" 更低成本的真实数据需求，更快迭代智驾系统。 " 作者丨田哲 编辑丨林觉民 4月23日，第二十一届上海国际车展上，商汤绝影发布了包括生成式智驾R-UniAD、4D世界模型"绝影开 悟2.0"、AI内核"绝影千机"在内的多个核心技术。 统观绝影发布的技术，一句话概括为：不是做"功能模型"，而是做"系统AI"。 过去几年，端到端技术被认为是智能驾驶实现路径中的"终极形态"，但大多数方案都卡在数据、性能、安 全三道门槛前。R-UniAD的思路，是在这三道门槛外，再造一层AI基础设施。 据介绍，R-UniAD结合世界模型和强化学习，以VLAR（视觉、语言、行动、强化学习）技术架构为核 心，支持仿真环境中进行闭环训练，让模型能在虚拟世界里"试错—强化—泛化"，最终形成系统策略能 力。 以车展现场展示的"施工占道场景"为例，R-UniAD先通过"绝影开悟"世界模型构建4D动态环境，再让模 型反复在这个世界里强化训练不同策略。最终结果是，模型不仅能精准避让，还能泛化到其它类似场景 中。 这种能力的意义在于，它不仅降低了数据成本，更提高了模型上限——在少数据场景里跑出更强泛化能 力，是当前行业少见的尝试。 如果说R-Uni ...

Group 1: Financial Performance - The company achieved an annual revenue of 14.08 billion yuan and an operating profit of 1.742 billion yuan in 2024, indicating steady operational progress and improved quality and efficiency [1] - The core membership segment saw significant growth, with the number of effective members reaching 73.31 million and membership revenue exceeding 5 billion yuan, a year-on-year increase of 19.3% [1] - The company expects to benefit from a tax exemption policy from 2024 to 2027, which will enhance profits by allowing a full refund of previously paid corporate income tax amounting to 260 million yuan [2] Group 2: Membership and International Expansion - The company focused on enhancing membership value, achieving a dual increase in membership scale and revenue through improved operational efficiency and innovative strategies [3] - The international app downloads surged from 130 million in 2023 to 261 million in 2024, with overseas revenue growing from 62 million yuan to 141 million yuan, exceeding the target of doubling [3] Group 3: Content and Innovation - The company made significant investments in original and customized series, with over 60% of its content being self-produced, leading to a notable increase in viewership [4] - The company has developed over 70 scenarios for generative AI applications, significantly reducing production costs and enhancing creative capabilities [4]

IDC总裁德尔普雷特在接受日本经济新闻（中文版：日经中文网）的独家采访时透露，2025年全 球IT支出增长率将从同比增长10％降至同比增长5％。2025年半导体市场的增长率也可能从原本 预计的11％降至9％左右…… IDC总裁克劳福德・德尔普雷特（Crawford Del Prete）在接受日本经济新闻（中文版：日经中文 网）的独家采访时透露了上述内容。此前，IDC预计2025年的IT支出规模为4.1万亿美元，同比增 长10％。德尔普雷特表示，由于美国启动关税政策，"被认为会出现增长率降至5％的情况"。 版权声明：日本经济新闻社版权所有，未经授权不得转载或部分复制，违者必究。 IT支出不仅包括智能手机和服务器等硬件，还涵盖云服务等软件产品。预计硬件和软件都将因特 朗普关税而导致成本增加，并转嫁到价格上。德尔普雷特预测称："价格上涨会让消费者难以购买 产品和服务，自2025年下半年起，IT支出将减速"。 除了IT支出外，IDC还汇总了半导体市场动向。包括特朗普政府正在考虑引进的半导体关税政 策，IDC还在研究2025年半导体市场的增长率从原本预计的11％降至9％左右的可能性。 半导体的设计、生产、制造设备、零部 ...

Core Insights - The emergence of open-source large models, such as DeepSeek-R1, has sparked a revolution in the AI industry, challenging the debate over the merits of open-source versus closed-source models [2][3] - Open-source AI is still in its early stages, with significant potential for industry transformation, but challenges related to security and commercialization remain [2][11] Group 1: Open-Source AI Impact - Open-source models are reshaping global industry dynamics, enabling lower-cost access to advanced AI capabilities for small and medium enterprises [3][4] - The cost of using DeepSeek-R1 for inference is only 1/30 of OpenAI's model, allowing developers to create applications at a fraction of the cost, such as a legal document generation tool that saw a 90% cost reduction [4] - The rise of "Model as a Service" (MaaS) is changing the service model of traditional cloud providers, making it easier for startups to deploy AI applications without building their own infrastructure [4][5] Group 2: Security Challenges - Security has become a major concern with the rise of open-source AI, with 57% of IT decision-makers citing privacy and data security as top issues [6][9] - High-profile security incidents, such as unauthorized access to Hugging Face's platform, highlight the vulnerabilities associated with open-source models [6][7] - DeepSeek has faced significant security threats, including DDoS attacks and data breaches, indicating the challenges that open-source platforms must address [7][8] Group 3: Future Considerations - The demand for computational power remains high, even with reduced costs, necessitating better observability and security measures from cloud service providers [5][9] - The integration of edge computing with AI is creating new security challenges, requiring companies to develop more complex security frameworks [10] - As data becomes a critical asset, ensuring data privacy and security in AI deployments is essential for companies [10][11]

近日， 未可知人工智能研究院副院长张孜铭老师受邀前往南方基金总部，开展了一场主题 为"DeepSeek+：企业的AI战略落地"的企业培训。 此次培训聚焦于 AI技术在企业战略中的应用与落地 ，旨在帮助南方基金的员工们深入理解AI 如何赋能企业运营、提升工作效率并推动业务创新，为企业的数字化转型提供有力支持。 张孜铭老师是未可知人工智能研究院副院长 ，北京大学与新加坡国立大学双硕士，著有 《DeepSeek使用指南》等多部作品，并参与起草了《生成式人工智能数据应用合规指南》等 行业标准。他在人工智能领域拥有深厚的学术背景和丰富的实践经验，为本次培训提供了坚实 的专业支撑。 培训中，张老师还分享了 未可知人工智能研究院在 AI培训、解决方案开发、咨询研究等方面 的丰富经验和成果。 他介绍了研究院的 专家团队 ，他们在人工智能领域拥有深厚的学术背景 和丰富的实践经验，为研究院的发展提供了坚实的人才支撑。 针对企业如何落地 AI战略，张老师提出了诸多建议。 他强调，企业应积极培养员工的AI技 能，掌握AI工具的使用方法；同时，要善于发现工作中的重复性任务，思考如何用AI替代，将 AI融入业务流程的标准化操作中。他还 ...

【环球网科技综合报道】随着ChatGPT、Deepseek等生成式AI聊天机器人在全球范围内掀起热潮，其应 用场景正不断拓展，如今甚至"跨界"至投资领域。近期，社交媒体上涌现出大量散户分享使用生成式AI 进行选股投资的经验，他们认为，相比咨询专业投资顾问，利用AI制定投资策略不仅成本低廉，而且 似乎颇具成效。然而，这一新兴趋势背后，是否真的隐藏着财富密码？还是潜藏着不为人知的风险？ 然而，专家也提醒广大散户，在使用AI进行投资研究时，务必保持警惕，仔细核查AI模型所引用的资 料来源。由于AI训练数据可能存在时效性不足或质量参差不齐的问题，其生成的内容有时可能包含过 时信息甚至错误观点。更为严重的是，AI在处理复杂或小众内容时，还可能出现"幻觉"现象，即输出与 事实不符或逻辑混乱的信息。 "AI幻觉本质上源于模型对知识点的掌握不足或概率计算错误。"一位不愿具名的AI领域专家解释 道，"对于不具备相应专业知识的散户而言，很难在短时间内准确识别这些错误信息，从而可能误导其 投资决策。" 面对AI在投资领域的应用热潮，专家呼吁广大散户保持理性态度，既要看到AI在信息处理方面的潜 力，也要清醒认识到其局限性和潜在风险。 ...

Core Viewpoint - The collaboration between SenseTime and Dongfeng Motor in the field of intelligent driving represents a deep integration of technology and industry, providing a practical model for the sector [1] Technological Breakthroughs - SenseTime showcased the R-UniAD solution based on the VLAR architecture, achieving near real-time interactive simulation and reducing the need for real data by two orders of magnitude [3] - The system demonstrated a perception accuracy of 98% compared to real data and improved simulation efficiency by five times over industry benchmarks [3] - The joint team has made significant progress in developing an end-to-end autonomous driving system, which features a "dual insurance" architecture combining a learning-based decision system with a rule-driven safety redundancy [3] Industry Collaboration - SenseTime has partnered with four car manufacturers, covering seven vehicle models, and plans to mass-produce an end-to-end solution based on the NVIDIA Thor platform by Q4 2025 [3] - The company has maintained a leading market share in intelligent cockpit software for five consecutive years, with cumulative deliveries exceeding 3.6 million units [3] Ecological Evolution - SenseTime introduced the first in-vehicle AI operating system, "Jueying Qianji," with a response latency of less than 300ms, supporting over 100 intelligent agents [6] - The upgraded "New Member" system showcases multimodal recognition and deep thinking capabilities, providing personalized services during real-time demonstrations [6] - The "Travel Medical" system integrates medical-grade health monitoring and connects with third-party medical services [8] Industry Observations - The recent regulatory adjustments are viewed as a turning point towards maturity rather than a pause in the intelligent transformation process [8] - The penetration rate of L2-level assisted driving reached 62% in 2024, with a 37% year-on-year increase in user complaints regarding functional safety [8] - The collaboration model between Dongfeng and SenseTime reflects a dual-track strategy of "independent control + open collaboration" in the intelligent transformation of traditional car manufacturers [8] Common Industry Challenges - The introduction of AI-native technologies like world models and reinforcement learning is breaking through the limitations of rule-driven systems [11] - There is an urgent need to establish compliant data collection systems and unified safety assessment standards within the industry [11] - The deep binding of AI companies and vehicle manufacturers may be a key pathway to address these common challenges [11]

生成式AI 一、 OpenAI 称刚刚对GPT 4o模型进行了升级，个性化更强 1. 优化了记忆存储机制，使AI能更智能地记忆和回忆对话信息； 2. STEM领域推理能力显著提升，可更好解决数学、科学、工程等复杂问题； 3. 对话风格更加主动自然，擅长引导对话方向，同时回复更贴近真实交谈。 https://mp.weixin.qq.com/s/oZVIP1hLb2ZZu5E9VNr5Zw 二、 实测免费DeepResearch！轻量版，速度更快重视脉络梳理 1. OpenAI发布基于o4-mini的轻量版DeepResearch，免费用户可使用，付费用户获额外使 用额度； 2. 轻量版与满血版相比，用时更短、内容更精简，但保持相近的智能水平； 3. 实测显示轻量版更注重梳理重点脉络，适合需要快速了解概况的场景。 https://mp.weixin.qq.com/s/0vZvNaAhEQQOqUfg3YiIdQ 2. 系统通过层级化分解和提交历史分析来理解代码全局结构，已索引3万个仓库，处理超40 亿行代码； 3. 使用方式简单，只需将github.com替换为deepwiki.com即可访问对应仓库的AI文档 ...

Core Viewpoint - Digital China Group Co., Ltd. reported a revenue of 31.78 billion yuan for Q1 2025, marking an 8.6% year-on-year increase, with a net profit of 224 million yuan, up 10.4% [1] Group 1: Financial Performance - Overall revenue reached 31.78 billion yuan, reflecting an 8.6% year-on-year growth [1] - Non-GAAP net profit was 224 million yuan, showing a 10.4% increase compared to the previous year [1] - Operating cash flow recorded a net inflow of 2.73 billion yuan [1] Group 2: AI Strategy and Implementation - The company focused on AI technology transformation, enhancing its digital technology stack and launching multiple AI projects [1] - New direct customer opportunities increased by 78% year-on-year, with significant growth in flagship AI projects [1] - The company successfully implemented AI solutions in various industries, including retail, automotive, and manufacturing [1] Group 3: R&D and Product Development - R&D expenses grew by 9.3% year-on-year, emphasizing the commitment to generative AI and AI accessibility [2] - The launch of the DeepSeek version of the KunTai integrated machine and the KunTaiCube toolbox supports private deployment and offers cost-effective smart upgrade solutions [2] - The company secured an 800 million yuan order for its self-branded products from China Telecom for the 2024-2025 procurement project [2] Group 4: Automotive Industry Focus - Digital China introduced the "AI for Process Maturity Model" to outline the AI evolution blueprint for the automotive industry [2] - The company launched an "end-to-end one-stop AI workspace" and the "Digital China Intelligent Vehicle Solution" to enhance productivity and reduce costs for automotive enterprises [2] - Multiple "ten million-level" orders were won for digital marketing and data compliance projects from leading automotive companies [2]

到2030年，全球半导体公司计划在新晶圆厂（fabs）建设上投资约1万亿美元，行业年收入也有望 突破1万亿美元。这一数字尚未包括生成式AI（gen AI）在中等乐观情境下可能带来的额外增长潜 力。除了满足日益增长的市场需求，这些投资也将增强各地区在半导体价值链上的供应弹性。 然而，尽管这场全球范围的大规模投资有望显著扩展半导体的产能版图，实现其预期效益的道路却 并不平坦。除了已公布的建设项目本身执行难度大之外，至少还有五大结构性障碍，特别在北美和 欧洲市场可能长期制约新增投资带来的实质进展：包括资本与运营成本结构、材料需求增长、关键 原材料及封装环节的离岸集中、物流与处理瓶颈，以及人才短缺等问题。若行业希望实现投资的长 期价值，就必须正视并逐一应对这些挑战。 底层半导体资本和运营成本动态 近年来，全球多国通过战略政策推动半导体制造和供应链的本土化。例如，美国推出《两党基础设 施法案》、《芯片与科学法案》、《通胀削减法案》及各州激励措施，试图吸引企业在本土建厂。 类似的激励措施也在欧洲、印度、日本、中国大陆、东南亚、韩国与中国台湾陆续推出，从而带动 全球晶圆厂建设热潮。 不过，从实际成本结构来看，美国本土建设先 ...