为什么中国在将像DeepSeek这样的AI应用快速集成到各种产品中，而美国的OpenAI在产品集成方面却毫无进展？-网帖翻译-龙腾网

Robert Quek
OpenAI is full of self-importance, being the first with the reasoning model, named ChatGPT o1. CEO Sam Altman tells investors that vast sums of money and computing power are needed to stay at the forefront of AI. It is about big investments for big returns, about dominance and control. He was the prime mover in the $500 billion AI fund.
It did not pay attention to applications, which it expects, would follow naturally given its leadership and its first-mover advantage. Competition in the US from Google Germini, Meta Llama, and Andropic Claude, was not intense enough to give it the extra kick.

OpenAI充满了自大，成为第一个推出推理模型的公司，命名为ChatGPT 01。首席执行官Sam Altman告诉投资者，保持在AI前沿需要巨额资金和计算能力。这关乎巨额投资和巨大的回报，关乎主导地位和控制。他是5000亿美元AI基金的主要推动者。
它没有关注应用程序，它认为，鉴于其领先地位和先发优势，应用程序自然会随之而来。在美国，来自 Google Germini、Meta Llama 和 Andropic Claude 的竞争还不够激烈，不足以给它带来额外的动力。

The industry’s strategy was to keep the market excited with incremental improvements of their models, and consolidate their gains domestically and overseas.
US deemed itself the world’s AI leader. China was a distant second. To make sure it can never catch up, it banned the export of advanced chips to China, the equipment to make them, and threatened non-US firms with penalties if they did not comply. Nvidia had to design special downgrades for the Chinese market.

该行业的战略是通过逐步改进其型号来保持市场的兴奋，并在国内外巩固其成果。
美国自认为是世界AI的领导者，中国则位居第二。为了确保中国永远无法赶上，美国禁止向中国出口先进芯片、制造这些芯片的设备，并威胁如果非美国公司不遵守，将会受到惩罚。英伟达不得不为中国市场设计特定的降级版本。

DeepSeek has a different origin. It is a spin-off from High-Flyer, a hedge fund, and its motivation is not purely commercial. Liang Wenfeng, the founder, made DeepSeek open-source, said he wanted to prevent the technology being monopolized by a few companies. He has democratized AI.
It spent only $5.6 million to train the v3 and R1 models. R1 comes with a set of smaller, cheaper, and faster “distilled” variants, for users with limited computing power. Not only was the training cost low, the cost of running it as well. Firms that make use of the models to create services are charged less than 1/10 the charges of the US models.

DeepSeek 有着不同的起源。它是 High-Flyer 对冲基金的衍生品，其动机并非纯粹的商业。创始人梁文峰表示，他将 DeepSeek 设为开源，希望防止这项技术被少数公司垄断。他实现了人工智能的民主化。它仅花费了 560 万美元来训练 v3 和 R1 模型。R1 提供了一组更小、更便宜、更快速的“精炼”版本，适用于计算能力有限的用户。训练成本不仅低，运行成本也很低。利用这些模型开发服务的公司，其收费不到美国模型的 1/10。

This is one reason for the rapid and widespread application and integration of the models. The other reason is the great publicity that DeepSeek generated on its release. Who can forget that the market value of US tech stocks plunged $1 trillion, the largest single-day loss in the history of Nasdaq.

这也是 DeepSeek 模型快速广泛应用和整合的原因之一。，另一个原因是DeepSeek一经发布就获得了极大的关注，谁能忘记当时美国科技股市值暴跌1万亿美元，创下纳斯达克历史上最大单日跌幅。

Therefore there is this rush by companies to adopt it in their systems and products - the telcos, auto companies, smartphones, all kinds of manufacturing, fund managements and consultancies, platform companies to integrate it into their own model, and even by the local governments, to enhance their services.

因此，各大公司纷纷在其系统和产品中采用区块链技术——电信公司、汽车公司、智能手机、各种制造业、基金管理和咨询公司、平台公司都将其整合到自己的模型中，甚至地方政府也将其纳入其中，以增强其服务。

But it does not have a free ride. Competition is intense. China has scores of AI companies. The more well known ones are Alibaba Qwen, Tencent Hunyuan, Baidu Ernie, ByteDance Doubao, and Kimi. A few days ago, Baidu released Ernie X1 and Ernie 4.5 to much acclaimed. Its stock price jumped. A bit earlier, Alibaba released Qwen 2.5, a multimodal that processes images and audio inputs. It beat DeepSeek in clarity, depth, reasoning, creativity, and transparency. They are all open-source.

但它并不是一路顺风。竞争非常激烈。中国有数十家人工智能公司。比较知名的有阿里巴巴的 Qwen、腾讯的 Hunyuan、百度的 Ernie、字节跳动的豆宝和 Kimi。几天前，百度发布了 Ernie X1 和 Ernie 4.5，受到了广泛好评，其股价也随之上涨。稍早前，阿里巴巴发布了 Qwen 2.5，一款处理图像和音频输入的多模态模型。它在清晰度、深度、推理、创造力和透明度方面超越了 DeepSeek。这些都是开源的。

You may therefor expect a lot more innovations and more applications and integrations going forward. There is in the current an AI industry boom in China.

因此，你可以预期未来会有更多的创新、更广泛的应用和整合。目前中国正处于人工智能产业的繁荣期。

Paul Denlinger
The US economy has become so thoroughly financialized by financial rent-seekers that the measure of success is no longer making products and services, but how much capital new companies can raise in their funding rounds.
OpenAI became so obsessed about raising funding to raise its company valuation that it lost sight of product development, and has now resorted to advising the Trump administration to block access to DeepSeek, OpenAI’s leading competitor.

美国经济已经被金融租金寻租者彻底金融化，成功的衡量标准不再是生产产品和提供服务，而是新公司在融资轮次中能够筹集多少资本。
OpenAI 太过痴迷于筹集资金以提高公司估值，以至于失去了对产品开发的关注，现在竟然转而建议特朗普政府建议封锁 DeepSeek 的访问，DeepSeek 是 OpenAI 的主要竞争对手。

When you can’t beat them, tell the government to block them because they are a national security threat to the US. Especially to US startups which have forgotten how to make products.
DeepSeek has the advantage of being in the country which is also the manufacturing powerhouse in the world, which produces electric vehicles, solar panels, space research and health devices, just to name a few. It was relatively easy for DeepSeek to win local traction, and after the Chinese government recognized their success, it went up from there.

当你打不过他们时，就告诉政府封锁他们，因为他们对美国构成了国家安全威胁，尤其是对那些忘记如何做产品的美国初创公司。
DeepSeek 拥有的优势是，它位于世界制造业强国——中国，生产电动汽车、太阳能电池板、航天研究和健康设备，仅举几例。DeepSeek 相对容易获得当地的支持，并且在中国政府认识到他们的成功后，事业也随之发展。
原创翻译：龙腾网 https://www.ltaaa.cn 转载请注明出处

The US doesn’t have any major manufacturers like BYD, DJI, CATL which can compete. Since there are no world-class manufacturers anymore, there is no way to integrate AI into their products.
It is just left with competing on fundraising, because that is all it can do.

美国没有像比亚迪、DJI、宁德时代这样的主要制造商能够竞争。由于现在没有世界级的制造商，根本无法将 AI 融入到他们的产品中。
美国只剩下在融资方面竞争，因为那是他们唯一能做的事情。

So what’s left?
I’m waiting for the next bubble from Silicon Valley and Wall Street.
Do you have any ideas?

那么接下来是什么呢？
我在等待硅谷和华尔街的下一个泡沫。
你有什么想法吗？

Eric
They key is synergy. Chinese entrepreneurs are open to share their core competence with a cooperative spirit for maximum synergy. One prominent example is that on September 13, 2023, Huawei and xiaomi reached a global patent cross-licensing agreement, resolving a patent dispute and covering multiple communications technologies, including 5G. They both flourished and kick American asses worlewide.
There is nothing open about OpenAI, a closed-source platform, because all American capitalists are territorial and selfish. They are on their own fighting battle royale to the last one standing. Good luck.

关键是协同效应。中国企业家愿意分享他们的核心竞争力，以合作精神实现最大化的协同效应。一个显著的例子是，2023年9月13日，华为和小米达成了全球专利交叉许可协议，解决了专利争议，涵盖了包括5G在内的多项通信技术。他们都繁荣发展，并在全球范围内击败了美国。
而 OpenAI 完全不是开放的，它是一个闭源平台，因为所有美国资本家都是领土性的，自私的。他们各自为战，最后一个站着的才算赢家。祝你好运。

Freddie
It is not only about AI, but mass of the new technology is fast integrated in all kinds of products and applications in China. Faster than most of other places.
The overall vibe of soceity is open and generous to the new technology. People like to give new things chance. And business will soon adopt new technology after found that it is really help to save cost and/or improve productive force.
There is so far no strong enough vested interest to block the road.
Like bank power is not strong enough to block the road for any epayment.

不仅仅是人工智能，中国大量新技术正在快速融入各类产品和应用，速度比其他大多数地方都要快。
社会整体氛围对新技术持开放和宽容的态度。人们愿意给新事物机会。企业在发现新技术确实有助于节省成本和/或提高生产力后，很快就会采用它。
目前还没有足够强大的既得利益集团来阻挡这条道路。
就像银行的力量还不足以阻挡任何电子支付的道路。

Karthik Bala

Open AI had a great breakthrough in 2022 with Chat GPT
They believed their product was so unique and so revolutionary that they set a massive price tag on it and proposed a model that could potentially lead to the next Skynet one day
The entire US Tech ecosystem envisioned billions of dollars into Data Centres, Powerful servers and Powerful Chips
When the investment is so high, it is obvious the applications where such AI is likely to be used would also be high revenue generating applications

OpenAI 在 2022 年取得了巨大的突破，推出了 ChatGPT。
他们相信自己的产品是如此独特且革命性，以至于给它定了一个高昂的价格，并提出了一个可能会导致未来某一天出现天网的模型。
整个美国科技生态系统预见到数十亿美元将投入数据中心、强大的服务器和强大的芯片。
当投资如此之高时，很明显，使用这种人工智能的应用也将是高收入的应用。

So OpenAI and it's vision of product integration was AI being used in a F-22 or a F-35 Or a brand new fighter bomber capable of being auto piloted by AI and capable of sexting and firing targets on its own
That level of complexity and that level of potential revenue generation to justify so many billions of dollars of investments

因此，OpenAI 及其产品整合的愿景是，人工智能将被用于 F-22 或 F-35，或者一款全新的战斗轰炸机，能够由人工智能自动驾驶，并能自主选择并打击目标。
这种程度的复杂性和潜在的收益水平证明了数十亿美元的投资是合理的

Deepseek came along and demonstrated that with a fraction of the investment, it was able to provide a functionality on par with ChatGPT
The Chinese Tech ecosystem envisioned integration of AI Models into other models, into products such as Toys, Robots, Drones and even Medical Equipment & integration with e commerce platforms, search engines and other basic low level applications used by 500 million people

Deepseek 的出现证明了，只需一小部分投资，它就能提供与 ChatGPT 相当的功能
中国科技生态系统设想将人工智能模型集成到其他模型中，应用于玩具、机器人、无人机，甚至医疗设备，并与电子商务平台、搜索引擎以及其他500百万用户使用的基础低级应用程序集成。

Much smaller level of application, much smaller level of revenue and returns but also a fraction of the huge investment that the Tech firms in the States have committed
This time the Chinese have triumphed because this method has led to better and better models emerging from China with more advanced applications such as Manus AI

这些应用的规模要小得多，收入和回报也更低，但相比美国科技公司承诺的巨大投资，所需的投资也是一小部分。
这一次，中国人取得了胜利，因为这种方法促使中国涌现出越来越优秀的模型，并产生了更多先进的应用，比如 Manus AI。

Meanwhile OpenAI with all those billions is struggling to evolve into the next iteration

与此同时，OpenAI 在投入了所有那些数十亿美元后，仍在努力发展到下一代。

Its like one side has invested into a huge, expensive Michelin style restaurant with the best chefs and most expensive dishes
Now they are waiting for the wealthy customers to arrive and start making money to justify all the expense by selling Noodles for 200 Dollars a plate

就像一方投资了一家巨大的、昂贵的米其林风格餐厅，拥有最好的厨师和最昂贵的菜肴，现在他们在等待富有的顾客前来，用200美元一盘的面条来开始赚钱，以证明所有的开支是值得的。

Meanwhile the other side has invested into small clean but inexpensive restaurants offering the same noodles with the same taste for 2.50 Dollars a plate and have opened to all members of the general public
With more and more customers, they have moved on to start making automated equipment to actually make noodles instead of relying on Chefs

与此同时，另一方投资了小而干净但价格低廉的餐厅，提供相同的面条，同样的口味，每盘 2.50 美元，并且向所有普通顾客开放。随着越来越多的顾客，他们开始制作自动化设备，实际上用机器来做面条，而不是依赖厨师。

—
OpenAI need to start from scratch on their financial modelling and this means saying goodbye to the heavy investments already made into the original vision of OpenAI

OpenAI 需要从头开始重新进行财务建模，这意味着要告别已经投入到 OpenAI 原始愿景中的大量投资。

Bethany Beaulac
China’s got the combo of a government that shoves tech down everyone’s throat and companies that sprint to kiss bureaucratic ring. No fuss about “ethics” or “privacy” clogging the pipes—just slap AI on everything, fix the bugs later. Meanwhile, the U.S. is too busy arguing if ChatGPT’s gonna write hate speech or whatever. OpenAI’s stuck in lawsuit purgatory, waiting for some committee to approve their app upxe. Also, Chinese consumers? They’ll beta-test your glitchy AI toilet paper dispenser if it scans their face first. Over here, everyone’s allergic to data harvesting unless it’s disguised as “personalized ads.” Different vibes, I guess.

中国有着一种政府强迫技术普及的模式，同时企业也争先恐后地去迎合官僚的要求。没有什么“伦理”或“隐私”之类的麻烦阻碍——只要把 AI 应用到一切上，先把问题解决再说。与此同时，美国则忙着争论 ChatGPT 会不会写仇恨言论之类的东西。OpenAI 被困在诉讼的泥潭中，等待某个委员会批准他们的应用更新。此外，中国的消费者？只要能扫描他们的脸，他们会帮你测试那些有故障的 AI 卫生纸分配器。而在这里，大家对数据收集过敏，除非它被伪装成“个性化广告”。气氛完全不同我猜。
原创翻译：龙腾网 https://www.ltaaa.cn 转载请注明出处

Kakan Chandra Dey
China's rapid integration of AI applications like DeepSeek into various products can be attributed to several factors. First, China's approach to AI development is heavily backed by government support, with significant investment in research, infrastructure, and AI startups. The government has a more centralized approach that allows for faster implementation of AI technologies into consumer products, often with fewer regulatory hurdles. In addition, China has a large and diverse consumer base, providing ample data for AI development and integration. Companies like Tencent, Alibaba, and Baidu also benefit from a more flexible regulatory environment, enabling them to deploy AI products more quickly and at scale.

中国将 AI 应用，如 DeepSeek，迅速融入各种产品的原因可以归结为几个因素。首先，中国在 AI 开发方面得到了政府的大力支持，政府在研究、基础设施和 AI 初创企业方面进行了大量投资。政府采取的是一种更加集中的方式，这使得 AI 技术能够更快地融入消费产品，并且通常面临的监管障碍较少。此外，中国拥有庞大且多样化的消费者基础，为 AI 的开发和整合提供了充足的数据。像腾讯、阿里巴巴和百度等公司也受益于更加灵活的监管环境，使得它们能够更快、更大规模地部署 AI 产品。

In contrast, while the US's OpenAI has made significant advancements in AI, its focus has primarily been on developing large language models (like GPT) and other foundational technologies. OpenAI’s pace of product integration is likely slower because of its more cautious approach to commercialization, ensuring safety, ethical considerations, and addressing potential misuse of AI. Additionally, OpenAI's structure, as a research-oriented organization, has led to more gradual implementation in products, as it partners with companies like Microsoft for broader deployment.

相比之下，尽管美国的 OpenAI 在人工智能领域取得了显著进展，但其主要关注的是开发大型语言模型（如 GPT）和其他基础技术。OpenAI 在产品整合方面的速度可能较慢，因为它采取了更为谨慎的商业化策略，确保安全性、伦理考虑，并应对人工智能潜在的滥用问题。此外，作为一个以研究为导向的组织，OpenAI 的结构导致其产品的实施较为渐进，因为它与微软等公司合作进行更广泛的部署。

As for Tencent Music Entertainment, as of their latest report, they had around 81 million paying users. Keep in mind that these numbers could change over time, so it's a good idea to check the most recent financial reports for the latest upxes.

至于腾讯音乐娱乐，根据他们最新的报告，他们大约有 8100 万付费用户。请注意，这些数字可能会随时间变化，因此最好查看最新的财务报告以获取最新信息。

MaryBeth VanderMeulen
Open AI was original in terms of producing a product before other tech companies got to market. The model they used that allowed everyone to try it gave them enough information to know where they needed to make changes which led to new upxed versions.

OpenAI 在其他科技公司进入市场之前就已经推出了产品，这一点非常具有原创性。他们采用的模型让每个人都可以尝试，从中获得了足够的信息来了解需要在哪些方面进行调整，从而推出了新的更新版本。

Deepseek did not come along for some period so it had plenty of time to take from what was already created and free to use. They could then tweak it to conform to their needs and not have the expense of an original concept. They made it to handle the needs of their own users.
Open AI had many upxes you will recall so there will not be any newer versions since that would mean a different model using LLM’s. I would assume they are also working on that.

Deepseek 的出现时间并不长，因此它有充足的时间从已经创建并免费使用的内容中汲取经验。然后他们可以对其进行调整以满足自己的需求，而无需花费原始概念的成本。他们制作它是为了满足自己用户的需求。
你会记得，Open AI 有很多更新，所以不会有任何更新版本，因为这意味着使用 LLM 的模型不同。我猜他们也在做这件事。
原创翻译：龙腾网 https://www.ltaaa.cn 转载请注明出处

"Jason Bulsa" Camp YouTube JBULSA
Press optics. OpenAi is evols in a group. DeepSeek is for the group. Evols: ww2 war continues: philosophy of com...sts vs. SOCIALISTS. group profits vs. Profits for the group

媒体公关效果。OpenAI 是一个群体中的演变。DeepSeek 是为群体服务的。Evols：第二次世界大战仍在继续：GCZY者与社会主义者的哲学。集团利润与为集团获取的利润。

Aqua Platycodon
Big businesses and small businesses operate very differently. DeepSeek is like a street vendor giving away free lemonade. Set up a stand, and people will come and grab a cup—it gets attention, but it’s not necessarily profitable.

大公司和小公司运作方式非常不同。DeepSeek 就像街头小贩免费赠送柠檬水。摆个摊，人们就会过来喝一杯——这会引起关注，但不一定能赚钱。

Large companies don’t work that way. Sometimes they stay quiet for years, then make a single move that dominates the market. OpenAI’s enterprise customers aren’t the type to make noise—they don’t need to brag about using AI. And individual paying users? Most don’t go around announcing what they’re subscribed to.

大公司不是这样运作的。有时他们会保持沉默多年，然后采取一个主导市场的举措。OpenAI 的企业客户不是那种喜欢制造噪音的人——他们不需要吹嘘自己使用人工智能。那么个人付费用户呢？大多数人不会到处宣布他们订阅了什么。
原创翻译：龙腾网 https://www.ltaaa.cn 转载请注明出处

So, from the outside, it might seem like OpenAI isn’t doing much. But if you want to know whether they’re actually making money, just look at their financial reports. The problem is, most people don’t read reports—they just follow headlines, or worse, internet rumors.
In that kind of environment, DeepSeek’s noise level might be high, but that doesn’t mean much.

因此，从外部来看，OpenAI 似乎没有做太多事情。但如果你想知道他们是否真的在赚钱，只需看看他们的财务报告。问题是，大多数人不看报告——他们只是追随头条新闻，或者更糟的是，追随网络谣言。在那种环境下，DeepSeek 的噪音水平可能很高，但这并不意味着什么。

The Fierce Independent
You should learn facts. DeepSeek isn’t in any known products, while OpenAI is now powering more of Apple’s Intelligence platform.

你应该了解事实。DeepSeek 没有出现在任何已知的产品中，而 OpenAI 现在正在为苹果的智能平台提供支持。