他认为人类和AI将演进出全新的协同方式，从Vibe Coding到Vibe Working。未来，每个家庭、工厂、公 司，都会有众多的Agent和机器人24小时为我们服务。正如电曾经放大了人类物理力量的杠杆，ASI将 指数级放大人类的智力杠杆，人会变得比历史上任何时候都强大。 同时，他表示目前阿里正积极推进3800亿的AI基础设施建设，并计划追加更大的投入。 9月24日，阿里巴巴集团CEO、阿里云智能集团董事长兼CEO吴泳铭在云栖大会演讲中表示，实现AGI 已是确定性事件，但这仅是起点，终极目标是发展出能自我迭代、全面超越人类的ASI（超级人工智 能）。 ...

阿里巴巴集团董事兼CEO吴泳铭发表主旨演讲。他表示，AGI是具备人类公共认知能力的智能系统，现 在在行业看来已经成为现实的世界。但吴泳铭指出，AGI并非是发展的终点，只是全新的起点，AI肯定 不会止步于AGI， 它将迈向超越人类智能能够自我迭代进化的超级人工智能行业，称为ASI超级人工智 能。吴泳铭称，资本开支会在3800亿基础上，增加更多的投入。 【免责声明】本文仅代表作者本人观点，与和讯网无关。和讯网站对文中陈述、观点判断保持中立，不 对所包含内容的准确性、可靠性或完整性提供任何明示或暗示的保证。请读者仅作参考，并请自行承担 全部责任。邮箱：news_center@staff.hexun.com 阿里巴巴短线拉升，现涨超4%，报价165.8港元，夜盘涨近3%。 | 09988 阿里巴巴-W | | | | --- | --- | --- | | 165.800↑+6.400 +4.02% | | 交易中 09/24 10:40 | | SEES BEBER | | ● 快捷交易 | | 最高价 165.900 | 开盘价 158.500 | 成交量 5635.01万 | | 最低价 157.600 | 昨 ...

(责任编辑：宋政 HN002) 【免责声明】本文仅代表作者本人观点，与和讯网无关。和讯网站对文中陈述、观点判断保持中立，不对所包含内容 的准确性、可靠性或完整性提供任何明示或暗示的保证。请读者仅作参考，并请自行承担全部责任。邮箱： news_center@staff.hexun.com 格隆汇9月24日｜阿里巴巴集团CEO、阿里云智能集团董事长兼CEO吴泳铭在云栖大会演讲中表示，实 现AGI已是确定性事件，但这仅是起点，终极目标是发展出能自我迭代、全面超越人类的超级人工智能 (ASI)，以解决气候、能源、星际旅行等重大科学难题。 通往超级人工智能之路分为三个阶段：1、智能涌现，AI通过学习人类知识具备泛化智能；2、"自主行 动"，AI掌握工具使用和编程能力以"辅助人"，这是行业当前所处的阶段；3、"自我迭代"，AI通过连接 物理世界的全量原始数据，实现自主学习，最终能够"超越人"。 ...

通往超级人工智能之路分为三个阶段：1、智能涌现，AI通过学习人类知识具备泛化智能；2、"自主行 动"，AI掌握工具使用和编程能力以"辅助人"，这是行业当前所处的阶段；3、"自我迭代"，AI通过连 接物理世界的全量原始数据，实现自主学习，最终能够"超越人"。 9月24日，阿里巴巴集团CEO、阿里云智能集团董事长兼CEO吴泳铭在云栖大会演讲中表示，实现AGI 已是确定性事件，但这仅是起点，终极目标是发展出能自我迭代、全面超越人类的超级人工智能 （ASI），以解决气候、能源、星际旅行等重大科学难题。 ...

9月24日，阿里巴巴集团CEO、阿里云智能集团董事长兼CEO吴泳铭在云栖大会演讲中表示，实现AGI 已是确定性事件，但这仅是起点，终极目标是发展出能自我迭代、全面超越人类的超级人工智能 （ASI），以解决气候、能源、星际旅行等重大科学难题。 （文章来源：第一财经） 通往超级人工智能之路分为三个阶段：1、智能涌现，AI通过学习人类知识具备泛化智能；2、"自主行 动"，AI掌握工具使用和编程能力以"辅助人"，这是行业当前所处的阶段；3、"自我迭代"，AI通过连接 物理世界的全量原始数据，实现自主学习，最终能够"超越人"。 ...

通往超级人工智能之路分为三个阶段：1、智能涌现，AI通过学习人类知识具备泛化智能；2、"自主行 动"，AI掌握工具使用和编程能力以"辅助人"，这是行业当前所处的阶段；3、"自我迭代"，AI通过连接 物理世界的全量原始数据，实现自主学习，最终能够"超越人"。 9月24日，阿里巴巴集团CEO、阿里云智能集团董事长兼CEO吴泳铭在云栖大会演讲中表示，实现AGI 已是确定性事件，但这仅是起点，终极目标是发展出能自我迭代、全面超越人类的超级人工智能 （ASI），以解决气候、能源、星际旅行等重大科学难题。 终极目标是发展出能自我迭代、全面超越人类的超级人工智能（ASI）。 ...

9月24日，在2025年云栖大会上，阿里巴巴集团董事兼CEO吴泳铭表示，AGI并非是发展的终点，只是全新的起点，AI肯定不会止步于AGI, 它将迈向超越人类智能能够自我迭代进化的超级人工智能行业，称为ASI超级人工智能。ASI将作为全面超越人类智能的系统，有可能创造出一批超级科学家来和潜在的高级工程师。ASI将以我们难以想象的思路解决现在未被解决的科学工程问题，比如攻克医学难题，发明新材料，解决可持续能源和气候问题，甚至未来的风险。 ...

AI发展与就业市场 - AGI（通用人工智能）可能导致知识工作仅受计算能力限制，资本所有者将能无限复制人才[2][4] - 斯坦福研究显示，在AI暴露度最高的职业中，22-25岁年轻人的就业率下降约13%[9] - Handshake数据显示，入门级职位需求下降15%，每个职位的申请人数同比增长30%[11] 永久下层阶级风险与对策 - AGI可能导致社会阶层固化，缺乏流动性，形成永久下层阶级[3][6] - 一些人认为UBI（全民基本收入）或UHI（全民高收入）是解决方案，但存在社会主义和通货膨胀的争议[14][15][16] AI应用与工作模式转变 - AI可能只能完成任务的中间环节，人类仍需参与提示、协调和验证，工作模式将转变为人机协作[18][19] - 企业应优先投资于使用AI的部门，提高效率，从而可能促进公司更快增长和创造新的工作岗位[24][26] - AI将创造目前无法想象的新工作岗位，并为新创始人提供与大型公司竞争的机会[27][28][29] 行业观点与未来展望 - Box CEO认为，企业应优先投资于使用AI的部门，提高效率[23][24] - OpenAI CTO认为，AI并非完美，存在错误，但AI将创造大量我们今天无法想象的工作[27] - 行业普遍认为，需要积极教授人们如何使用AI工具，以提高生产力和价值[32]

Core Insights - The paper predicts that in the era of AGI, while the economy may experience exponential growth due to the expansion of computing power, ordinary people's wages will be "locked" by the cost of computing power, becoming completely decoupled from economic growth [1][2][15]. Group 1: Wage Dynamics in the AGI Economy - In traditional economics, wages are linked to "skill scarcity," where individuals can command a premium for unique skills [4]. - The paper argues that in an AGI economy, wages will depend on the computing power required to replicate a person's skills, rather than the scarcity of those skills [5][10]. - The value of professions, such as that of a surgeon, will be determined by the computing power needed for AGI to simulate their tasks [9][11]. Group 2: Wealth Distribution and Power Shift - The model suggests that as AGI automates all bottleneck jobs, the share of labor in GDP will approach zero, with nearly all new wealth flowing to computing capital [22][23]. - Computing power will become the core asset determining wealth distribution, akin to land and machinery during the Industrial Revolution [25]. - Companies like Microsoft are already investing heavily in AI infrastructure, with plans to spend approximately $80 billion on AI-driven data centers in the 2025 fiscal year [26]. Group 3: The Role of Accessory Work - While AGI will automate many tasks, there will still be "accessory work" that holds social value but does not drive economic growth, such as caregiving and artistic endeavors [35][36]. - These roles may not provide increasing income but will retain a necessary social significance, as they are less likely to be automated due to their complexity or the value of human interaction [41][42]. Group 4: Future Wealth Distribution Mechanisms - The paper highlights a critical issue: as labor's share of GDP diminishes, a new mechanism for wealth distribution must be established to address the concentration of wealth among computing power owners [51][52]. - Potential solutions include universal dividends from computing profits or treating computing resources as public capital, similar to Norway's oil fund model [56]. - The concentration of computing power among a few tech giants raises concerns about increasing social inequality if wealth distribution mechanisms are not implemented [59][60].

Core Insights - Nvidia is investing $100 billion in OpenAI to build a large-scale AI data center with a capacity of at least 10GW, enhancing its ecosystem and competitive position against AMD [2][3] - The collaboration with OpenAI will utilize millions of Nvidia GPUs to train next-generation AI models, addressing OpenAI's computational resource constraints [3][4] - Nvidia's investment in Intel, amounting to $5 billion, aims to develop AI infrastructure and personal computing products, indicating a strategic defense against market competition [6][7] Nvidia and OpenAI Collaboration - The partnership will leverage Nvidia's hardware resources to establish a 10GW AI data center, which is projected to meet a significant portion of global AI workload demands [3][4] - Nvidia's CEO highlighted the historical collaboration with OpenAI, emphasizing the importance of this data center for the next era of intelligence [3] - The project will utilize up to 5 million Nvidia AI chips, equivalent to the company's total expected shipments for the year [3] Market Reaction - Following the announcement of the collaboration, Nvidia's stock rose nearly 4%, pushing its market capitalization above $4.5 trillion, reflecting investor optimism [4] - OpenAI, facing computational bottlenecks, will benefit significantly from the partnership, which will enhance its capacity to support its growing user base [4][5] Nvidia and Intel Partnership - Nvidia and Intel's collaboration focuses on three main areas: connecting CPUs and GPUs using Nvidia's NVLink technology, custom CPU development based on Nvidia's needs, and launching new x86 SoCs integrated with Nvidia's RTX GPUs [6][7] - Nvidia's investment in Intel is seen as a strategic move to alleviate its high-end chip production bottlenecks and penetrate the AI acceleration market [7] Industry Trends - The partnerships reflect a broader trend among tech giants to collaborate in response to external pressures, including the escalating AI competition and tightening global antitrust regulations [8][9] - The collaborations are expected to accelerate the development of AI infrastructure, enhancing the industry's growth potential [9][10] - The Chinese AI industry faces challenges from these international collaborations but also has opportunities for technological innovation and industry upgrades [9][10]