&D investments in artificial intelligence and robotics are becoming crucial for new energy vehicle companies, with Li Auto planning to allocate a significant portion of its R&D budget to AI technologies [20][26] - Tesla's R&D expenditures have been steadily increasing, focusing on AI and robotics, while Xpeng and Li Auto are also ramping up their investments in these areas [21][23][26] - The competitive pressure in the market necessitates that new energy vehicle companies maintain a strong focus on their core automotive sales while gradually increasing their investments in AI and technology [26][27]

Core Viewpoint - The discussion emphasizes the importance of AI applications in various industries and the need for innovation and challenges to drive progress in technology and education [1][2][3]. Group 1: AI and Technology Applications - AI is seen as crucial for transforming industries such as agriculture and manufacturing, with examples including optimizing iron production and enhancing coal mining safety through data collection and modeling [3][4]. - The use of AI in healthcare is highlighted, with models improving diagnostic capabilities in hospitals and remote areas [4]. - The potential for AI to automate processes in ports and mining operations is discussed, showcasing advancements in efficiency and safety [3][4]. Group 2: Education and Talent Development - The shift from traditional education to online learning is noted, allowing students in remote areas to access quality education [5][6]. - The importance of nurturing talent within China is emphasized, with a focus on creating a strong educational foundation to support technological advancements [7][8]. - The company recognizes the need for diverse pathways in education, encouraging students to pursue various career paths, including skilled labor in technology [9]. Group 3: Future of AI and Workforce - The discussion includes the anticipated impact of AI on job markets, with a need for re-education programs to help workers transition to new roles as automation increases [15][16]. - The company acknowledges the potential for AI to enhance productivity across sectors, leading to overall economic growth [16][24]. - The importance of collaboration between academia and industry is highlighted to ensure that technological advancements align with market needs [8][30]. Group 4: Global Collaboration and Innovation - The company emphasizes the value of international collaboration in scientific research and technology development, recognizing contributions from various countries [20][30]. - The need for an open approach to learning from global advancements is stressed, with a focus on integrating diverse knowledge and practices [30]. - The potential for breakthroughs in fields like quantum computing and AI is acknowledged, with a call for continued investment in research and development [25][27].

憨态可掬的导盲机器狗 未来感十足的脑控轮椅 科技助残产品展示体验区时 吸引不少人驻足、互动 △科技助残产品展示体验区上的一台智能调脊机器人 这些"黑科技" 用于康复的智能调脊机器人 融合AI大模型与5G通信技术的 无障碍服务系统 这些代表着残疾人辅具个性化、智能化 发展方向的装备与服务 日前出现在全国残特奥会主媒体中心的 有不少其实已经运用在了 本届残特奥会运动员们比赛、训练 及生活的各个场景中 赛场上 六足导盲机器狗 引导盲人门球选手进场与退场 赛场外 以数字孪生技术打造的全时空指挥体系 让场馆信息 尤其是无障碍设施使用情况一目了然 科技力量的注入 润物细无声地为残疾人的出彩人生路 插上了"隐形的翅膀" 借助残特奥会这一窗口和平台 粤港澳大湾区作为我国科技创新 和辅具研发生产高地 用更精细化的管理和更先进的装备 让科技成为残疾人运动员 比赛与生活的好帮手 也让"有爱无碍"的口号具象化 残特奥会组委会在开赛前就提出 科技助残是本届残特奥会最大的亮点 无论身在何地 从微信小程序就能呼出 残特奥会专项手语翻译服务 2024年12月 中国残联、国家发展改革委、 不少参赛运动员和教练员表示 无障碍的参赛体验 已经贯穿 ...

AI给智驾的赋能触及瓶颈了？ "明星高管"苏菁再次"大胆开麦"引发圈内热议。 "大家最近都很嗨，一段式（端到端）搞定了，各种乱七八糟的概念都跑出来了，智驾是不是会进入一 个大发展时代？"近日，在2025地平线技术生态大会，地平线副总裁兼首席架构师苏箐表示："我 要'浇'点冷水。" 他认为，这一代深度学习技术可能碰到天花板了，所以很大概率在未来三年内，智驾的发展是在现有的 系统上做极致优化，而不是理论内核重构，所以"大家别太嗨，又进入苦日子的阶段了"。 公开资料显示，苏箐曾担任华为车BU智能驾驶产品部部长，负责华为自动驾驶系统方案。在华为车BU 时，苏菁就因为"敢说"在智驾圈广为人知，曾发表"基础是计算机，车是计算机控制的外设"的观点，并 锐评特斯拉智驾事故"讲难听点就是'杀人'"。 在此次地平线技术生态大会上，苏菁回顾了过去几年其在智驾行业的历程，这也映射了整个智驾行业近 年来的发展。2022年他准备加入地平线时，心态是"不想再做自动驾驶了"，原因一是太痛苦，二是看不 到希望，"那个时代的自动驾驶看似不错，但与真正人类司机相比差距极大。" 当时的智驾技术范式主要是规则驱动。分水岭出现在2024年初特斯拉FSD ...

Core Insights - The conference "Horizon Together 2025" focuses on the transition of smart driving from commercialization to widespread adoption, emphasizing collaboration among industry leaders [1][3] - Dr. Yu Kai, CEO of Horizon, highlighted the company's journey from "high aspirations" to "collaborative growth," aiming to democratize advanced technology for a broader audience [3][5] Group 1: Technological Advancements - Horizon has introduced three technological pillars: BPU®, compilers, and algorithms, which are essential for driving advancements in smart driving and robotics [3][10] - The newly released fourth-generation BPU architecture, named Riemann, boasts a tenfold increase in key operator computing power and supports full floating-point calculations, enhancing efficiency for large language models [14][16] - The compiler technology has evolved into an "AI-driven compilation" era, significantly improving compilation speed and model performance [14][16] Group 2: Market Strategy and Collaboration - Horizon is transitioning from a traditional model to an "HSD Together" algorithm service model, allowing partners to focus on their strengths while Horizon provides comprehensive algorithm services [20][21] - The company aims to reduce the time and costs associated with product development by 90% through this collaborative approach, enabling more companies to leverage advanced smart driving capabilities [20][21] - Horizon's products have achieved significant market penetration, with the latest series reaching over one million units shipped in just 12 months [20][29] Group 3: Industry Impact and Vision - Horizon positions itself as a foundational player in the robotics era, aspiring to be the "Wintel of the robotics age," focusing on ecosystem collaboration rather than competing as a vehicle or robot manufacturer [9][24] - The company has expanded its influence beyond automotive applications, becoming a leading platform for consumer robotics with over 100 products and extensive partnerships [24][25] - Horizon emphasizes the importance of making advanced technology accessible to the masses, aiming to transform high-end innovations into everyday solutions for consumers [28][29]

在人工智能应用竞争激烈的2025年末，一匹"黑马"正以加速度闯入更多人视野。 12月10日，阿里巴巴官方信息显示，旗下AI应用"千问"自11月17日全面公测仅23天，月活跃用户数（涵盖 APP、Web、PC全端）已破3000万大关，一举成为全球增长最快的AI应用之一。 有业内观察人士指出，这展示出市场对AI应用的旺盛需求，也体现出AI应用的竞争焦点正从炫技式的"会聊 天"转向务实的"能办事"。 不只是流量狂欢 千问的快速起势是多种因素共同作用的结果，其背后既体现了技术积累的价值，也反映出战略定位在市场竞争 中的关键作用。 首先，底层模型的技术支撑即基于阿里千问（Qwen）大模型的长期积累，为快速发展奠定基础。实际上，自 2023年起，阿里便采取了积极的开源策略，其Qwen系列模型在Hugging Face等全球开发者社区获得了广泛认 可。 截至目前，阿里千问Qwen已开源300多款模型，囊括文本、编程、图像、语音、视频等全模态，覆盖0.5B到 480B等全尺寸，在全球主要模型社区的下载量已经突破6亿，衍生模型突破17万个。 其次，明确的功能场景聚焦加速了其用户增长。千问发布伊始便定位"会聊天、能办事"这一方向 ...