公开课程
美国限制AI出口，中国却敞开大门——这剧情反转也太快了吧？ News English
重播当前句 0.75x 1x 1.25x 隐藏中文
access noun 访问权；进入 model noun 模型 available adjective 可获得的 consequences noun 后果 paradox noun 悖论 trade noun 交易；行业 speeding up verb phrase 加速 frontier 待补充词义 cybersecurity 待补充词义 heating up verb phrase 升温；加剧 opening noun 机会；缺口 workload noun 工作负载 shift verb 转移 open source adjective 开源的 restrict verb 限制 backfire verb 适得其反 block verb 阻止；封锁 developer noun 开发者 position verb 定位；安置 alternative noun 替代方案 ecosystem 待补充词义 risk noun 风险 foundation noun 基础 default adjective 默认的 influencer noun 影响者 standard noun 标准 national security noun phrase 国家安全 counter verb 对抗；反击 capable adjective 有能力的 pursue verb 追求；从事 incentivized adjective 受激励的 complicated adjective 复杂的 It's 待补充词义 well-established 待补充词义 Anthropic 待补充词义 matched 待补充词义 comparable to mythos 待补充词义 bug-finding 待补充词义 ironically 待补充词义 company 待补充词义 kind of frustrating 待补充词义 know 待补充词义 themselves 待补充词义 approach 待补充词义 thought 待补充词义 talk to says 待补充词义 place 待补充词义 good 待补充词义 actually 待补充词义 hearing 待补充词义 that's 待补充词义 willing to build 待补充词义 Reflection 待补充词义 mentioned 待补充词义 happened 待补充词义 hard to monetize 待补充词义 sort of strategic 待补充词义 treats 待补充词义 ambition 待补充词义 business 待补充词义 charging 待补充词义 charging for usage 待补充词义 However 待补充词义 Chinese 待补充词义 getting 待补充词义 companies 待补充词义 decision 待补充词义 making 待补充词义 enough 待补充词义 really 待补充词义 going 待补充词义 models out 待补充词义 leading 待补充词义 almost 待补充词义 talk to they 待补充词义 needs to take 待补充词义 sort of been 待补充词义 seriously 待补充词义 Typically 待补充词义 America 待补充词义 left up 待补充词义 Universities 待补充词义 happening 待补充词义 There's 待补充词义 funding 待补充词义 Welcome back to The Exchange. Just as the U.S. is trying to slow access to American AI models, 欢迎回到《交易所》。正当美国试图减缓对美式AI模型的访问时， China is making them more available, and that could have big consequences. Deirdre Bosa has more in today's Tech Check. Deirdre? 中国却在让这些模型更易获取，这可能带来重大影响。迪尔德丽·博萨将在今天的科技前沿节目中带来更多信息。迪尔德丽？ Hey, Kelly. So this is really the paradox at the heart of the AI trade right now. America is 嘿，凯利。这确实是当下AI交易的核心矛盾。美国正在 slowing just as China is speeding up. And now there are signs that Chinese labs are no longer competing on cost. They may actually be reaching the frontier in some of the most sensitive AI 放缓，而中国在加速。现在有迹象表明，中国实验室不再仅靠成本竞争。他们可能在某些最敏感的AI能力上逼近前沿， capabilities like cybersecurity. So the policy fight here, it is likely just heating up, limiting access to anthropics and open AI's most powerful models. That has given Chinese 比如网络安全。因此，这里的政策斗争可能才刚刚升温，限制了对Anthropic和OpenAI最强大模型的访问。这给了中国 labs an opening. And we have seen a number of companies like Coinbase, Airbnb, Shopify, 实验室一个突破口。我们看到许多公司，如Coinbase、Airbnb、Shopify， shift at least some of their workloads to these open source Chinese models. And now people I talk to say that the next focus in Washington, it could be the Chinese models 将部分工作负载转移到这些中国开源模型上。现在与我交谈的人说，华盛顿的下一个焦点可能正是中国模型本身， themselves, including a decision whether to restrict or even ban them. But that could backfire 包括是否限制甚至禁止它们的决定。但这可能产生反效果， if the Chinese government is not able to do that. So I think it's a good idea to have a U.S. blocks Chinese models at home. That will not stop developers, companies, governments outside 如果美国政府无法做到这一点。所以我认为，如果美国在国内禁止中国模型，那将无法阻止美国以外的开发者、公司、政府 of the U.S. from building on them. In fact, it could push China to move even faster and position 在此基础上构建。事实上，这可能推动中国更快行动，并将自身定位 itself as the alternative to U.S. controlled AI. And so the ecosystem is the bigger risk here. 为美国控制AI的替代者。所以生态系统是这里更大的风险。 If Chinese open source models become the foundation, the way that Android became the default mobile operating system for much of the world, then China doesn't just get users, 如果中国开源模型成为基础，就像Android成为全球大部分地区默认的移动操作系统那样，那么中国获得的不仅仅是用户， but it gets influencers, say, in the standards, the defaults, the rules of the next AI stack, 还会获得影响力，比如在标准、默认设置、下一套AI系统的规则方面， or system. So, guys, the question for Washington, it's whether it can protect national security without handing the next AI ecosystem to China. And for American AI labs, Kelly, the question is, 或者说系统。所以，各位，华盛顿的问题在于，它能否在不将下一个AI生态系统拱手让给中国的情况下保护国家安全。而对于美国的AI实验室，凯利，问题是， will they go open source to counter these very capable models out of China like GLM 5.2? 他们是否会选择开源，以对抗像GLM 5.2这样非常有能力的中国模型？ Right now, it is just NVIDIA pursuing this route with Nemotron because it is incentivized to do so. 目前，只有英伟达在通过Nemotron走这条路，因为它有这样的动力。 The major AI labs, less incentivized to do so. It's complicated. We have the journal saying 主要的AI实验室动力较小。情况很复杂。我们此前有报道称 that China has matched the AI system. It's not just the AI system. It's the AI system. It's a very well-established anthropic in cybersecurity, Deirdre. You know, with this 待补充翻译：that China has matched the AI system. It's not just the AI system. It's the AI system. It's a very well-established anthropic in cybersecurity, Deirdre. You know, with this company, 360 Security, their new bug-finding tool called Tulong Fang, comparable to mythos in finding bugs. You know, ironically, the Chinese stock market is not even 待补充翻译：company, 360 Security, their new bug-finding tool called Tulong Fang, comparable to mythos in finding bugs. You know, ironically, the Chinese stock market is not even doing that well, which is kind of frustrating people who thought that might be a great play on this. But in terms of the models themselves, you know, I don't know what a better U.S. approach 待补充翻译：doing that well, which is kind of frustrating people who thought that might be a great play on this. But in terms of the models themselves, you know, I don't know what a better U.S. approach would be here. What are you hearing? So, you know, for you to say that China is a good place to be, I think that's a good place to be. years, actually, a lot of the people I talk to says that the U.S. must pursue open source. We 待补充翻译：would be here. What are you hearing? So, you know, for you to say that China is a good place to be, I think that's a good place to be. years, actually, a lot of the people I talk to says that the U.S. must pursue open source. We must have an ecosystem that people are willing to build on. That just hasn't happened. There's NVIDIA, as I mentioned. Reflection AI is another company that is looking at this. And the problem 待补充翻译：must have an ecosystem that people are willing to build on. That just hasn't happened. There's NVIDIA, as I mentioned. Reflection AI is another company that is looking at this. And the problem is that open source is hard to monetize. So China treats this as sort of a strategic and 待补充翻译：is that open source is hard to monetize. So China treats this as sort of a strategic and national ambition to have the world build on their open source models, whereas, you know, our frontier labs, OpenAI and Anthropic, their business model depends on charging for their APIs, 待补充翻译：national ambition to have the world build on their open source models, whereas, you know, our frontier labs, OpenAI and Anthropic, their business model depends on charging for their APIs, charging for usage of the best AI that there is. However, the Chinese now are getting close. So 待补充翻译：charging for usage of the best AI that there is. However, the Chinese now are getting close. So a lot of companies are making that decision. We are going to use AI that is good enough or really 待补充翻译：a lot of companies are making that decision. We are going to use AI that is good enough or really almost as good as the leading models out of the U.S. And that has sort of put us in this place. 待补充翻译：almost as good as the leading models out of the U.S. And that has sort of put us in this place. So a lot of the folks I talk to, they still say that America needs to take open source seriously. Typically, that has sort of been in the public sphere, right? 待补充翻译：So a lot of the folks I talk to, they still say that America needs to take open source seriously. Typically, that has sort of been in the public sphere, right? Universities. But that's not happening right now. There's funding issues. So it's left up to companies. And we'll see how that evolves. 待补充翻译：Universities. But that's not happening right now. There's funding issues. So it's left up to companies. And we'll see how that evolves.
课程说明与学习重点 2026-07-09｜中美AI竞速：美方限制出口 中方加速开源 全球生态面临重塑
本课适合谁
中高级英语学习者（B2及以上）
对科技、国际关系感兴趣的学习者
希望掌握经济新闻词汇和逻辑的学习者
你会学到什么
理解并区分中美两国在AI领域的政策差异
掌握经济新闻中常见的矛盾、影响类表达
学会分析技术出口管制对全球产业链的潜在影响
重点表达
slow access to American AI models：减缓对美式AI模型的访问（这是美国策略的核心动作，通过限制技术出口延缓他国发展。'slow access' 比简单的 'restrict' 更强调渐进性，值得模仿。）
make them more available：让这些模型更易获取（与中国模式形成直接对比，'more available' 体现了开放与抵制的对立，是理解新闻脉络的关键短语。）
the paradox at the heart of the AI trade：AI交易的核心矛盾（paradox' 一词点出了看似矛盾却真实存在的现象，是新闻评论常用的修辞手法，有助于提升语言深度。）
big consequences：重大影响（泛用于预示结果，强调事态严峻性，在听力中常作为过渡信号词出现。）
听力难点
快速识别 'while' 引导的对比结构，尤其是前后主语和动词的切换
paradox at the heart of the AI trade' 中的连读与弱读可能导致理解困难
掌握 'access'、'available'、'consequences' 等核心词在不同语境中的准确含义
中文背景说明
自2022年起，美国商务部对先进AI芯片和模型实施出口管制，重点限制NVIDIA等企业向中国销售高端GPU。
中国科技企业如百度、字节跳动、阿里等加速了开源大模型的发布，如文心一言、通义千问等，意图通过开源生态吸引全球开发者。
全球AI生态可能因此出现分裂：一方依赖美国闭源模型，另一方转向中国开源平台，影响国际技术标准和合作框架。