Could China Topple America's AI Throne?

中国能否颠覆美国的人工智能霸主地位？

Helen Toner, former OpenAI board member and director of strategy at Georgetown Center for Security and Emerging Technology, says Chinese AI models are "getting plenty of traction" after the rise of DeepSeek. She shares her outlook on the tech rivalry between the US and China in an interview with Bloomberg's Haslinda Amin.

海伦·托纳曾担任OpenAI董事会成员，现任乔治城大学安全与新兴技术中心战略总监。她指出，随着深度求索（DeepSeek）等模型的崛起，中国AI模型正“获得大量关注”。在接受彭博社记者哈斯林达·阿明（Haslinda Amin）的采访时，她分享了对中美科技竞争态势的看法。

Recently, China's momentum has dominated headlines. Zippo. Asia is expanding into Southeast Asia and the Middle East, while Deep Sea and Minimax continue to roll out models. They say rival American counterparts despite ongoing U.S. chip restrictions. OpenAI said in a blog post last week that state-backed Zippo. AI has made notable progress in getting several governments across the world to adopt its AI models. And the tech rivalry just saw a new development today. Siemens says the U.S. has lifted export curbs on chip design software for China, restoring access for Chinese clients. That move comes weeks after Washington had ordered electronic design automation firms to halt shipments, sparking backlash from Beiing.

最近，中国在人工智能领域的发展势头频频登上头条。Zippo.Asia 正加速向东南亚和中东市场扩张，而深度求索（Deep Sea）和 Minimax 也在持续推出新型模型。尽管美国仍在实施芯片限制，但这些中国企业的表现已对美国同行构成竞争。上周，OpenAI 在一篇博客文章中提到，由政府支持的 Zippo.AI 已在推动多个国家政府采用其人工智能模型方面取得显著进展。而就在今天，科技竞争格局又迎来新动态 —— 西门子称美国已取消对中国芯片设计软件的出口限制，恢复中国客户的使用权限。这一举措距离美国政府此前要求电子设计自动化公司暂停对华出货仅过去数周，当时该禁令曾引发中方强烈反对。

Great to be here. Company has been pretty open about, you know, the impact of the likes of ZG. I am just wondering how much traction Chinese AI models are making right now.

They're certainly getting plenty of traction and it's anyone's game and an important thing to understand as well about how AI development is progressing is that there are really not just one AI competition or one AI race, as people sometimes say, but actually many. So there's a question of who is leading at the very cutting edge of the technology, making the most advanced models. There's different questions about who is leading in terms of adoption, in terms of B2B use cases, B2C use cases in different countries. And we're seeing Chinese companies certainly making a strong showing across many of those fields. Even if, you know, I would I would say that the US is still generally in the lead.

很高兴来到这里。大家对 ZG 等中国企业的影响力已有所耳闻，我现在想知道，中国 AI 模型目前的市场接受度究竟如何？

它们确实正在获得广泛关注，而且这场竞争并非一家独大。值得注意的是，当前 AI 发展的竞争并非像人们常说的 “一场 AI 竞赛”，而是多维度的较量。比如，谁在技术最前沿占据领先地位，能开发出最先进的模型；又比如，在不同国家的企业级应用（B2B）和消费者应用（B2C）场景中，谁的落地能力更强。我们看到中国企业在多个领域都展现出强劲势头，尽管我仍认为美国整体上仍处于领先地位。

But how quickly, though, are Chinese companies eroding China's leadership or rather U.S. leadership?

It really depends on where you look. So I think if you're looking at adoption across enterprise, then certainly in China, Chinese models have obviously a big advantage. Most of the US companies are not even attempting to go for Chinese adoption. On the flip side, if we're looking at US adoption, corporate adoption, customer adoption, then the US companies have a natural advantage there. Something that opening I said about Japan in their blog post recently was that Japan was really trying to make a play for those third party countries, third countries. So opening line has been on a big spree going across Southeast Asia and other countries, looking to sell their products, looking to set up what they call Opening Eye for countries initiative. So partnering with specific governments and what it sounds like is they have heard from those governments as they've been touring the area, that if you pull this Chinese company, one of the Chinese six little dragons that gets called that, they're also going and having those conversations with governments trying to come in and say, look, we can give you our full stack solution, we can give you white glove service to get not just, you know, a specific like a broad based product, but get something specific to your needs. And so what it sounds like is that Japan was a company in addition to other companies that we know from the Chinese side are trying to do this like Alibaba, Huawei, that they are they are starting to show up in these conversations as well.

但中国企业究竟以多快的速度在削弱 —— 或者说，美国的领先地位正在以多快的速度被中国企业追赶？

这完全取决于观察的领域。如果关注企业级应用的落地情况，中国本土模型显然占据巨大优势，多数美国公司甚至从未尝试进入中国市场。但反过来看美国本土的企业应用和消费者市场，美国公司又拥有天然优势。值得注意的是，OpenAI 近期在博客中提到，日本企业正积极争夺第三方国家市场。OpenAI 自身也在大举进军东南亚等地区，推行所谓的 “国家 AI 计划”，寻求与特定政府合作。他们在走访这些地区时发现，当提及中国企业（即所谓的 “中国六小龙”）时，当地政府也在与这些中企接触。中企会向政府提议：“我们能提供全栈解决方案，提供定制化服务，不仅是通用型产品，更能满足你们的特定需求。” 除了阿里巴巴、华为等知名中企，日本的一些企业也开始参与到这类合作洽谈中。

Hmm. How long? Talk to us about the countries companies already adopting the Chinese models because we hear the adoption is pretty much throughout the world, including the Middle East, Asia and Europe as well.

我们听说中国 AI 模型在全球多个地区都有应用，包括中东、亚洲和欧洲，能否谈谈哪些国家或企业已开始采用中国模型？

So what we're seeing, I think, is a lot of experimentation. It's difficult to know how much actual adoption and how much at scale used that we're seeing. So, for example, in China, in in the wake of this big deep SIK moment in January where that company really broke through and had a huge new cycle. There are a lot of stories about Chinese companies, Chinese, local and provincial governments starting to adopt deep seek, but it's really not clear if you kind of scratch beneath the surface. What does that look like? Does it look like they ran one trial? Does it look like some of their employees started using the app or did it really look like a deep partnership and really serious adoption? It's very difficult to get that in-depth data. One thing we can say is that if you look at, again, these open source models, actually a really big player there is Alibaba. So they have a model series called Q1 and that is very, very popular. So on even on US platforms or international platforms like Hugging Face is a very popular place to go to find models to use. They are showing that Alibaba's models are really very popular, outstrip any of the US models in terms of just sheer amount of use on that particular platform.

目前更多是试验性应用，实际落地规模和深度尚难量化。例如，今年 1 月深度求索（DeepSeek）取得重大突破并引发广泛关注后，有不少报道称中国本土企业和地方政府开始采用其模型，但深入来看，这些合作究竟是单次测试、员工试用，还是真正深度的合作伙伴关系，仍缺乏详细数据。不过有一点可以确定：在开源模型领域，阿里巴巴是重要参与者。其推出的 Q1 系列模型非常受欢迎，甚至在美国 Hugging Face 等国际平台上，阿里巴巴模型的使用量已超过任何美国模型，展现出惊人的市场接受度。

以下是各国网友的评论：

ianchu8232
America has the AI throne? It's the Chinese in America competing with the Chinese in China at this time.

美国还占着AI霸主地位呢？现在明明是在美国的华人在和中国本土的华人卷。

Iwantalloftheinformation
sounds like a possible prevention of ww3.

这听着像是能阻止第三次世界大战的苗头啊。

eugenec7130
You are so bad! Don't say that. The American politicians can very much freak out over your statement.

别这么说啊，美国政客听了你的话能吓破胆。

xushenxn
US models do not allow criticize Israel

美国的AI模型都不让批评以色列。

No1careabU
"Absolutely. China's AI development is powered by its manufacturing base, enabling a true Internet of Things with ten times more engineers than the US. America's ChatGPT has no industrial foundation—it's just building castles in the sky. And when a single US-made AI robot costs $200,000, who can afford that?"

“就是！中国AI发展有制造业打底，工程师数量是美国的十倍，这才能搞出真正的物联网。美国的ChatGPT根本没有工业基础，纯玩空中楼阁。而且美国造个AI机器人要20万美元，谁买得起啊？”

BLACKNETRESEARCH
OpenAI (Chat GPT) just got a $200 million contract with the military. The USA will not lose.

OpenAI刚和军方签了2亿美元的合同，美国不会输的。

df

China gives us quality service with minimum cost. And America provides nothing but imposing other countries to adopt their model. America always did that. American firms must be competitive in quality and price point. Threatening other countries is not going to work anymore

中国用最低成本给我们提供优质服务，美国除了强迫别国接受他们的模式啥也不干。美国一直这样。美国公司该在质量和价格上卷起来了，威胁别国这招现在不好使了。

Nom_AnorVSJedi
American AI companies are all staffed by Chinese AI researchers though.

可美国AI公司全是中国研究员在撑着啊。

passby8070
Absolutely! I am close follower of the development of AI industry, I think China is already on par with the best US in terms of model for capacity and it has far exceed the value metrics. China also is well ahead in smart devices and robotic adoption. China will disrupt and dominate the AI space like it has done to motor transport with its dominance with EVs and battery tech.

必须的！我是AI行业的死忠粉，我觉得中国模型的能力已经和美国顶尖水平持平了，性价比更是甩美国几条街。中国在智能设备和机器人应用上也遥遥领先，肯定会像用电动车和电池技术颠覆汽车行业那样，统治AI领域。

ashflownpv
STEM-Which country focuses on these areas and who has the sheer numbers in terms of students/grads/industry/military? China of course can do it and is currently executing this strategy as we speak

STEM领域哪家强？论学生、毕业生、行业人才和军方投入的数量，谁能比得过中国？中国当然能行，而且正在落实这个战略。

glenwjohnson809
China isn't racing with the US, it's the US that is racing with China.

不是中国在追美国，是美国在追中国。

Thailova
Deep seek is better than chat GPT

深度求索比ChatGPT好用。

lzasnw
We’re really still pretending censorship is a cna only problem?? Helen should try criticizing uncle sam and see if she still has a job tomorrow

我们还在装模作样说审查只是某国的问题吗？海伦要是敢批评美国政府，看她明天还能不能保住工作。

duydatyds
China has AI development at grassroots level, with vast ecosystems of infrastructures and manufacturers join the field, while the US corporations is just throwing money at it to attract talents (mostly from China) to build it. Needless to say who will win in the long terms.

中国AI是从基层开始发展的，有庞大的基础设施和制造商生态入局，美国公司就知道砸钱挖人才（还大多是从中国挖）。长期来看谁会赢还用说吗。

iamsheep
She doesn’t know the details because she can’t read Chinese.

她不了解细节，因为她看不懂中文。

samanthaw4955
Yes it will because they are a lot smarter and have more highly educated people to do this already

当然能颠覆，因为他们更聪明，而且已经有更多高学历人才投身AI了。

lambertgiang
Just Remember one thing, GPS, Space Technology, 3rd and 4th Generation Nuclear Power Reactors, Nearlink from Huawei, 8K standards in Broadcasting, DC catapult ejection on China's own design aircraft carriers, Chinese MRI using different approach all or most resulted from the non-participation allowed under the US and its Alliances. On top of that China has 2 million engineering graduates in comparison to US 200,000 yearly and China alone has more Graduates than the G7 Combined. Is it that hard to topple America's AI and half of the American AI engineer are from Asian Chinese background. Is it that hard to figure out?

记住一点：GPS、航天技术、三四代核电反应堆、华为的 Nearlink、8K 广播标准、中国自主设计航母的电磁弹射，还有中国用不同技术路线造的核磁共振仪，这些大多都是美国及其盟友搞技术封锁逼出来的。再说了，中国每年有200万工程专业毕业生，美国只有20万，中国一国的毕业生比G7加起来都多。颠覆美国AI很难吗？而且美国一半的AI工程师都是华裔，这还想不明白？

Penaming
The most important feature of open source AI is that OpenAI does NOT have access to the proprietary, confidential, secret data you are going to run thru the LLMs.

开源AI最重要的特点是：你用大语言模型处理的专有数据、机密数据、秘密数据，OpenAI根本接触不到。

Artificial_intelligence2-m8d
There is no throne, world needs more competition.

根本没有什么霸主地位，世界需要更多竞争。

alexma5848
Because DeepSeek，OpenAI became open AI.

因为深度求索，OpenAI才开始搞开源AI的。

mokulashi
they already win

他们已经赢了。

Luther84695
Plot Twist: China will win the AI race, but it is THE OTHER China.

反转来了：中国会赢得AI竞赛，但不是你们想的那个中国。

eanerickson8915
If you have the money, you can buy the talent. No matter how smart you are, nobody is going to work hard for peanuts.

有钱就能买到人才。再聪明的人，也不会为了一点点钱拼命干活。

FootFixAIASMR
China have something that hard to copy, they really have a lot talent. if gpu not restrict by america, the story maybe different

中国有难以复制的优势，真的有很多人才。要是美国不限制GPU，剧情可能就不一样了。

Wileytiger-el6pu
Nobody will win. It’s a long game. Perhaps one that lasts forever. They have a lot of talent but so do we and we have a lot more money and influence . Never underestimate the US

没人会赢，这是场持久战，可能永远不会有胜负。他们有很多人才，我们也有，而且我们钱更多、影响力更大，别低估我们美国。

AntOne-y7t
They already have

他们已经做到了。

Artificial_intelligence2-m8d
Legacy media still obsessing about Empires, thrones, colonies power centres and influences. The world has moved on

传统媒体还在纠结帝国、霸权、殖民中心和影响力，世界早都变了。

foosimon7537
when u can't win, just call for gov help to ban. I saw the signal of losing..

赢不了就找政府帮忙禁掉，我已经看到失败的信号了

gfan003
Meta recently snatched top AI specialists from google and open AI, but China produces plenty of top IT engineers that fill the roles in new start ups and big corporate labs. China will likely win as majority of AI specialists are Chinese.

Meta最近从谷歌和OpenAI挖走了顶尖AI专家，但中国能培养出大量顶尖IT工程师，填补初创公司和大企业实验室的岗位。因为大部分AI专家都是中国人，中国很可能会赢。

Octavius8080
AI rapidly reaching the point that nobody is going to be leading. I can hop on Gemini, ChatGPT, Perplexty etc. and get pretty decent answers from all of them. Sure if I want to do coding or high level scientific research I would go with a better trained model but 99% of people will never use those features. I have never asked ChatGPT to sequence a genome and I never will.

AI正在迅速进入无人领先的阶段。我用Gemini、ChatGPT、Perplexty这些模型都能得到不错的答案。当然，如果要编程或做高级科研，我会选训练更好的模型，但99%的人根本用不到这些功能。我从没让ChatGPT给基因组测序，以后也不会。

rohullahkarimi744
Chinese researchers in us laugh loudly Just look at meta latest talents that they recruit hahahah

在美国的中国研究员已经笑出声了，看看Meta最近招的人才就知道了哈哈哈。

KhoiNguyens
deepseek failed miserably in planning a recent trip to Dubai. never again!

深度求索最近规划迪拜旅行时惨败，再也不用了！

Akhanbharat123
Deep seek is not real times based . Use it where it is suitable for analysing studies . Not like a friend to chat like chat gpt

深度求索不是实时的，适合用来分析研究，别像ChatGPT那样当朋友聊天。

BruceJ999
You're the problem for using deepseek which is not real times based, even the app clearly states that its information is restricted to 2024. Don't blame Deepseek for being slow.

问题出在你自己用了非实时的深度求索，人家app都明说了信息截止到2024年，别嫌人家慢。

继续阅读：人工智能   美国

此文由 三泰虎 编辑，未经允许不得转载！：首页 > 资讯 » 美媒：中国能否颠覆美国的人工智能霸主地位