M11:16
不过，把这些先放一边，让我们来谈谈AI创造世界。
With that out of the way though, let's talk about AI creating worlds.
欢迎回到《AI Daily Brief》头条版，每天大约5分钟，带你了解所有你需要的人工智能新闻。
Welcome back to the AI Daily Brief Headlines Edition, all the daily AI news you need in around 5 minutes.
Anthropic正在加大对生物技术的投入，收购了一家非常小的公司。
Anthropic is doubling down on biotech with a big acquisition of a very small company.
周二，Anthropic发布了一篇博客文章，详细阐述了他们用人工智能增强药物发现的方法。
On Tuesday, Anthropic published a blog post talking through their approach to AI-enhanced drug discovery.
他们写道：“开发一种药物涉及许多不同的任务：进行早期研究以确定药物的靶点和合适的药物类型，设计具有正确特性的分子，优化其以便大规模可靠生产，起草开展临床试验所需的监管文件，设计和管理这些试验，以及更多。”
They wrote, "Developing a drug involves many varied tasks: doing early research to identify a target for a drug and the right type of drug to engage it, designing a molecule with the right properties, optimizing it to be manufactured reliably at scale, drafting the regulatory filings needed to run clinical trials, designing and administering those trials, and much more."
我们实现AI医疗效益的战略是基于让AI越来越多地参与所有这些任务。
Our strategy to realize AI's medical benefits is based on having AI increasingly involved in all of these tasks.
没有一个单一的问题需要解决。
There is no single problem to solve.
瓶颈分布在药物开发过程的各个阶段。
The bottlenecks are distributed all throughout the drug development process.
识别并解决这些瓶颈的最佳方式是深入参与真实的药物项目，涵盖许多不同的疾病领域和开发阶段。
The best way to identify and work through these bottlenecks is to get deeply involved in real drug programs, across many different disease areas and stages of development.
为此，Anthropic表示他们已经与6家组织合作，并启动了自家的药物发现工作。
To that end, Anthropic said that they've been partnering with 6 organizations as well as launching drug discovery efforts of their own.
他们附上了几个案例研究，展示Claude如何加速了开发有前景的药物候选所需的科学和运营工作。
They included several case studies showing how Claude had accelerated both the scientific and operational work required to develop promising drug candidates.
此外，他们宣布与Bristol-Myers Squibb建立新的合作伙伴关系，利用AI解决瓶颈并加强早期临床试验的规划。
In addition, they announced a new partnership with Bristol-Myers Squibb to use AI to address bottlenecks and enhance planning in early-stage clinical trials.
但最引人注目的消息是另一项生物技术收购。
But the news that caught everyone's attention was another biotech acquisition.
他们写道：“我们收购了Mithril Medicine，这是一家一人公司，一直在为2种罕见的肌营养不良症开发治疗方法。”
They wrote, "We've acquired Mithril Medicine, a one-person company that has been developing treatments for 2 rare forms of muscular dystrophy."
其创始人Cody Tranbarger利用Claude来组织合同、物流以及与11家外部实验室的实验，并将Mithril运营的许多其他方面委托给Claude。
Its founder, Cody Tranbarger, used Claude to organize contracts, logistics, and experiments with 11 external labs, and delegate many other aspects of Mithril's operations to Claude.
结果，Mithril在不到18个月的时间里，将一种可能的肌营养不良症治疗方法从设想推进到在小鼠中取得有希望的结果。
As a result, Mithril took a possible treatment for one form of muscular dystrophy from an idea to promising results in mice in less than 18 months.
Anthropic希望在2028年底前准备好一种药物用于临床试验，这将显著加速通常需要4到7年才能准备新药进行人体测试的时间线。
Anthropic hopes to have a drug ready for clinical trials by the end of 2028, and this would be a significant acceleration of the typical 4 to 7 year timeframe for preparing a new drug for human testing.
这主要归功于药物发现阶段的大幅缩减。
Largely due to a big reduction in the drug discovery phase.
最让许多人印象深刻的是一个人创办生物技术初创公司的可行性。
The thing that stood out most to many folks was the feasibility of a single-person biotech startup.
新闻聚合者Andrew Curran评论说，Sam Altman关于一人公司的预测正在成真。
News aggregator Andrew Curran commented, Sam Altman's one-person company prediction is coming true.
Anthropic刚刚收购了一家这样的公司。
Anthropic just bought one.
Mithril Medicine的创始人利用Claude，仅靠他自己、Claude和兼职顾问，在18个月内将一种肌营养不良症抗体治疗从概念推进到在小鼠中证明原理。
The founder of Mithril Medicine used Claude to take a muscular dystrophy antibody treatment from concept to proof of principle in mice in 18 months, with just himself, Claude, and part-time consultants.
真正的一人十亿美元公司。
The real one-person billion-dollar company.
一个聪明的、高自主性的单人CEO加上AI，可以对世界产生超乎寻常的影响。
A solo smart high-agency CEO and AI can have an outsized impact on the world.
不过话说回来，也不是所有人都买账。
Then again, not everyone is bought in yet.
激进做空机构Scorpion Capital写道，Mithril Medicine的创始人，一家被Anthropic收购的一人制药公司，并非我们以为的科学家或学者，而是哈佛MBA，前高盛银行业分析师，以及前对冲基金人士。
Activist short seller Scorpion Capital wrote, founder of Mithril Medicine, one-person drug company acquired by Anthropic, is not a scientist or academic as we assumed, but a Harvard MBA, ex-Goldman Sachs banking analyst, and ex-hedge fund guy.
不过，也有人指出这位创始人拥有生物技术硕士学位，意味着他们对科学并非完全陌生，但多半这更像一个罗夏测试，看你是认为这是专业知识的消亡，还是民主化带来的更多创新。
Now, some did note that the founder has a master's in biotech, meaning they are not completely foreign to the science, but mostly this is a Rorschach test for whether you see the death of expertise or a demonstration of democratization that could lead to more innovation.
其他人则质疑一人AI公司这个前提本身。
Others question the entire premise of the one-person AI company.
Peter Yang评论说，说实话，有些这种一人公司在我看来非常可疑。
Peter Yang commented, some of these one-person companies look highly suspicious to me, to be honest.
Lenny's Podcast的Lenny Rachitsky也同意，补充道，我非常怀疑一个人能撑起一家大规模公司。
Lenny Rachitsky from Lenny's Podcast agreed, adding, I'm very skeptical of a one-person company reaching massive scale.
光是客服时间就让人无法应付，除非你不算那些帮你的外包人员。
Just the support time alone makes it impossible to handle, unless you don't count contractors helping you with support.
别指望AI能替一人公司解决这个问题。
Don't think AI will ever solve this for a one-person company.
在X上其他地方，有人把这看作Anthropic在IPO前的一次非常精明的营销策略。
Elsewhere on X, people viewed this as a very shrewd marketing move from Anthropic ahead of their IPO.
从根本上说，这是重申他们的论点，即AI能极大加速药物研发，并让低频疾病的研究在经济上更可行。
Fundamentally, this was a restatement of their thesis that AI can massively accelerate drug discovery and make research into low-frequency diseases more financially viable.
不过，Anthropic强调，AI在生物科技中的应用并非全自动化，而是减少整个过程中的摩擦和瓶颈。
Still, Anthropic hammered home that AI use in biotech isn't about full automation— it's about reducing friction and bottlenecks throughout the process.
在X上，Chubby对围绕IPO的元策略评论道，药物研发将成为未来最重要的收入来源之一。
Commenting on the meta-strategy of promoting the IPO, Chubby on X wrote, Drug discovery will be one of the most significant revenue streams in the future.
这里的目标是早早定下正确的方向。
The goal here is to set the right course early on.
大多数疾病都会变得可治疗，这是一个覆盖70亿人的市场。
Most diseases will become treatable, a market encompassing 7 billion people.
时机显然是刻意的，赶在IPO前宣布，向投资者传递信号，表明业务未来会持续蓬勃发展。
The timing is clearly intentional, announced just before the IPO to signal to investors that the business will continue to thrive in the future.
这很可能会让那些不混科技圈的人也明白，为什么AI这么重要。
This will likely make it clear, even to those outside the tech nerd sphere, why AI is so important.
这不仅仅是写代码而已。
It goes beyond just writing code.
它在为人类解决问题。
It solves problems for humanity.
接着我们过去几周讨论过的一个主题，对一些人来说，这又是AI公司收入来源多样化、不局限于卖token的一个例子。
And picking up a theme that we've explored a bit in the past few weeks, for some, this is yet another example of an AI company diversifying their revenue streams outside of just selling tokens.
Martin Varsovsky写道，更优分子带来的数千万美元归该分子所有者所有。
Martin Varsovsky writes, the tens of millions from the better molecule go to whoever owns the molecule.
如果AI实验室想按价值收费，他们就必须拥有成果——发现药物、进行试验、持专利。
If AI labs want to be paid for value, they will have to own the outcome— discover the drug, run the trial, hold the patent.
这就是为什么Isomorphic Labs，即Alphabet从DeepMind分拆出来的药物研发公司，比任何定价页面都重要。
That is why Isomorphic Labs, the drug discovery company Alphabet spun out of DeepMind, matters more than any pricing page.
卖智能正在变成商品生意。
Selling intelligence is becoming a commodity business.
而拥有智能所发现的事物则不然。
Owning what intelligence discovers is not.
哇，你完全可以指望这话题继续争论下去，而且当Anthropic IPO时只会更加激烈。
Boy, you can count on that to continue to be a debate and one that gets nothing but more fierce when Anthropic IPOs.
说到这个，市场对Anthropic S-1整体正面的评价似乎得到了验证。
Speaking of which, the assessment that the market was generally positive on the Anthropic S-1 seems to have been validated.
Anthropic IPO文件公开亮相是在科技股经历三周回撤和半导体四个月熊市之后。
The public reveal of Anthropic's IPO filing came after a 3-week drawdown for tech stocks and a 4-month bear market in semiconductors.
不过本周到目前为止，AI交易全线飘红。
So far this week, however, the AI trade has been nothing but green.
自周一开盘以来纳斯达克指数上涨2%，费城半导体指数反弹近6%。
The Nasdaq index is up 2% since trading opened on Monday, and the Philadelphia Semiconductor Index has rebounded by almost 6%.
当然，市场并未认定Anthropic业务毫无弱点，但对于那些原本担心文件会确认AI资本支出无法收回的投资者来说，一个业务蓬勃、有现实发展路径的画面足以让许多投资者重回阵营。
Certainly, the market has not assessed that Anthropic has no weaknesses in the business, but for investors who were expecting the filing would confirm no path to realizing a return on AI CapEx, the picture of a booming business with a realistic path forward was enough to get many of those investors back on side.
Anthropic里程碑式的IPO文件，加上Meta激进的资本支出推动，重新点燃了更广泛的风险偏好。
Anthropic's landmark IPO filing alongside Meta's aggressive CapEx push has reignited broader risk appetite.
资金从防守姿态中撤出来，芯片股又重新稳稳地回到了领跑位置。
Pulling capital out of its defensive crouch and firmly restoring chip stocks back to the driver's seat.
老虎证券首席市场策略师James Wei表示，这完全归功于Anthropic的现金储备。他评论说，截至6月，Anthropic拥有超过1120亿美元的现金及现金等价物，即使不算任何潜在的IPO募资所得，它似乎也拥有相当雄厚的财务实力来支撑其扩张。
James Wei, the chief market strategist at Tiger Brokers, said it was all about Anthropic's cash pile, commenting, with more than $112 billion in cash and cash equivalents as of June, even before any potential IPO proceeds, Anthropic appears to have considerable financial capacity to support its expansion.
与此同时，《华尔街日报》正在印证我在周一节目中提出的观点：这次IPO顺利进行，确实符合AI行业所有人的最大利益。
Meanwhile, the Wall Street Journal is reinforcing the point that I raised on Monday's show, that it's really in everyone in the AI industry's best interest for this IPO to go well.
我当时说，这可能是OpenAI选择按兵不动、搁置下一代模型发布，让Anthropic凭借Claude 5.5独占几周聚光灯的绝佳理由。
I argued that that could be a very good reason for OpenAI to just sit back, mothball their next model release, and let Anthropic have the limelight for just a couple of weeks with Claude 5.5.
《华尔街日报》更进一步，指出这次IPO将在很大程度上说明整个行业的资本获取能力。
The Journal took it a step further, pointing out that the IPO will say a lot about access to capital for the entire industry.
如果Anthropic能够轻松募资1000亿美元——那将是史上最大规模的IPO——那么对于资金需求更迫切的小公司来说，这是一个好兆头。
If Anthropic can raise $100 billion with room to spare, which would be the largest IPO in history, then it bodes well for smaller companies with more pressing fundraising needs.
在另一篇文章中，《华尔街日报》探讨了华尔街将如何为存在性风险定价。
In a separate article, the Journal grappled with how Wall Street will price existential risk.
《华尔街日报》写道，逻辑很简单。
The logic is simple, writes the Journal.
正如Anthropic一位研究员上个月所说，如果人工智能在未来十年内有10%的可能杀死全人类，那就有90%的可能它真的非常非常好，只是设计者设法抑制了它的种族灭绝本能。
If there's a 10% chance that artificial intelligence could kill all humans in the next decade, as an Anthropic researcher said last month, there's a 90% chance it's really, really good, but designers managed to rein in its genocidal instincts.
一个聪明到足以杀死我们所有人、却更愿意帮助我们的AI，是极其有用的，应该为它的发明者带来可观收益。
An AI smart enough to kill us all, but which prefers to help us, is supremely useful and should generate healthy gains for its inventors.
我不知道作者James Mackintosh到底有多少是玩笑成分，但至少从本周来看，华尔街似乎已经不再关注P doom，而是开始把P boom计入价格了。
Now, I don't know exactly how firmly author James Mackintosh's tongue is in his cheek, but it does appear that this week at least, it looks like Wall Street has stopped focusing on P doom and is starting to price in P boom.
最后，简单跟进一下昨天关于OpenAI新科学突破的报道：一些科学家仍然对OpenAI的披露方式感到不满。
Lastly, in a brief follow-up to yesterday's story about new scientific breakthroughs from OpenAI, a group of scientists are still unhappy with how OpenAI went about their disclosure.
本月初，OpenAI一次性发布了大量新颖的数学证明，当时的批评是OpenAI在验证结果方面做得不够。
During the giant dump of novel math proofs earlier this month, the criticism was that OpenAI didn't do enough to verify results.
这一次，OpenAI试图通过任命多组科学专家来把关这些突破性成果，以回应这一批评。
This time around, OpenAI tried to address this criticism by appointing groups of scientific experts to shepherd the breakthrough results.
这些专家针对每项成果都推荐了不同的处理方式。
These experts recommended a different approach for each result.
数学证明只是作为学术论文公开发布了。
The math proof was just released publicly as an academic paper.
计算机科学方面，推翻exponential time hypothesis的成果，在发布时附带发布了证明和AI会话记录，还任命了一个科学家团队做进一步的工作，解释这项成果并探索其影响。
The computer science result, disproving the exponential time hypothesis, had proofs and AI sessions released alongside, with a team of scientists appointed to do further work explaining the result and exploring its implications.
然而，物理方面的成果——据称包含Yang-Mills theory的存在性证明——则被保密处理，同时一个物理学家团队正在验证和形式化这一成果。
However, the physics result, which was said to include an existence proof for Yang-Mills theory, was kept private while a group of physicists worked on verifying and formalizing the result.
在OpenAI发布公告之后，其自家的数学与人工智能顾问组公开反对了这一做法。
Following OpenAI's announcement, their own advisory group on mathematics and artificial intelligence publicly disagreed with this approach.
在一份公开建议中，数学顾问组表示，他们非常尊重被任命来把关这些成果的物理学家，但理解的过程应该在公开的环境中展开，让每个人都拥有相同的获取权限。
In a public recommendation, the mathematics advisory group said they had great respect for the physicists appointed to shepherd the results, but that understanding should develop in the open, with everyone given the same access.
周二晚些时候，相关物理学家团队发布了一份回应，评论说：OpenAI手稿的长度和复杂性令人望而生畏，需要一些时间来消化其方法论，以及它在物理学和数学两方面的意义。
Later on Tuesday, the group of physicists in question released a response, commenting, the length and complexity of the OpenAI manuscripts are formidable, and some time will be needed to digest its methodology and meaning in both physics and mathematics.
在我们发布之前，我们希望先确信它是正确的。
Before we release it, we would like to be convinced that it is correct.
这是一场关于学术界如何应对AI快速创造新知识的实验。
This is an experiment in how the academic community can deal with the rapid development of new knowledge by AI.
我们打算非常谨慎地推进。
We intend to proceed very carefully.
尽管如此，对延迟发表的沮丧情绪似乎正在全国各地的教师休息室里蔓延。
Still, the frustration with the delayed publication seems to be playing out in faculty lounges across the country.
The Verge采访了几位物理学教授，其中许多人呼吁立即发布。
The Verge spoke with several physics professors, with many of them calling for an immediate release.
高等研究院的Konstantin Koegler说：我接受他们像今天这样发布这些问题，他指的是发布一小批3个极其重要的成果。
Konstantin Koegler from the Institute for Advanced Study said, I'm fine with them releasing the problems as they did today, referring to releasing a smaller batch of 3 extremely important results.
但Koegler补充说，只有几位杰出科学家能查看这些成果，这完全没有道理。
However, Koegler added, it doesn't make any sense that only a few distinguished scientists can look at them.
德克萨斯大学奥斯汀分校的Andreas Karch说，他们亲手挑选这个团队这件事本身就是一场灾难。
Andreas Karch from the University of Texas at Austin said, the fact that they handpicked the group is a disaster.
那些大型实验室现有的权力已经够大了。
The large labs have enough power as it is.
他们不应该挑选赢家和输家。
They should not be picking winners and losers.
卡奇最初提出了一个基于 merit 的筛选标准，不过他承认这很可能会产生同样的群体，包括诺贝尔奖得主。
Karch initially suggested a merit-based selection criteria, although he acknowledged that this would likely produce the same group that included Nobel laureates.
相反，他只是呼吁 OpenAI 公布证据，说这样既能民主化访问权限，又能表明他们关心让结果有意义且易于理解。
Instead, he simply called on OpenAI to release the proof, saying, this would democratize access while still showing that they care about making results meaningful and understandable.
我不知道，老兄，我觉得这件事上 OpenAI 怎么做都不讨好。
I dunno, man, I think on this one OpenAI simply can't win.
如果他们尽快发布所有东西，就会被批评给学术界制造了一个巨大的验证问题。
If they release everything they have as soon as possible, they're criticized for creating a gigantic verification problem for academics to deal with.
如果他们安排自己的验证工作组，把重要的结果扣着直到没问题才发布，就会被批评在科学上搞门禁。
If they arrange their own verification working groups and hold back important results until they're fine, they're criticized for gatekeeping science.
完全可以理解的是，人们对这些实验室拥有这么大的权力感到不安，这不仅仅基于实验室所做的具体决策，而是基于他们相对于这种新智能所处的棘手现实。
The thing that's totally understandable is that people are uncomfortable with how much power these labs have, and that is not solely based on specific decisions that labs make, it is based on the intractable reality of where they sit vis-à-vis this new intelligence.
但具体到这件事上，他们基本上怎么做都是错。
But when it comes to this sort of specific, they're pretty much damned if they do and damned if they don't.
斯坦福大学的数学家贾里德·杜克·利希特曼评论说，我的感觉是，无论采用什么方法，总会有人不高兴。
Stanford mathematician Jared Duker Lichtman commented, my sense is that for any approach, there is going to be someone who is upset.
所以问题只是找到最不冒犯的那个选项。
So it's just a question of finding the least offensive option.
我很高兴这不是我的工作，今天就到这里，新闻头条部分结束。
I am glad that is not my job, and for now, that's gonna do it for the headlines.
接下来是主节目。
Next up, the main episode.
毕马威和德克萨斯大学奥斯汀分校的一项新研究发现，当人们与AI合作时，相似的技能并不能保证相似的结果。
A new study from KPMG and the University of Texas at Austin found that when people work with AI, similar skills don't guarantee similar outcomes.
研究人员研究了500多名早期职业专业人士，发现表现最好的人通过引导、评估和优化AI的输出，持续放大了AI的价值。
Researchers studied more than 500 early-career professionals and found that the best performers consistently amplified the value of AI by guiding, evaluating, and refining its outputs.
这些顶尖表现者被称为AI放大器，他们并不是单纯由知识定义，而是由与AI合作的方式定义。
These top performers, called AI amplifiers, weren't defined by what they knew alone, but by how they worked with AI.
了解更多关于是什么将AI放大器与其他区分开来的信息，请访问 kpmg.com/us/ai-amplifiers。
Learn more about what separates AI amplifiers from everyone else at kpmg.com/us/ai-amplifiers.
本期节目由 CData 提供赞助。
This episode is brought to you by CData.
推广AI的公司总是碰到同一堵墙。
Companies rolling out AI keep hitting the same wall.
Token消耗在攀升，安全限制着AI代理，而且没人知道AI到底是怎么被使用的。
Token spend is climbing, security is holding agents back, and no one knows how AI is truly being used.
根本原因是什么？
The root cause?
AI与你业务运行所需的系统和上下文是割裂的。
AI is cut off from the systems and context your business runs on.
这就是为什么CData构建了Connect AI Gateway，一个位于AI和系统之间的控制点。
That's why CData built Connect AI Gateway, one control point between your AI and your systems.
基于为Palantir、Google、Salesforce等数百家公司提供支持的连接层构建，每个代理都能获得它所需的确切数据和工具。
Built on the connectivity layer powering Palantir, Google, Salesforce, and hundreds more, each agent gets exactly the data and tools it needs.
它可以管理到记录级别，并让你最终在一个地方看到整个组织的AI使用情况。
It governs down to the record and lets you finally see your org's complete AI usage in one place.
准确的答案，更低的token成本，以及一个每次交互都在学习你业务的门户。
Accurate answers, a fraction of the token cost, and a gateway that learns your business with every interaction.
详情请访问 cdata.ai。
See how at cdata.ai.
你整天都在开一个接一个的会。
You're in back-to-back meetings all day.
你努力保持专注，但同时又担心忘记决策、行动项或重要的下一步。
You're trying to stay present, but you're also worried you'll forget the decision, the action item, the important next step.
这正是我们在AI Daily Brief这里经常遇到的场景，因为一切变化太快，我们也在快速迭代我们所做一切。
This is a scenario that we absolutely live here at the AI Daily Brief, given how fast things move and change and how quickly we're evolving everything that we do.
这时候Granola就派上用场了。
That's where Granola comes in.
Granola 是一个用于会议的 AI 记事本。
Granola is an AI-powered notepad for meetings.
你可以像平时一样随手记下粗略的笔记，后台的 Granola 会转写这些内容，并在会议结束时把它们变成清晰、有用的笔记。
You jot down rough notes like you always do, and, in the background, Granola transcribes and turns them into clear, useful notes when the meeting ends.
没有机器人加入你的通话，没有干扰，只有一个干净的记事本，帮你保持专注。
No bots joining your calls, no distractions, just a clean notepad that helps you focus.
在通话期间或通话结束后，你可以和笔记对话，让 Granola 帮你提取行动项、协助谈判、写跟进邮件，甚至可以用预设提示词来对你进行指导。
During or after the call, you can chat with your notes, ask Granola to pull out action items, help you negotiate, write a follow-up email, or even coach you using recipes which are premade prompts.
只要你在第一次会议上试过，之后就很难不用它了。
Once you try it on a first meeting, it is very hard to go without.
前往 granola.ai/AIDaily，使用代码 AIDaily，新用户可享 3 个月 100% 优惠。
Head to granola.ai/AIDaily and use code AIDaily to get 100% off for 3 months for new users.
也就是 granola.ai/AIDaily，使用代码 AIDaily，新用户即可享受 3 个月全免。
That's granola.ai/AIDaily and code AIDaily to get 100% off for new users for 3 months.
这就是为什么大多数遗留系统现代化项目会失败。
Here's why most legacy modernization projects fail.
因为负责工作的 AI 无法理解大规模代码库。
The AI doing the work can't understand codebases at scale.
它只看到很小的上下文片段，检查语法，却忽略了分布在整个全球应用生态系统中长达数年的决策。
It sees a small slice of context, examines syntax, and misses years of decisions distributed across the global application ecosystem.
Blitzy 解决这个问题的方式，和它解决一切问题的方式一样。
Blitzy solves this the way it solves everything.
在任何迁移开始之前，Blitzy 的智能体会先以你的代码为基础，把整个遗留系统逆向工程成一个持久的知识图谱。
Grounded in your code before any migration begins, Blitzy's agents reverse engineer the entire legacy system into a persistent knowledge graph.
包括每一个依赖、每一个约束、每一段曾经只存在于某个工程师脑子里的经验知识。
Every dependency, every constraint, every piece of tribal knowledge that used to live in one engineer's head.
基于这种理解，Blitzy 能够自主执行语言迁移、框架升级，以及从单体架构到微服务的转型，而且整个过程都经过端到端验证。
From that understanding, Blitzy autonomously executes language migrations, framework upgrades, and monolith-to-microservices transformations, all validated end-to-end.
有一个 Blitzy 客户在 16 周内完成了价值一千万美元的单体保险系统现代化，而使用编码智能体的基线是 137 周。
One Blitzy customer modernized a $10 million monolithic insurance stack in 16 weeks against a 137-week baseline with coding agents.
这相当于压缩了 9 倍的时间。
That's 9x compression.
在加速你的路线图的同时，清除技术债务。
Retire technical debt while accelerating your roadmap.
到 blitzy.com 看看实际效果。
See how at blitzy.com.
也就是 blitzy.com。
That's blitzy.com.
欢迎回到 AI Daily Brief。
Welcome back to the AI Daily Brief.
在进入今天的节目之前，先提醒一下：今天会有很多视觉内容。
Quick note before we dive into today's episode: there is going to be a lot of visual content.
和往常一样，我会尽最大努力让内容即使你只是听、不看也有意义，但这期节目可能更适合去 Spotify 的视频版或 YouTube 上看。
As always, I will do my best to make it relevant even if you are just listening, not watching, but this might be a good one to go check out either on Spotify with video or on YouTube.
Apple 在继续推广视频播客，但还没有对我在用的这个托管平台开放这个功能。
Apple continues to roll out video podcasting, but has not yet made it available for my particular host.
希望很快就能有。
Hopefully that will come soon.
今天我们聊的是 Genie 4。
Today we are talking about Genie 4.
这是谷歌世界模型的最新迭代，它可能对机器人技术、游戏等领域产生重大影响。
It's the latest iteration of Google's world model, and it could have some big implications for robotics, gaming, and more.
在 AI Daily Brief 上，我们从 2024 年世界模型刚开始出现时，就一直在追踪它的发展。
On the AI Daily Brief, we've been tracking the development of world models since they first began showing up in 2024.
其中一个原因是，像 Yann LeCun 和 Fei-Fei Li 这样的 AI 领军人物认为，世界模型比大型语言模型更能通往 AGI。
One of the reasons for that was because AI luminaries like Yann LeCun and Fei-Fei Li believed that world models were a more viable path to AGI than large language models.
他们的论点基本上是，要实现 AGI，你必须对周围世界有不同的物理理解，而不仅仅是能读很多资料。
The argument was basically that you had to have a different physical understanding of the world around to achieve AGI, not just be able to read a bunch of stuff.
显然我这说得过于简化了，但你明白我的意思。
Obviously I'm being wildly over-reductive, but you get the idea.
世界模型的早期实验是令人叹为观止的技术演示——它们向我们展示了与以往所见AI能力完全不同的东西。
The early experiments in world models were breathtaking tech demos— they showed us something completely different to anything we'd seen AI do before.
当时，生成视频还处于起步阶段，所以能够创建完整的3D世界本身就是一个巨大的突破。
At the time, generative video was still in its infancy, so the ability to create fully 3D worlds at all was a massive breakthrough.
然而即使在那时也很清楚，这些演示真正展示的不是当前能做什么，而是未来将会如何。
And yet even then it was clear that these demos really weren't about what was available now but what was to come.
在接下来的一年左右，Google在Genie模型上取得了快速的进展。
Over the next year or so, Google made rapid progress with their Genie models.
Genie 2首次亮相是在2024年12月，已经非常令人印象深刻，但当Genie 3在2025年1月预览时，很明显Google取得了巨大的飞跃。
Genie 2, which was first revealed in December 2024, was very impressive, but when Genie 3 was previewed in January 2025, it was clear that Google had made a huge leap.
该模型最终在2025年8月向公众开放，首次允许用户通过提示词创建世界，然后用键盘和鼠标在这个世界中导航。
The model was eventually served to the public in August of 2025, and allowed users to prompt worlds into existence and then navigate those worlds with a keyboard and a mouse for the first time.
最令人印象深刻的技术突破之一是我们前所未见的持久性水平——用户可以改变世界中的某些东西，比如在墙上画一个符号，把视线移开，再回头看，符号仍然保持不变。
One of the most impressive technical breakthroughs was a level of persistence that we hadn't seen before— users could change something about the world, like paint a symbol on a wall, look away, and then look back, and the symbol would be the same.
当时，生成视频模型在努力达到这种一致性水平上遇到了极大的困难。
At the time, generative video models were struggling mightily to achieve that level of consistency.
Google表示，这是模型工作方式中涌现出的特性，而不是他们添加的功能。
Google said that this was an emergent property of how the models worked rather than a feature they added.
Genie-3基于之前生成的内容，以自回归方式生成每一帧视频，因此模型在生成世界的同时会保留世界的记录。
Genie-3 generated each frame of video autoregressively based on what had come before, so the model kept a record of the world as it was being generated.
然而，我们仍然明显处于世界模型的技术演示阶段——非常酷的技术演示，但距离实际应用还太远、太有限。
And yet still we were very clearly in the tech demo stage of world models— incredibly cool tech demos, but far, far too limited for practical use.
技术专家建议世界模型可以帮助训练机器人或实时生成完全可玩的游戏，但Genie-3由于高计算需求，生成长度仅限于60秒。
Technologists suggested world models could help train robots or generate fully playable games in real time, but Genie-3 was limited to 60 seconds of generation because of the high compute requirements.
与此同时，在许多人的眼中，Google一直没有顶级语言模型来参与当前AI竞争的主要领域，比如智能体编码。
Now, alongside all of this, Google has of course in the eyes of many been languishing without a state-of-the-art language model to compete in the main areas of AI competition right now like agentic coding.
然而，对于大多数密切关注的人来说，Google在先进多模态AI方面的能力从未有过疑问。
And yet for most people who were paying close attention, there was never any question of Google's proficiency in advanced multimodal AI.
现在Google发布了Genie-4，并随之提供了一条让世界模型更加实用和有用的路径。
Now Google has revealed Genie-4, and with it, a pathway to making world models far more practical and useful.
在这一版本中，DeepMind对模型的架构进行了巨大的改变。
With this iteration, DeepMind has made a huge change to the model's architecture.
Genie-4不再是独立模型，而是整合了其他Google模型来提供Genie自身无法完成的功能。
Rather than being a standalone model, Genie-4 integrates other Google models to provide things that Genie can't do by itself.
Google的视频模型Gemini Omni将模拟锚定在真实世界的外观和行为上。
Gemini Omni, Google's video model, anchors the simulations in real-world appearance and behavior.
大概这意味着Omni会创建一段视频片段作为参考，然后Genie将其扩展成一个完全可玩的世界。
Presumably that means something like Omni creates a video clip for reference, which Genie then expands out into a fully playable world.
此外，Gemini 3.8 Flash提供了物理交互的推理引擎，为模型带来了可靠的因果关系。
In addition, Gemini 3.8 Flash provides a reasoning engine for physical interactions, giving the model reliable causality.
在此基础上，一些未指明的Google音频模型组合首次为生成的世界带来了声音。
On top of that, some unspecified combination of Google's audio models brings sound to generated worlds for the first time.
结果就是，Google的整个AI生态系统被利用起来，创造出迄今为止我们见过的最逼真的世界模拟。
The result is Google's entire AI ecosystem being leveraged towards creating the most realistic world simulations we've seen to date.
在公告文章中，Google重点介绍了他们认为Genie-4的四大关键进展。
In their announcement post, Google highlighted 4 key features that they viewed as the big advances for Genie-4.
首先，用户现在可以实时提供提示词来引导模拟。
First, users can now deliver prompts to guide the simulation in real time.
之前版本的Genie只允许用户在开始时用提示词创建世界，然后在一个稳定的生成世界中导航。
Previous iterations of Genie only allowed users to create the world with a prompt upfront, then navigate whatever was generated as a stable world.
现在，用户可以给出实时提示来引导镜头，或者通过改变天气、引入新物体或角色来修改环境。
Now, users can give live prompts to guide the camera, or to modify the environment by changing the weather or introducing a new object or character.
其次，Google强调了自Genie-3以来在真实感方面的巨大提升。
Second, Google highlighted the huge jump in realism since Genie-3.
他们写道：“基于对世界的先进物理理解，Genie-4模拟了真实的物理场景，如空气动力学、碰撞等。”
They wrote, "Built on an advanced physical understanding of the world, Genie-4 simulates real-world physics scenarios such as aerodynamics, collisions, and more."
最显著的飞跃是Genie-4如何模拟人物，创造出移动自然、对环境做出有机反应、并且可以实时交互操控的角色。
The most dramatic leap is how Genie-4 simulates people, creating characters that move naturally, respond organically to their environments, and can be interactively steered in real time.
第三个亮点是生成音频，正如我提到的，这是我们在世界模型中从未见过的。
The third highlight was generative audio, which as I mentioned we haven't seen in world models before.
现在谷歌的 Veo 视频模型和 Lyria 音乐模型都有出色的音效，所以在 Genie 的新版本里出现惊人的音频也不奇怪。但关键在于，这些世界不仅能包含音乐或人声，还能加入音效，这给了创作者一个完全不同的、更高层次的工具。
Now Google's Veo video model featured impressive sound, as did their Lyria music model, so it's not all that surprising to find amazing audio in the new version of Genie, but the fact that the worlds can not only include music or human speech but also sound effects gives creators a whole different level of tooling than was available before.
比如，看看那些与抽象景观相匹配的超凡脱俗的声音，就像他们示例中彩色玻璃荒野场景所伴随的那种声音。
Check out for example the otherworldly sounds to match an abstract landscape, such as those that go alongside the stained glass wilderness scene that they used as an example.
不过语音部分感觉稍微落后于当前最先进的水平，仍然带有那种空洞、带回声的风格，那是大约三个月前生成视频对话的标志性特征，比如这对情侣骑摩托车时聊天的例子。
Now the speech does feel a little behind the state of the art, still featuring that hollow and echoey style that was the hallmark of dialogue in generative video until about 3 months ago, such as this example of a couple talking while riding on a motorcycle.
你真的觉得我们能成功吗？
Do you really think we're gonna make it?
最后，谷歌强调了在后台加入推理模型会带来怎样的效果。
Finally, Google highlighted what adding a reasoning model in the background brings to the table.
Genie 现在可以模拟复杂的多步骤动作，其真实感比我们以前见过的要高得多。
Genie can now simulate complex, multi-step actions with a much higher level of realism than we've seen before.
为了解释其工作原理，谷歌写道：‘为了实现实时模拟，Gemini 3.8 Flash 在循环中充当引导代理。’
Explaining how this works, Google wrote: "To achieve this in real-time, Gemini 3.8 Flash serves as a guiding agent in the loop."
当 Genie-4 实时生成一个世界时，Gemini 持续观察场景，将目标分解成一系列步骤，并审查帧的流入，以跟踪进度并验证每一步是否成功。
As Genie-4 generates a world in real-time, Gemini continuously observes the scene, breaks down goals into a sequence of steps, and reviews incoming frames to track the progress and verify that each step succeeded.
随着世界变化或用户介入，Gemini 会即时调整计划，从而解锁具有可靠因果关系、程序化行为和多步推理的世界。
As the world shifts or the user intervenes, Gemini adapts its plan in the moment, unlocking worlds with reliable causality, procedural behaviors, and multi-step reasoning.
不可否认，这是世界模型的一大巨大进步。
The result is undeniably a huge step up in world models.
我们现在拥有了现代视频模型的真实感，结合现代 LLM 的推理能力，全部内置到一个完全可玩的世界模拟器中。
We now have the realism of modern video models combined with the reasoning capabilities of modern LLMs baked into a fully playable world simulator.
一个直接的影响是机器人技术的新范式。
One immediate implication is a new paradigm in robotics.
谷歌一直在尝试使用 Gemini 4 Argon 而不是 Gemini 3.8 Flash 作为尖端机器人研究的后端推理模型。
Google has been experimenting with using Gemini 4 Argon rather than Gemini 3.8 Flash as the backend reasoning model for cutting-edge robotics research.
在展示一个机器人在 Genie 模拟中下棋的演示时，他们写道：‘Argon 将用户指定的任务分解为步骤，供 Genie 渲染。这些步骤作为文本字幕交错输入给 Genie-4，同时模拟正在进行渲染。’
Showing a demonstration of a robot playing chess inside a Genie simulation, they wrote, "Argon breaks down a user-instructed task into steps for Genie to render. These steps are interleaved as text captions to Genie-4 as the simulation is rendering."
因为 Genie-4 可以通过 Gemini 用文本字幕实时引导，这现在开辟了一条路径，可以在模拟朝着任务完成推进的过程中对模拟进行调整。
Because Genie-4 can be steered in real time with text captions using Gemini, this now unlocks the path of adapting the simulation as it progresses towards the completion of that task.
这种能力成为一条路径，可以用 Gemini 4 Argon 和 Genie-4 在想象世界中模拟各种各样的场景。
This capability becomes a path for simulating a breadth of scenarios in imagined worlds using Gemini 4 Argon and Genie-4.
说白了，谷歌现在可以为机器人创建完全模拟的训练环境，以软件速度进行训练，完全不受现实世界限制的束缚。
Put a little less densely, Google can now create entirely simulated training environments for robots, unlocking training at software speed completely untethered from real-world limitations.
社区的普遍反应是‘惊掉下巴’。
The general response from the community was jaws hitting floors.
我认为值得指出的是，因为我之前也批评过，那就是每个人都有动机去关注 3D 渲染之类的内容，因为模型刚发布时这些东西在社交媒体上很流行。
And I think it's worth noting, since I've critiqued before, the fact that everyone has an incentive to focus on 3D renderings and things like that because they're popular social media shares when a model first gets released.
但在这种情况下有点不同，因为那正是这个模型的全部功能。
That's a little bit different in this case because that is the entirety of what this model does.
换句话说，它不是那种用于代理编码或写文章的工具；人们是依据它制作 3D 渲染的能力来评判它的。
In other words, it's not a model that you're going to use for agentic coding or writing essays; people are judging it on the basis of its ability to make 3D renders.
你创建和分享的视觉与互动世界，才是这个模型的核心目的。
The visual and interactive worlds that you're creating and sharing are the purpose of the model at its core.
Genie 4 是一种相当疯狂的经历，它会随着你的操作生成类似游戏的环境。
Genie 4 is quite an insane experience, generating game-like environments as you go.
它并非完美无缺，但我可以输入‘新哥特式粗野主义城市，居住着机器人和马’，它就能在一分钟内生成一个。
It isn't flawless, but I can type, neo-gothic brutalist city inhabited by robots and horses, and it creates one in a minute.
然后我让它加上龙，再加一个友好的水獭 NPC，全部实时完成。
Then I asked it to add dragons, then a friendly otter NPC, all in real time.
Ethan 还创建了一个维多利亚时代下午茶的场景，主角是一只穿着得体服装的老鼠，然后实时添加了一只路过的豪猪、老鼠长出翅膀、老鼠飞向茶杯。
Ethan also created a scene of a Victorian high tea where the main character was a mouse in appropriate garb, and then in real time added a porcupine walking by, the mouse growing wings, the mouse flying to the teacup.
他说，虽然有点小毛病，但和几年前相比，已经先进得不可思议了。
He says, it's glitchy, but also impossibly advanced compared to a couple of years ago.
如果必须用一个表情符号来总结人们的反应，那肯定是‘惊掉下巴’的表情，正如 a16z 的 Justine Moore 分享的那样。
If you had to sum up people's reactions with a single emoji, it would certainly be the mind-blown emoji, as shared by Justine Moore from a16z.
她分享了一个洛夫克拉夫特风格游戏的片段，写道：‘这是一个完全生成的互动世界，来自一个起始提示词。’
Sharing a clip of a Lovecraftian-style game, she writes, this is a fully generated interactive world from a single starting prompt.
你现在可以通过提示词改变世界，并与它进行原生音频的互动。
You can now prompt to change the world and interact with it with native audio.
她还做了一个新版本的Grand Theft Auto，然后说：‘我们居然在GTA 6之前就先搞出了GTA Waymo！’
She also created a new version of Grand Theft Auto and said, "We got GTA Waymo before GTA 6!"
ChrisGPT试图解释为什么这件事对AI整体的进步也很重要。
ChrisGPT tried to explain why this matters for the larger advancement of AI as well.
他写道，很多人可能不明白世界模型有什么用，但我觉得Genie 4让我们更清楚地看到了，它们怎么帮助我们更接近AGI。
He writes, a lot of people may not understand the purpose of world models, but I think Genie 4 gives us a clearer glimpse of how they could help us get closer to AGI.
Google已经在把Genie和Gemini结合起来——可惜在实时循环里用的是3.8 Flash。
Google is already combining Genie with Gemini— unfortunately 3.8 Flash in the live loop.
Gemini可以观看模拟过程，把目标拆解成更简单的步骤，还能随着世界的变化调整自己的计划。
Gemini can watch the simulation, break goals down into simpler steps, and adjust its plan as the world changes.
最终，把大语言模型的推理能力和世界模型学到的物理动态结合在一起，也许能让智能体有办法测试自己的行动，提升它们的时空推理能力——就像人类在做某件事之前，可以想象一个真实场景，先预演一下可能会发生什么。
Eventually, pairing an LLM's reasoning with a world model's learned physical dynamics could give agents a way to test actions and improve their spatiotemporal reasoning, similar to how humans can imagine a real-world scenario and think through what might happen before doing something.
这就给了智能体一个场所，在去真实世界尝试之前，先比较不同的行动、从模拟结果中学习——前提是这些模拟足够准确。
This gives agents a place to compare actions and learn from simulated outcomes before trying them in the real world, assuming these simulations are accurate enough.
我觉得这能帮助解决模型目前仍然比较吃力的空间推理和物理推理问题。
I think this could help with some of the spatial and physical reasoning problems models still struggle with.
人类可以在脑子里先预演一遍自己要做的事。
Humans can mentally rehearse what they're about to do.
但大语言模型只能做基于文本的推理，能用文字描述物理和数学的可能性，而交互式世界模型给了AI一种把它们实际演出来的方式。
But LLMs are only able to do text-based reasoning, and can describe physics and math possibilities in text, but an interactive world model gives AI a way to play them out.
显然，物理效果和长期一致性还需要改进，但这是朝着给AI提供更丰富的环境去互动、去学习迈出的一大步。
The physics and long-term consistency still need work, obviously, but this is a great step towards giving AI richer environments to interact with and learn from.
现在，你大概也能猜到，人们看到的一个非常明确的应用领域就是游戏。
Now, as you might imagine, one place that people are seeing very clear implications is for the world of gaming.
事实上，从世界模型诞生之初，大多数人就认为那会是这项技术最先产生重大影响的地方。
In fact, since the earliest days of world models, most have assumed that that was where this technology would have its first big impacts.
想法是，有一天——也许很快——玩家就能用世界模型来生成自己专属的个性化游戏，获得完全量身定制的生成式游戏体验。
The idea is that one day, perhaps one day soon, gamers will be able to prompt their own personalized game using a world model and create a fully tailored generative gaming experience.
事实上，Genie 3发布的时候，游戏类股票直接暴跌；不过这次Genie 4的发布，对这个行业的冲击要小一些。
In fact, the release of Genie 3 sent gaming stocks plummeting, although this Genie 4 release was less damaging to the sector.
看来投资者的理解更成熟了一些，他们意识到这项技术距离颠覆游戏行业还有一段路要走。
It seems investors have gotten a little bit more sophisticated in their understanding that the technology is still a little way off for being able to disrupt the gaming industry.
不过，这个行业已经在经历一场巨大的AI转型了。
Still, there's already a huge AI transition underway in that industry.
这得益于Opus 5.5反编译和重新打包游戏的能力。
Driven by Opus 5.5's ability to decompile and repackage games.
最初几周出现了一些奇怪的混搭，比如把Mario塞进Dark Souls里。
The first few weeks of that saw weird mashups, like dropping Mario into Dark Souls.
然后就是一波移植热潮，人们发现Opus基本上能让任何游戏兼容几乎任何硬件。
Then there was the wave of ports, with people realizing Opus could make basically any game compatible with pretty much any hardware.
一位科技记者把Tim Sweeney的第一款游戏从PC移植到了NES上，之后这位Epic Games的CEO评论说：‘这太不可思议了，过去需要耗费大量人力的事，现在作为一个业余爱好项目就能非常高效地完成。’
After a tech reporter ported Tim Sweeney's first game from PC to the NES, the Epic Games CEO commented, "This is amazing and shows what would have taken a vast manual effort in the past can be done as a hobby project very efficiently."
游戏和软件的新时代真的要来了！
Truly a new age of gaming and software ahead!
现在，人们正在把一些真正古怪疯狂的想法变成现实。
Now people are bringing some truly zany ideas to life.
Lucas Martinich成功地把Mario 64变成了一款混合现实游戏。
Lucas Martinich managed to turn Mario 64 into a mixed reality game.
游戏画面显示在VR眼镜里的虚拟屏幕上，玩家可以伸手进去，在Mario跑来跑去的时候和他互动，比如用手当额外的平台。
The gameplay is presented on a virtual screen inside of VR glasses, and the player can reach in and interact with Mario as he's running around, for example using their hand as an additional platform.
玩家还能抓住Mario，把他从游戏里拉出来，放到自家的厨房桌子上，把现实空间变成Mario可以到处奔跑的另一个世界。
The player can also grab Mario, pull him out of the game, and place him on their kitchen table, turning their real-world space into another world for Mario to run around in.
总之，虽然技术可能还没准备好实现完全生成式的游戏，但AI用户正在用我们已有的技术做出一些奇怪而惊人的事情，而这必然会塑造接下来的发展方向。
In short, while the tech might not be ready for fully generative games, AI users are figuring out how to do some strange and incredible things with the technology we already have, which will inevitably shape what comes next.
和Genie 3一样，Google这次也向Gemini Ultra的订阅用户开放了完整的Genie 4，让大家试用。
As with Genie-3, Google is making Genie-4 fully available for Gemini Ultra subscribers to try out.
它目前仍然只能做2分钟的模拟，虽然已经是Genie 3的两倍了，但距离完全生成式的游戏体验还有差距。
It's still only capable of 2-minute simulations, which is double the limits of Genie-3, but still short of a fully generative gaming experience.
所以现在来看，这东西多少还像个科技玩具。
So for now, this is still a bit of a tech toy.
虽然这项技术开始显露出一些非常有用的苗头。
Although one that's starting to show glimmers of a very useful technology.
不过，Genie-4 的发布还有一个方面值得我们深入聊一聊。
There is one more aspect of the Genie-4 announcement, however, that's worth digging into a little bit.
简而言之，有些人解读说，Genie-4 预示着 DeepMind 还有另一张牌要打。
In short, it's the interpretation by some that Genie-4 is a sign that DeepMind still has another act.
过去六个月，坦白说，Google 在大语言模型方面的表现相当令人失望。
The past 6 months have been, putting it fairly mildly, disappointing for Google on the LLM front.
他们最新的模型 Gemini 4 Argon 算是及格了，但肯定没有让 Google 重新回到 AI 竞赛的第一梯队。
Gemini 4 Argon, their most recent model, got a passing grade but certainly didn't put Google back in the AI race.
不过在 Anthropic 的 S-1 文件里，他们只把 Google 和 OpenAI 列为两大主要竞争对手。
Still, in Anthropic's S-1 paperwork, they named Google as one of only 2 other major competitors alongside OpenAI.
如果说 AI 的前沿是高低不平的，那么 DeepMind 的产出可能是最参差不齐的那个。
If the AI frontier is jagged, DeepMind's output is perhaps the most jagged of all.
他们在追赶大语言模型方面还没完全跟上，但在多模态上的进展依然无人能敌。
They haven't been able to fully catch up on LLMs, but their progress on multimodal is still unrivaled.
通过 Genie-4，Google 证明了把一系列不同的模型架构组合起来，可以创造出全新的、潜力巨大的 AI 系统。
With Genie-4, Google is demonstrating that combining a series of disparate model architectures can produce entirely novel AI systems with huge amounts of potential.
而且，作为一家 AI 企业，Google 还有一些分散布局的投资正在看到回报。
What's more, as an AI business, Google also continues to have some diverging bets that are paying off.
去年 Google 开始加大 TPU 产量，并传出要对外销售这些芯片的讨论时，很多半导体分析师对此嗤之以鼻——Nvidia 在 AI 芯片上牢牢把控着市场，而 Google 在销售和业务拓展方面是从零开始。
When Google started ramping up TPU production last year and leaked discussions about selling the chips externally, many semiconductor analysts scoffed at the idea— Nvidia had a stranglehold on AI chips, and Google was starting from zero on sales and business development.
不过，TPU 的销售如今已经成长为营收表上一个重要的项目，虽然几乎全部是卖给了 Anthropic。
Still, TPU sales have grown to become a significant line item in the revenue statement, albeit almost entirely through sales to Anthropic.
但这周，The Information 报道说 Google 赢得了两个新客户：OpenAI 和 Oracle。
This week, however, The Information reports that Google has won 2 new customers: OpenAI and Oracle.
OpenAI 的交易条款没有披露，所以很难知道会采购多少 TPU、如何部署；但 Oracle 将成为继 Google 之后第一家向外部客户出租 TPU 的大型云服务商。
Terms weren't disclosed for the OpenAI deal, so it's difficult to know how many TPUs will be purchased and how they will be deployed, but Oracle is set to become the first large cloud provider aside from Google to rent TPUs to outside customers.
除了硬件销售，未来几个月 Google 另一个值得关注的大看点，就是他们能否在大语言模型上快速迭代。
And in addition to their hardware sales, the other big thing to watch for Google in the coming months will be whether they can rapidly iterate on their LLMs.
甚至在 Gemini 4 Argon 发布之前，Google 就已经准备好了后续模型，用于内部测试。
Even before Gemini 4 Argon was released, Google already had their follow-up model ready for internal testing.
这个模型的代号是 Gemini 4 Carbon，而且试过的 Google 员工似乎对它评价不错。
The model is known by the codename Gemini 4 Carbon, and it seems like it's getting positive reviews from the Googlers who've tried it.
关键的是，Gemini 4 Argon 不仅仅是一个新的商业模型。
Critically, Gemini 4 Argon isn't just a new commercial model.
它同时也是一个新的超大规模基础模型，Google 可以在此基础上继续训练。
It's also a new ultra-large base model that Google can train on top of.
他们不需要再完成一次漫长的预训练——可以在 Argon 的基础上应用相对快速的后期训练。
They don't need to complete another long pre-training run— they can apply relatively fast post-training on top of Argon.
我们已经看到了这对 Meta 意味着什么——MuseSpark 发布时，第一版表现相当糟糕，但他们在五个月内又发布了三个版本，最终成功打造了一个有竞争力的中型模型。
We've seen what that meant for Meta with the release of MuseSpark— the first version was pretty dismal, but they released 3 additional versions in 5 months and bootstrapped their way to a competitive mid-sized model.
在超大规模模型上做强化学习不会那么快，但这仍然给 Google 提供了一条清晰的快速改进模型的路径。
Reinforcement learning runs on an ultra-large model won't be quite as fast, but it still gives Google a clear pathway to rapid model improvement.
另外，这个播客的创作者和听众可能太深入 AI 领域了，反而很难理解现有的分发渠道比模型能力本身更重要这一点。
There's also the fact that the creators and listeners of this podcast may be a little too deep in AI to understand how much existing distribution matters over even model capabilities.
一周前，Google 的 Logan Kilpatrick 发帖说，Gemini 4 Argon 上线仅仅 72 小时，我们就处理了大约 25 万亿个 token。
A week ago, Google's Logan Kilpatrick posted, in just the first 72 hours since Gemini 4 Argon launched, we processed around 25 trillion tokens.
对智能的需求是惊人的，而且仍然被低估了。
The demand for intelligence is incredible and still underappreciated.
归根结底，Genie 4 在世界模型方面仍然处于技术演示阶段。
At the end of the day, Genie 4 is still not out of the tech demo stage when it comes to world models.
但你能够感觉到，我们离那个技术演示变成实际应用的关键转折点已经越来越近了，而在世界模型、尤其是 Genie 这个案例上，那可能真的会是一个巨大的变化。
But you can feel us getting tantalizingly close to that inflection where tech demo becomes real-world application, and in the case of world models and Genie specifically, that could be a big change indeed.
今天的 AI Daily Brief 就到这里。
For now, that's gonna do it for today's AI Daily Brief.
感谢你的收听或观看，我们下次再见，Peace！
Appreciate you listening or watching as always, and until next time, peace!