邢自强指，AI需要四大支柱去支撑，中国在AI人才、基础设施及数据方面均具优势。有人才才能改进 演算法，独辟蹊径，节省效率，而全球AI相关的一半人才是中国培养的；基础设施就是算力中心，需 要有大量的电网、发电站及储能设备，连同冷却、基本维护等设施，这些设施在中国均较便宜。要训练 大模型变成最好的AI，便需要大量数据，中国老百姓(603883)一直使用的微信、制造业产业链上的第一 手资料，均为训练AI提供大量数据，故此这方面中国亦具有优势。 格隆汇12月16日｜摩根士丹利中国首席经济学家邢自强认为，中美对AI的观点与发展路径截然不同， 中国长远能否超越美国拭目以待，但凭借可靠的人才、基础设施及数据，可弥补GPU的不足。 邢自强表示，美国的AI企业均是超大规模服务商(Hyperscaler)，大家都拼命的砸钱、拼算力，就是要拼 谁能先通过这些军备竞赛，能先达到AGI(通用人工智能)，所以美国走的是一个重量化。中国的AI模型 普遍都属开源性质，一开始想到的是AI本身未必要那么赚钱，而是要让这些AI工具尽快落地市场化， 故中国走的是一个轻量化、比较便宜，但是又可以大量铺开的AI应用模型之路。 ...

Group 1 - OpenAI's Codex product lead indicates that human typing speed will become a bottleneck for the development of Artificial General Intelligence (AGI) due to the need for users to write prompts and verify AI outputs [1] - Zhiyuan Robotics announced the launch of the "Qingtian Rental" platform on December 22, aiming to promote standardization and scalable development in the robot rental industry [1] - Yuejiang Robotics has been included in the Hong Kong Stock Exchange's newly launched "HKEX Technology 100 Index," which selects constituents based on R&D investment, innovation capability, and industry representation [1] Group 2 - Donghua Software has established Lingyao Intelligent Technology Company with a registered capital of 300 million yuan, focusing on various AI-related businesses including AI infrastructure and software development [2] - Guangdong's core AI industry is projected to reach a scale of 220 billion yuan by 2024, reflecting a year-on-year growth of approximately 22%, with the smart robotics sector expected to generate 99.2 billion yuan in revenue [2] Group 3 - The 2025 World Robot Skills Competition, themed "Intelligent Creation for the Future," is set to take place from December 26 to 30, 2025, in Longgang District, Shenzhen, inviting global innovation teams and skilled talents in the robotics field [3]

Core Insights - The article discusses the transition from the "Scaling Law" to the "Densing Law," emphasizing the need for sustainable development in AI models as data growth slows and computational costs rise [2][3][15]. - The "Densing Law" indicates that model capability density increases exponentially, with capability density doubling approximately every 3.5 months, while the parameter count and inference costs decrease significantly [11][28]. Group 1: Scaling Law and Its Limitations - The "Scaling Law" has faced challenges due to bottlenecks in training data and computational resources, making it unsustainable to continue increasing model size [15][16]. - The current training data is limited to around 20 trillion tokens, which is insufficient for the expanding needs of model scaling [15]. - The computational resource requirement for larger models is becoming prohibitive, as seen with LLaMA 3, which required 16,000 H100 GPUs for a 405 billion parameter model [16]. Group 2: Introduction of Densing Law - The "Densing Law" proposes that as data, computation, and algorithms evolve together, the density of model capabilities grows exponentially, allowing for more efficient models with fewer parameters [11][28]. - For instance, GPT-3 required over 175 billion parameters, while MiniCPM achieved similar capabilities with only 2.4 billion parameters [24]. Group 3: Implications of Densing Law - The implications of the Densing Law suggest that achieving specific AI capabilities will require exponentially fewer parameters over time, with a notable case being Mistral, which achieved its intelligence level with only 35% of the parameters in four months [32][33]. - Inference costs are also expected to decrease exponentially due to advancements in hardware and algorithms, with costs for similar capabilities dropping significantly over time [36][39]. Group 4: Future Directions and Challenges - The future of AI models will focus on enhancing capability density through a "four-dimensional preparation system," which includes efficient architecture, computation, data quality, and learning processes [49][50]. - The article highlights the importance of high-quality training data and stable environments for post-training data, which are critical for the performance of models in complex tasks [68][70]. Group 5: End-User Applications and Market Trends - By 2026, significant advancements in edge intelligence are anticipated, driven by the need for local processing of private data and the development of high-capacity edge chips [11][45][76]. - The article predicts a surge in edge applications, emphasizing the importance of privacy and personalized experiences in AI deployment [76][77].

Group 1 - Wall Street strategists are advising clients to shift investment focus from "Tech Giants" to traditional cyclical sectors such as healthcare, industrials, energy, and finance by 2026, driven by concerns over the sustainability of tech valuations and rising optimism about the U.S. economic outlook [1] - A report from BCA Research indicates that the biggest threat to the U.S. economy in 2026 may stem from the financial markets themselves, suggesting that a potential stock market crash could directly push the economy into recession, challenging the prevailing market view [3][4] - The report highlights that approximately 2.5 million "excess retirees" are a key support for the current U.S. economy, as their consumption is closely tied to stock market performance, creating a demand-side that is sensitive to market fluctuations [3][5] Group 2 - The "excess retiree" phenomenon, which has seen 2.5 million individuals retire early due to the pandemic and stock market gains, injects strong demand into the economy but does not contribute to labor supply, leading to a tight labor market and persistent inflation around 3% [5][8] - BCA Research predicts that the Federal Reserve will prioritize avoiding a market crash over achieving its 2% inflation target, potentially leading to a tolerance for higher inflation and aggressive rate cuts in response to any signs of economic or market weakness [4][10] - The report notes that the current market rally is historically concentrated, with about two-thirds of global market capitalization in U.S. stocks, and 40% of U.S. market value concentrated in just ten stocks, which are heavily reliant on the success of the generative AI narrative [11][13] Group 3 - Recent market trends show a divergence among leading tech stocks, with significant market value fluctuations indicating that not all tech stocks are viewed as a single entity, allowing for potential value investment opportunities [13] - BCA Research suggests that as the era of U.S. tech stocks outperforming the market may be ending, funds could rotate into undervalued sectors and regions, such as healthcare and European markets, which are not facing the same inflation pressures as the U.S. [13]

Core Viewpoint - A recent report from BCA Research indicates that the biggest threat to the U.S. economy in 2026 may stem from the financial markets themselves, rather than an economic recession dragging down the stock market. The report suggests that a potential stock market crash could directly push the U.S. economy into recession, challenging conventional market views [1][2]. Economic Structure and Risks - The report highlights a significant structural change in the U.S. labor market, with approximately 2.5 million "excess retirees" whose consumption is closely tied to stock market performance. This group has retired early due to the pandemic and the subsequent stock market boom, creating a demand side that is sensitive to stock market fluctuations [1][3][5]. - The consumption of these retirees injects strong demand into the U.S. economy, but their retirement means they do not contribute to labor supply, leading to a constrained labor market. This situation creates a delicate balance where strong demand exists alongside limited supply, preventing a recession driven by weak demand [5][7]. Federal Reserve's Dilemma - BCA Research outlines a dilemma for the Federal Reserve: maintaining a 2% inflation target while avoiding a recession. The report predicts that the Fed will prioritize preventing a market crash over its inflation target, potentially allowing inflation to rise above 2% and adopting aggressive rate cuts in response to any signs of economic or market weakness [2][8]. Market Concentration and Challenges - The report notes that the current market rally is historically concentrated, with about two-thirds of global stock market value concentrated in U.S. stocks, and 40% of that in just ten stocks. This concentration poses risks, as the fortunes of these stocks are heavily tied to the success of the generative AI narrative [9][11]. - However, there are signs of divergence among leading tech stocks, indicating that the market is not treating all tech stocks as a single entity. This divergence suggests that value investors are still assessing individual company valuations [11][12]. Investment Opportunities - BCA Research suggests that as the era of U.S. tech stocks outperforming the market may be coming to an end, funds could rotate into undervalued sectors and regions, such as healthcare and European markets. Unlike the U.S., Europe does not face inflationary pressures caused by labor market distortions, creating a favorable environment for the bond market [12].

一份来自华尔街的最新展望报告颠覆了传统认知，指出2026年美国经济面临的最大威胁可能源自金融市场本身。 根据投资研究机构BCA Research的最新展望，2026年投资者面临的核心风险已经发生反转：不再是经济衰退拖垮股市，而是股市的潜在崩 盘可能直接将美国经济推入衰退。这一观点挑战了市场的普遍看法，并指出美国经济的韧性正悬于一个由股市财富支撑的脆弱平衡之上。 BCA Research在报告中明确指出，当前美国经济的一个关键支撑来自于约250万"超额退休"人群的消费支出。这部分人群因新冠疫情后的 股市繁荣而提前退休，他们的消费能力与股市表现直接挂钩，形成了一个"对股市敏感"的需求侧。 报告分析，这种结构性变化给美联储带来了棘手的两难。一方面，这批高技能退休人员的离场加剧了劳动力短缺，使通胀顽固地维持在 3%左右；另一方面，若为抑制通胀而维持高利率，则可能刺破股市泡沫，摧毁这部分关键消费，从而引发经济衰退。 因此，BCA Research预测，美联储将把避免市场崩盘置于其2%通胀目标之上，选择容忍更高的通胀率，并可能在任何经济或市场疲软的 迹象出现时采取激进的降息措施。这一政策路径，叠加史上最集中的市场涨势，为 ...

中新网海口12月13日电 (记者张茜翼)2025第十四届上市公司发展年会暨海南自贸港开放机遇交流大会12 日在海口举办。超过400家上市公司、20多家机构赴会，政产学研精英共话封关新局下的政策路径、全 球化机遇。 苏波表示，未来5年工业企业数字化转型是推进智能制造、实现制造业创新发展的主战场，需促进实体 经济与数字经济深度融合，加快数智技术创新，强化算力、算法、数据高效供给，建立健全数据要素基 础制度，以数智技术赋能千行百业。 中国国际经济交流中心副理事长胡晓炼聚焦海南开放特质指出，海南的开放是高水平开放，核心是制度 型开放——通过推进规则、规制、标准、管理与国际高标准经贸规则和良好实践相容衔接，从降低关 税、放宽市场准入等边境开放，转向更注重对接国际通行规则、完善国内制度体系的边境后开放。 大会还邀请国内行业领军企业围绕"AI价值锚点"与"新出海范式"展开智慧风暴，分享一线经验与思考。 (完) （文章来源：中国新闻网） 12月18日，海南自贸港将迎来全岛封关运作。海南国际经济发展局局长唐华在主会场致辞与推介环节表 示，海南自贸港的政策核心是"一线放开、二线管住、岛内自由"的制度型开放，叠加"零关税、低税 率 ...

Core Insights - The next stage of artificial intelligence (AI) development requires overcoming two major challenges: the Transformer architecture and the backpropagation algorithm [1][7][54] - The focus should shift from larger models to creating "living" models that possess native memory, autonomous learning, and continuous evolution capabilities [2][4][48] - This transition signifies a move from centralized cloud computing to decentralized learning, where each device can contribute to knowledge generation [3][5][70] Group 1: Hardware Awakening - The concept of "hardware awakening" suggests that devices can learn and adapt in real-time, transforming them from mere tools into active intelligent agents [4][64] - A multitude of such intelligent agents collaborating in the real world can lead to the emergence of collective intelligence [5][71] - The current reliance on the Transformer model limits the potential for true intelligence, as it does not facilitate autonomous learning or native memory [21][30][76] Group 2: Redefining Value - The future of AI will redefine the value of hardware, moving beyond traditional metrics like memory and processing power to focus on the co-creation of value between users and devices [64][66] - Users should pay for intelligence rather than token consumption, as the latter is seen as an inefficient model [15][19][21] - The emergence of devices with autonomous learning capabilities will enhance user experience and privacy, as data remains localized [68][69] Group 3: Collective Intelligence - Collective intelligence arises when each device possesses its own intelligence and can learn from the physical world, similar to human collaboration [71][76] - True intelligence is characterized by the ability to generate knowledge rather than merely disseminating it, which is a limitation of current large models [75][77] - The path to general artificial intelligence is through collective intelligence rather than the centralized model exemplified by companies like OpenAI [77]

Core Insights - The establishment of the Spiking Intelligence Lab (SIL) aims to develop brain-like models with neuro-dynamic characteristics, focusing on the integration of artificial intelligence and human intelligence [1][3] - The lab is led by Professor Li Guoqi and is part of the Tianqiao Brain Science Research Institute, which seeks to provide key capabilities for the "discovery-based intelligence" proposed by founder Chen Tianqiao [1][3] Research Focus - The lab emphasizes the importance of neuro-dynamics, contrasting with mainstream AI models that rely on scaling parameters, and aims to create a full-brain architecture with strong perception, memory, and thinking capabilities [3] - The research will support the construction of a full-brain architecture that operates with approximately 20 watts of power, similar to the human brain, which has complex operations supported by billions of neurons [3] Funding and Resources - The Tianqiao Brain Science Research Institute plans to invest over $1 billion to build dedicated computing clusters to provide resources for young scientists, focusing on structural exploration rather than scaling [4] - This investment aims to foster interdisciplinary innovation and support the verification of memory mechanisms and new neuro-dynamic hypotheses [4] Technological Advancements - The lab has developed the first brain-like pulse model, "Shunxi 1.0," which achieves breakthroughs in brain-like computing and large model integration, offering a new technical route for the next generation of AI [4][5] - The "Shunxi 1.0" model requires only about 2% of the data volume compared to mainstream models while achieving comparable performance in various language understanding and reasoning tasks [5]

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 ...