Whither AI? Pt 3. AI and Profit（双语）人工智能向何处去？第三部
Whither AI?
Part 3. AI and Profitability
https://blog.creaders.net/user_blog_diary.php?did=NTMwNDIw
https://www.facebook.com/andrew.colesville
https://x.com/mwsansculotte
References:
Part 2. AI, the Class Struggle and Human Intelligence (HI)
https://blog.creaders.net/user\_blog\_diary.php?did=NTI4NDI4
For Part 1, See: https://blog.creaders.net/user\_blog\_diary.php?did=NTI2MTU5
In Part 2, it was mentioned that capitals in the "AI industry" could not attain profits as much as the non-AI industry could, because the labor inputs immersed in the AI data centers such as the GPU's, electricity, cooling liquids such water, constructions, compute machinery were mostly dead and accounted for and dead labor did not produce surplus value hence profit for capitals. Only rather limited and gradually diminishing living labor involving staff members for training of AI, data-collection and maintenance contributed to the generation of rather small and negligible AI-profit for capitals.
*
GenAI (Generative AI) is a new content creation method following Professional-Generated Content (PGC) and User-Generated Content (UGC). It leverages technological advantages in creativity, expression, iteration, dissemination, and personalization to create new forms of digital content generation and interaction. With technological advancements, AI writing, AI music composition, AI video generation, AI speech synthesis, and the recently popular AI drawing have all generated considerable buzz in the creative field. Simply inputting a few keywords can generate a painting in seconds.(Source: google.com)
Stanford AI Club: Jason Wei on 3 Key Ideas in AI in 2025
https://www.youtube.com/watch?v=b6Doq2fz81U
Stanford AI Club Oct 18, 2025
This talk was recorded as part of the Stanford AI Club's Speaker Series.
Jason Wei is a prominent AI researcher currently working at Meta Superintelligence labs. Before Meta, Jason helped co-create the o1 model and Deep Research product at OpenAI. He also was one of the inventors of Chain of Thought reasoning and documented important research on emergent phenomena at Google Brain.
My comment:
Jason Wei's Stanford speech told us that
Intelligence is becoming a commodity and two of the three core ideas of AI are first, the cost of computation and knowledge is approaching zero and second, rapid takeoff of AI is unlikely to be achieved.
According to MIT's July 2025 research report:
"Despite $30–40 billion in enterprise investment into GenAI (AI Generated Content (Generative AI)), this report uncovers a surprising result in that 95% of organizations are getting zero return. The outcomes are so starkly divided across both buyers (enterprises, mid-market, SMBs) and builders (startups, vendors, consultancies) that we call it the GenAI Divide. Just 5% of integrated AI pilots are extracting millions in value, while the vast majority remain stuck with no measurable P&L impact.
"This divide does not seem to be driven by model quality or regulation, but seems to be determined by approach.
"Tools like ChatGPT and Copilot are widely adopted. Over 80 percent of organizations have explored or piloted them, and nearly 40 percent report deployment. But these tools primarily enhance individual productivity, not P&L performance. Meanwhile, enterprise-grade systems, custom or vendor-sold, are being quietly rejected. Sixty percent of
organizations evaluated such tools, but only 20 percent reached pilot stage and just 5 percent reached production. Most fail due to brittle workflows, lack of contextual learning, and misalignment with day-to-day operations." [
https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf ]
My comment:
The report did not present a comparison in manpower utilization between the 5 percent reaching production and the 95% that failed. It is highly possible that the one which used the most manpower would be the most successful in profitability as only living labor could produce surplus value or profit to be extracted by capital.
AI has the advantage of streamlining the extraction of workers' surplus value by means of increased productivity. The other side of the coin is that the more productive their labors become, the more will be unemployed, causing less surplus value hence profit. Hence AI capitalism has fallen into a self-inflicted irreconcilable contradiction and therefore will be doomed to failure all by itself.
According to "Secret Plans Reveal Amazon Plot to Replace 600,000 Workers With Robot Army" By Frank Landymore
https://futurism.com/category/artificial-intelligence Published Oct 23, 2025 2:10 PM EDT
*
In the talk of Interesting Times with Ross Douthat, the outline was as follows:
Is the A.I. revolution keeping the entire economy afloat? This week on “Interesting Times,” Ross talks with Jason Furman, an economist from the Harvard Kennedy School and a contributing writer for NYT Opinion, about how investors, policymakers and consumers should think about the boom — and potential bust — of the fastest growing segment of the American economy and look to past bubbles for answers.
https://www.youtube.com/watch?v=KoJH9cGtv1s
In the late 1990s, investors poured their money into internet-based companies. Telecom firms built out infrastructure — tens of millions of miles of fiber-optic cables — to support the anticipated boom.
It was a bubble. The profits never materialized, so investors sold off their shares and the companies collapsed. The Nasdaq lost more than three-fourths of its value, and it took 15 years to recover.
This time, the companies involved are much bigger. Yet unlike their dot-com forerunners, they’re actually earning some money. Just not as much — for now — as investors hope.(NYTimes, 10/27/2025)
Eric Lee for The New York Times
Nvidia Is Now Worth $5 Trillion as It Consolidates Power in A.I. Boom
The A.I. chip maker, which has become a linchpin in the Trump administration’s trade negotiations in Asia, is the first publicly traded company to top $5 trillion in market value.
My overall comment:
AI has arrived at a across road - its well-hyped efforts have shown its catastrophic illness that condemned them to years of penury because its LLM-generated hallucinations cannot be eliminated by compute alone. AI's survival depends solely on human's widespread and indispensable involvements. Its dangerous products of falsehood and fabrication have to be safely eliminated by human before any meaningful applicability can be realized . This necessarily applies to all AI applications, including both capitalist and non-capitalist ones. The former possibly will take advantage of much lower labor-cost in the third world than in US to hire workers overseas to do the job in order to save CAPEX, and then workers in those AI-centered countries will have to suffer serious unemployment dilemma.
Just as Joseph Schumpeter claimed in the 1920's, capital will have to produce destruction before achieving anything, AI will become a typical example of capitalist "creative destruction," from which a human intelligence(HI) will replace it.
On the other hand, it is gratifying that Elon Musk Takes On Wikipedia With AI-Generated ‘Grokipedia’—What To Know
https://www.youtube.com/watch?v=GeSjqnn0im0&list=RDNSGeSjqnn0im0&start_radio=1
Forbes Breaking News Oct 28, 2025
Elon Musk on Monday night unveiled “Grokipedia,” an online encyclopedia built entirely using AI-generated articles designed to take on Wikipedia, which Musk and his allies have claimed is biased and “woke.”
This is what I claimed that AI should rightly be - Smart Encyclopedia Based On Machine Evolution (S.E.B.O.M.E.)
For Smart Encyclopedia Based On Machine Evolution (S.E.B.O.M.E.), See https://blog.creaders.net/user_blog_diary.php?did=NTExNjgz
and
https://blog.creaders.net/user_blog_diary.php?did=NTEyMDM0
https://www.facebook.com/andrew.colesville
https://x.com/mwsansculotte
[Mark Wain 11/01/2025]
*
汉语译文
人工智能往何处去？
第三部分：人工智能与盈利能力
https://blog.creaders.net/user_blog_diary.php?did=NTMwNDIw
https://www.facebook.com/andrew.colesville
https://x.com/mwsansculotte
参考文献：
第二部分：人工智能、阶级斗争与人类智能 (HI)
https://blog.creaders.net/user\_blog\_diary.php?did=NTI4NDI4
第一部分请见：https://blog.creaders.net/user\_blog\_diary.php?did=NTI2MTU5
第二部分提到，“人工智能产业”的资本无法像非人工智能产业那样获得丰厚的利润，因为人工智能数据中心投入的大量劳动力，例如GPU、电力、冷却液（如水）、建筑和计算设备等，大多处于闲置状态，无法创造剩余价值，因此资本无法从中获利。只有数量有限且逐渐减少的、涉及人工智能训练、数据收集和维护等工作的员工劳动，才能为资本创造微乎其微的人工智能利润。
*
生成式人工智能（GenAI）是继专业生成内容（PGC）和用户生成内容（UGC）之后的一种新型内容创作方法。它利用技术在创意、表达、迭代、传播和个性化方面的优势，创造出全新的数字内容生成和互动形式。随着技术的进步，人工智能写作、人工智能作曲、人工智能视频生成、人工智能语音合成以及近期流行的人工智能绘画等，都在创意领域引起了广泛关注。只需输入几个关键词，即可在几秒钟内生成一幅画作。（来源：google.com）
斯坦福人工智能俱乐部：Jason Wei 谈 2025 年人工智能的三大关键理念
https://www.youtube.com/watch?v=b6Doq2fz81U
斯坦福人工智能俱乐部 2025 年 10 月 18 日
本次演讲是斯坦福人工智能俱乐部系列讲座的一部分。
Jason Wei 是一位杰出的人工智能研究员，目前就职于 Meta Superintelligence 实验室。加入 Meta 之前，Jason 曾参与 OpenAI 的 o1 模型和 Deep Research 产品的联合开发。他也是“思维链”推理的发明者之一，并在 Google Brain 记录了关于涌现现象的重要研究。
我的评论：
Jason Wei 在斯坦福的演讲告诉我们：
智能正在成为一种商品，人工智能的三大核心理念中的两个是：第一，计算和知识的成本正在接近于零；第二，人工智能不太可能实现快速发展。
根据麻省理工学院2025年7月的研究报告：
“尽管企业在生成式人工智能（GenAI）领域投入了300亿至400亿美元，但这份报告揭示了一个令人惊讶的结果：95%的组织机构的回报为零。无论是买家（企业、中型企业、中小企业）还是开发者（初创公司、供应商、咨询公司），其收益都存在着巨大的鸿沟，我们称之为‘生成式人工智能鸿沟’。只有5%的集成人工智能试点项目能够创造数百万美元的价值，而绝大多数项目仍然停滞不前，无法对损益表产生任何可衡量的影响。
“这种鸿沟似乎并非由模型质量或监管造成，而是由方法决定。
“ChatGPT 和 Copilot 等工具已被广泛采用。超过 80% 的组织已经探索或试用过这些工具，近 40% 的组织表示已部署。但这些工具主要提升的是个人生产力，而非损益表现。与此同时，企业级系统，无论是定制的还是供应商提供的，都正被悄然拒绝。60% 的组织评估过此类工具，但只有 20% 进入了试点阶段，仅有 5% 进入了生产阶段。大多数失败的原因在于工作流程脆弱、缺乏情境学习以及与日常运营不匹配。” [https://mlq.ai/media/quarterly_decks/v0.1_State_of_AI_in_Business_2025_Report.pdf]
我的评论：
该报告没有对进入生产阶段的 5% 和失败的 95% 的人力利用率进行比较。很可能，人力利用率最高的系统在盈利能力方面也最成功，因为只有活的劳动力才能创造可供资本提取的剩余价值或利润。
人工智能的优势在于，它可以通过提高生产力来简化工人剩余价值的榨取过程。但另一方面，劳动生产率越高，失业人数就越多，剩余价值和利润也就越少。因此，人工智能资本主义陷入了自我造成的不可调和的矛盾之中，注定会走向失败。
摘自弗兰克·兰迪莫尔 (Frank Landymore) 的文章《秘密计划揭露亚马逊用机器人大军取代 60 万工人的阴谋》
https://futurism.com/category/artificial-intelligence 发布于 2025 年 10 月 23 日下午 2:10（美国东部时间）
*
在罗斯·杜特 (Ross Douthat) 主持的《有趣的时代》节目中，讨论的要点如下：
人工智能革命是否支撑着整个经济？本周的“有趣时代”节目中，罗斯与哈佛肯尼迪学院的经济学家、纽约时报评论版撰稿人杰森·弗曼探讨了投资者、政策制定者和消费者应该如何看待这波经济繁荣及其潜在的冲击。——美国经济增长最快的领域之一，并从过去的泡沫中寻找答案。
https://www.youtube.com/watch?v=KoJH9cGtv1s
上世纪90年代末，投资者将大量资金投入互联网公司。电信公司铺设了数千万英里的光纤电缆等基础设施，以支持预期的繁荣。
这是一个泡沫。利润从未实现，投资者纷纷抛售股票，公司也随之倒闭。纳斯达克指数市值缩水超过四分之三，花了15年才恢复过来。
这一次，涉事公司的规模要大得多。然而，与它们的互联网先驱不同，它们现在确实在盈利。只是——就目前而言——盈利还不如投资者预期的那么多。（《纽约时报》，2025年10月27日）
埃里克·李，《纽约时报》
英伟达市值现已达5万亿美元，巩固了其在人工智能领域的统治地位。轰动一时！
这家人工智能芯片制造商已成为特朗普政府在亚洲贸易谈判中的关键人物，也是首家市值突破5万亿美元的上市公司。
我的总体评论：
人工智能已经走到了一个十字路口——它被大肆宣传的努力已经暴露出其灾难性的缺陷，注定使其陷入多年的困境，因为其由大预言模型（LLM）产生的幻觉无法仅靠计算消除。人工智能的生存完全依赖于人类广泛且不可或缺的参与。在实现任何有意义的应用之前，必须由人类安全地消除其产生的虚假和捏造的危险产物。这一必要性适用于所有人工智能应用，包括资本主义和非资本主义应用。前者可能会利用第三世界远低于美国的劳动力成本，雇用海外工人来完成工作以节省资本支出，而这些以人工智能为中心的国家的工人将面临严重的失业困境。
正如约瑟夫·熊彼特在 20 世纪 20 年代所宣称的那样，资本在取得任何成就之前都必须先造成破坏，人工智能将成为资本主义“创造性破坏”的典型例子，而人类智能(HI)将取代它。
另一方面，令人欣慰的是，埃隆·马斯克推出了人工智能生成的“Grokipedia”，挑战维基百科——你需要了解什么
https://www.youtube.com/watch?v=GeSjqnn0im0&list=RDNSGeSjqnn0im0&start_radio=1
福布斯突发新闻 2025年10月28日
周一晚间，埃隆·马斯克发布了“Grokipedia”，这是一个完全由人工智能生成文章的在线百科全书，旨在挑战维基百科。马斯克及其盟友声称维基百科存在偏见且“觉醒”。
这就是我所设想的人工智能应有的样子——基于机器进化的智能百科全书（S.E.B.O.M.E. 或 Smart Encyclopedia Based On Machine Evolution，即赛博梅）。
关于基于机器进化的智能百科全书（S.E.B.O.M.E.），请访问：https://blog.creaders.net/user_blog_diary.php?did=NTExNjgz
以及
https://blog.creaders.net/user_blog_diary.php?did=NTEyMDM0
https://www.facebook.com/andrew.colesville
https://x.com/mwsansculotte
Reply from Chat.DeepSeek.com about S.E.B.O.M.E. (bilingual 双语）与深度求索对话）关于 S.E.B.O.M.E. （赛博梅）的回复
For Smart Encyclopedia Based On Machine Evolution (S.E.B.O.M.E.), See https://blog.creaders.net/user_blog_diary.php?did=NTExNjgz
https://blog.creaders.net/user_blog_diary.php?did=NTEyMDM0
https://www.facebook.com/andrew.colesville
https://x.com/mwsansculotte
【Mark Wain 2025年十一月一日】