Who Is Leading AI? Ruth Porat, Vimal Kapur, Mike Sicilia, Tareq Amin — FII9
谁在领导人工智能?Ruth Porat、Vimal Kapur、Mike Sicilia、Tareq Amin — FII9
Context | 背景
The global race for AI leadership has moved beyond a simple geopolitical contest and is now being fought on multiple fronts simultaneously: the raw compute power of sovereign states, the platform dominance of hyperscale corporations, and the disruptive potential of open-source models. In this complex new era, will true AI supremacy be defined by owning the foundational models, by controlling the data and distribution, or by setting the world's ethical and regulatory standards, and can any single entity—corporate or sovereign—truly win them all?
全球人工智能领导权的竞争已经超越了简单的地缘政治竞赛,现在正在多个战线上同时展开:主权国家的原始计算能力、超大规模企业的平台主导地位,以及开源模型的颠覆潜力。在这个复杂的新时代,真正的人工智能霸权是由拥有基础模型来定义,还是由控制数据和分发来定义,或是由设定全球伦理和监管标准来定义?单一实体——无论是企业还是主权国家——能否真正赢得所有这些?
The Opening: Setting the Stage | 开场:设定舞台
Good afternoon everybody and welcome to what is the first day of FII 9. I'm sure you've had a terrific morning. Let's crack on with this afternoon session. If 2023 was the year AI captured the world's imagination and 2024 was the year that it became a fact of life, then 2025 is the year that it is being built out at a staggering scale.
大家下午好,欢迎来到FII 9的第一天。我相信大家度过了一个美好的上午。让我们开始下午的会议吧。如果说2023年是人工智能捕获全世界想象力的一年,2024年是它成为生活现实的一年,那么2025年就是它以惊人规模建设的一年。
According to Gartner, worldwide spending on AI is expected to reach one and a half trillion dollars by the end of this year, fueled by what is, as we are all aware, an unprecedented surge in infrastructure investment to sustain AI compute. Today I'm joined by the leaders shaping this AI race from every angle: from Google's investments in data and computing centers to its foundational and enterprising models, to Oracle's data and cloud muscle, to Humane's full homegrown full stack, and Honeywell's industrial heft.
根据Gartner的数据,到今年年底,全球在人工智能上的支出预计将达到1.5万亿美元,这是由我们都知道的、为维持人工智能计算而进行的前所未有的基础设施投资激增所推动的。今天,我与从各个角度塑造这场人工智能竞赛的领导者们一起:从谷歌在数据和计算中心的投资到其基础和企业模型,到甲骨文的数据和云实力,到Humane的完全本土化全栈,以及霍尼韦尔的工业实力。
At a moment when the sheer velocity of the industry is being met with global concerns over return on investment and when an AI bubble is also part of this conversation, it is well worth unpacking where we are at, where this race stands and where it is headed.
在这个行业的惊人速度遇到全球对投资回报的担忧,以及人工智能泡沫也成为这场对话一部分的时刻,值得深入探讨我们现在的位置、这场竞赛的现状以及未来的方向。
I couldn't be better served with my panelists today. Tareq is the CEO of Humane. Ruth Porat is the president and CEO of Alphabet and Google. Mike is the CEO of Oracle. And Vimal Kapur is the chairman and CEO of Honeywell.
今天我的小组成员再合适不过了。Tareq是Humane的首席执行官。Ruth Porat是Alphabet和谷歌的总裁兼首席执行官。Mike是甲骨文的首席执行官。而Vimal Kapur是霍尼韦尔的董事长兼首席执行官。
Quick Fire Round: Is There an AI Bubble? | 快速问答:是否存在人工智能泡沫?
Question: There is a lot of talk of an AI bubble. Concerned?
问题:有很多关于人工智能泡沫的讨论。担心吗?
Tareq (Humane):
We're underestimating demand.
Tareq(Humane):我们低估了需求。
Ruth (Google):
No concerns. Long-term value creation is very big. Very early days of implementation in the upside.
Ruth(谷歌):不担心。长期价值创造非常巨大。实施仍处于非常早期的阶段,上升空间很大。
Mike (Oracle):
No concerns. There'll be winners and those who don't make it like any technology revolution, but very excited.
Mike(甲骨文):不担心。会有赢家,也会有失败者,就像任何技术革命一样,但我非常兴奋。
Humane's Vision: Building the Full AI Stack | Humane的愿景:构建完整的人工智能堆栈

Moderator: Let's start with you, Tareq, because Humane is chartering a Saudi end-to-end full stack: data centers, cloud, models, applications. I think it's important for this audience to have you briefly explain how this enormous and audacious project is going and what's the vision.
主持人:让我们从你开始,Tareq,因为Humane正在建设沙特端到端的全栈:数据中心、云、模型、应用程序。我认为让这个观众群了解这个庞大而大胆的项目如何进行以及愿景是什么很重要。
Tareq: Vicki and everyone, nice to be here today. Humane came together through the consolidation of public and private entities. When we envisioned what the company should do, maybe the easy thing would have been to say let's just focus on a single vertical, but nothing is easy in this journey. So we took an approach where Humane has to focus on the entire AI total value chain.
Tareq:Vicki和各位,很高兴今天来到这里。Humane是通过公共和私人实体的整合而形成的。当我们设想公司应该做什么时,也许简单的做法是说让我们只专注于单一垂直领域,但在这个旅程中没有什么是容易的。所以我们采取了一种方法,即Humane必须专注于整个人工智能的总价值链。
A company that builds gigawatt capacity data centers, a company that has great partnerships for cloud, diversified chipset supply. We needed to really be the owners of our own foundation model and build true value realization on the AI application. It came together. It feels like I've been doing this for 10 years, but honestly, it's been only like five months since May 5th when we launched the company during the US president's visit to Saudi Arabia.
一家建设千兆瓦容量数据中心的公司,一家拥有出色云合作伙伴关系、多样化芯片组供应的公司。我们确实需要成为自己基础模型的所有者,并在人工智能应用上建立真正的价值实现。这一切都汇聚在一起。感觉我做这件事已经10年了,但实际上,自5月5日我们在美国总统访问沙特阿拉伯期间推出公司以来,只有大约五个月的时间。
I would tell you, it's been the most exciting thing to see a company like Humane stand on its feet in a relatively short time, with an opportunity to even do large impact projects. The stuff we're doing with Google is a first opportunity to truly build an AI hub that is not serving only the captive demand in the kingdom but to really achieve the objective of attracting global offtakers to take advantage of what the country has to offer.
我想告诉你,看到像Humane这样的公司在相对较短的时间内站稳脚跟,甚至有机会做大型影响项目,这是最令人兴奋的事情。我们与谷歌所做的事情是真正建立人工智能中心的第一次机会,不仅服务于王国内的固定需求,而且真正实现吸引全球承购商利用这个国家所提供的优势的目标。
Everybody today in the world is chasing power, and Saudi Arabia has led the power through energy exports via oil. I am optimistic that Saudi Arabia could lead the world through energy exports via tokens.
今天世界上每个人都在追逐能源,沙特阿拉伯通过石油能源出口引领了能源。我乐观地认为,沙特阿拉伯可以通过代币的能源出口引领世界。
Google's Perspective: The Four Pillars of AI Value | 谷歌的视角:人工智能价值的四大支柱

Moderator: Ruth, let me bring you in here because I'm curious for your thoughts on Humane's proposition from Google's vantage point as an industry leader and hyperscaler, and for your broader thoughts on this moment of enormous global AI investment. How is Google working to ensure that this AI boom really translates to meaningful results?
主持人:Ruth,让我把你带进来,因为我很好奇你从谷歌作为行业领导者和超大规模企业的角度对Humane提议的看法,以及你对这个巨大的全球人工智能投资时刻的更广泛想法。谷歌如何确保这场人工智能热潮真正转化为有意义的结果?
Ruth: It's a great question. It goes back to our very early discussions with the predecessor of Humane and Tareq and the Humane team. When we think about the upside from AI and the reason I said it's very early is there are four major opportunities for us collectively.
Ruth:这是一个很好的问题。这可以追溯到我们与Humane的前身、Tareq和Humane团队的早期讨论。当我们思考人工智能的上升空间以及我说它还处于非常早期阶段的原因时,我们有四个主要的集体机会。
Advancing Science
推进科学
Better Social Services
更好的社会服务
Healthcare & Education
医疗保健与教育
Fortifying Cybersecurity
加强网络安全
Economic Uplift
经济增长
There's advancing science. There's better delivery of social services around healthcare and education. There's fortifying cyber security. And there's the economic uplift. To realize that takes investment in the right kind of infrastructure to power it. It takes more than that. It also takes delivering the products and services that enable everyone in the kingdom, everyone more broadly to actually realize both delivery of improved healthcare, education, other services and the economic upside.
有推进科学的机会。有更好地提供围绕医疗保健和教育的社会服务的机会。有加强网络安全的机会。还有经济增长的机会。要实现这一点,需要对正确类型的基础设施进行投资来为其提供动力。不仅如此。它还需要提供产品和服务,使王国中的每个人,更广泛地说每个人都能真正实现改善医疗保健、教育、其他服务的提供以及经济上升空间。
And importantly, it requires training people throughout the kingdom, training people around the globe so that everyone can participate in this upside. And I think one of the things we're really proud of when we launched this is we talked about all three of those elements. I think of it as a triangle. You need to pull all of them together in order to not just have the infrastructure which is obviously so core, but to ensure that you're building a society where everyone can benefit from it.
重要的是,它需要在整个王国培训人员,在全球范围内培训人员,以便每个人都能参与这个上升空间。我认为当我们启动这个项目时,我们真正引以为豪的事情之一是我们谈论了所有这三个要素。我把它想象成一个三角形。你需要把它们全部结合在一起,不仅要拥有显然如此核心的基础设施,而且要确保你正在建设一个每个人都能从中受益的社会。
And that's what we're proud of. We're pursuing each of those. There are proof points along the way. We made announcements this week about investments in education. I was at King Saud University. We launched a healthcare platform yesterday. So, it's really building the vision of what we were ruminating about quite some time ago. The speed of this is quite phenomenal.
这就是我们引以为豪的。我们正在追求每一个目标。在这个过程中有证明点。我们本周宣布了对教育的投资。我在沙特国王大学。我们昨天推出了一个医疗保健平台。所以,这真的是在建设我们很久以前就在思考的愿景。这个速度相当惊人。
AI vs. The Internet: Vint Cerf's Perspective | 人工智能与互联网:Vint Cerf的观点

Moderator: Ruth, just before I come to you, Mike, you've been imbued in this industry now for some time. You work with people who've been deeply involved in both the internet and AI. When you talk to those who've been at the frontier over what is decades now, what is ultimately more powerful, the internet or AI?
主持人:Ruth,在我找你之前,Mike,你在这个行业已经浸淫了一段时间。你与那些深度参与互联网和人工智能的人一起工作。当你与那些几十年来一直处于前沿的人交谈时,最终什么更强大,互联网还是人工智能?
Vint Cerf's Answer
Vint Cerf的答案
AI is more profound
人工智能更加深远
Ruth: Well, I am, as you know, privileged to work with someone who's described as the father of the internet. His name is Vint Cerf and if you Google him, that is actually what you will find. And so, I went to him because who better to answer that question, I think, than the father of the internet.
Ruth:嗯,如你所知,我有幸与被描述为互联网之父的人一起工作。他的名字是Vint Cerf,如果你谷歌他,这实际上就是你会发现的。所以,我去找他,因为我认为,谁能比互联网之父更好地回答这个问题呢?
AI leverages human creativity and human ingenuity and there's no limit to that.
人工智能利用人类的创造力和人类的聪明才智,而这是没有限制的。
And his answer was that AI is more profound for all of us. And the reason I thought was so beautiful and so consistent with FII and the spirit of it, he said AI leverages human creativity and human ingenuity and there's no limit to that. And to me, that's just a beautiful place to be. And it's why when we talk about the upside from AI, it's the economic upside, but it's also the ability to better deliver for our people, the types of work we do around education that we're talking about together as well as around healthcare, around food security and other elements.
他的答案是,人工智能对我们所有人来说更加深远。我认为这个原因非常美好,与FII及其精神非常一致,他说人工智能利用人类的创造力和人类的聪明才智,而这是没有限制的。对我来说,这只是一个美好的地方。这就是为什么当我们谈论人工智能的上升空间时,它是经济上的上升空间,但它也是为我们的人民更好地提供服务的能力,我们一起谈论的围绕教育的工作类型以及围绕医疗保健、食品安全和其他要素的工作。
Oracle's View: Real-World AI Impact | 甲骨文的观点:现实世界的人工智能影响

Moderator: Mike, Oracle is a leading partner here in this region in data and cloud infrastructure both in the kingdom and through Stargate UAE. You've also made headlines recently for your mega contracts to train OpenAI, Meta and XAI models. What tangible impacts are you seeing from this AI push?
主持人:Mike,甲骨文是该地区数据和云基础设施的领先合作伙伴,无论是在王国还是通过阿联酋的Stargate。你最近也因为与OpenAI、Meta和XAI模型的大型合同而成为头条新闻。你从这次人工智能推动中看到了什么具体影响?
Mike: Well, we're actually seeing excellent impact already by combining these large language models or in some cases frontier models with existing applications. So as Ruth mentioned in healthcare, for example, we have now for about a year and a half had a cloud service called AI service available that's embedded within our electronic medical records, electronic health records, clinical applications.
Mike:嗯,通过将这些大型语言模型或在某些情况下的前沿模型与现有应用程序结合起来,我们实际上已经看到了出色的影响。所以正如Ruth在医疗保健中提到的,例如,我们现在大约一年半以来一直有一个名为AI服务的云服务,该服务嵌入在我们的电子病历、电子健康记录、临床应用程序中。
100
Minutes Returned
节省的分钟数
Per shift for doctors and nurses
每班次为医生和护士节省
40%
Energy Reduction
能源减少
In restaurant operations
在餐厅运营中
We've had customers report that on an average shift for a doctor or a nurse or some provider, they've actually returned 100 minutes back to that doctor. That was time that they had to spend with the systems, time that they had to spend with obligatory documentation. And as everybody said, we're still in very early days. So I actually think that there's already dramatic impact with the AI that's set to go, set to deliver.
我们的客户报告说,在医生或护士或某些提供者的平均轮班中,他们实际上为该医生节省了100分钟。这是他们必须花在系统上的时间,必须花在强制性文档上的时间。正如每个人所说,我们仍处于非常早期的阶段。所以我实际上认为,已经准备好、准备交付的人工智能已经产生了巨大的影响。
We see the same thing in the banking industry. So AI is a perfect complement to existing core banking systems to help automate financial crimes investigations, anti-money laundering investigations, all of which are very laborious processes, very manual. I think the next piece and the even more exciting piece is taking all the existing private data—healthcare data, banking data, all of which should and will remain private data—and marrying that together with these models, whether they're traditional large language models or frontier models.
我们在银行业看到同样的情况。所以人工智能是现有核心银行系统的完美补充,有助于自动化金融犯罪调查、反洗钱调查,所有这些都是非常费力的过程,非常手动。我认为下一个更令人兴奋的部分是获取所有现有的私人数据——医疗保健数据、银行数据,所有这些都应该而且将保持为私人数据——并将其与这些模型结合在一起,无论它们是传统的大型语言模型还是前沿模型。
And coming up with either an inferencing or retrieval augmented generation strategy where you're taking the best of these very large high-powered models and powering them with existing private data. Then I think you get into the healthcare industry, for example, which is one that everybody can always relate to—clinical assistance, clinical diagnosis and clinical intervention at the point of care, which traditionally would take providers quite a lot of time to research on their own time.
并提出推理或检索增强生成策略,在该策略中,你采用这些非常大的高功率模型中的最佳模型,并用现有的私人数据为它们提供动力。然后我认为你进入了医疗保健行业,例如,这是每个人都可以联系到的行业——在护理点的临床协助、临床诊断和临床干预,传统上这将需要提供者花费相当多的时间来研究他们自己的时间。
But how do you make intervention at the point of care? You need gigawatt scale data centers. You need sovereign infrastructure. You need to get all the data in one place.
但是你如何在护理点进行干预?你需要千兆瓦规模的数据中心。你需要主权基础设施。你需要将所有数据放在一个地方。
But how do you make intervention at the point of care? You need gigawatt scale data centers. You need sovereign infrastructure. You need to get all the data in one place. And I think that all of that is very neatly coming together to have just major impact for businesses, for governments and most importantly for humanity.
但是你如何在护理点进行干预?你需要千兆瓦规模的数据中心。你需要主权基础设施。你需要将所有数据放在一个地方。我认为所有这些都非常整齐地汇集在一起,对企业、政府,最重要的是对人类产生重大影响。
Honeywell: Industrial AI's Unique Challenges | 霍尼韦尔:工业人工智能的独特挑战

Moderator: Vimal, there's an untold story of the industrial sector, isn't there, and why it moves slower to transform. Can you just explain why and what Honeywell is doing to gain an edge in the competitive industrial AI space?
主持人:Vimal,工业部门有一个不为人知的故事,不是吗,为什么它转型速度较慢。你能解释一下为什么以及霍尼韦尔正在做什么来在竞争激烈的工业人工智能领域获得优势吗?
Vimal: So the physical AI and industrial AI builds on all the work done by colleagues on this. If they don't do their work on infrastructure, core capability of AI, we won't even exist. But where the industrial world—think about a hospital, think about an airport, think about a refinery—what's different is, first, the data is hard to get. The data friction is very high because the data is captured into very older systems, the protocols are unknown, so we need to solve for that.
Vimal:所以物理人工智能和工业人工智能是建立在同事们在这方面所做的所有工作之上的。如果他们不做基础设施、人工智能核心能力的工作,我们甚至都不会存在。但是工业世界在哪里——想想医院,想想机场,想想炼油厂——不同的是,首先,数据很难获得。数据摩擦非常高,因为数据被捕获到非常旧的系统中,协议是未知的,所以我们需要解决这个问题。
Data Friction
数据摩擦
Hard to access legacy systems
难以访问遗留系统
Domain Knowledge
领域知识
Unique expertise required
需要独特的专业知识
Determinism
确定性
Must work 100% of the time
必须100%工作
Second challenge is domain knowledge is very unique because each of them are very different, so just data alone is not a path to find the solution. You need to know what problem to solve for. And finally, our solutions have to be what we call deterministic—means they exactly have to work, not almost, not 95% probability, it has to work 100%. We call it six nines.
第二个挑战是领域知识非常独特,因为每一个都非常不同,所以仅仅数据本身不是找到解决方案的途径。你需要知道要解决什么问题。最后,我们的解决方案必须是我们所说的确定性的——意味着它们必须完全工作,不是几乎,不是95%的概率,它必须100%工作。我们称之为六个九。
So when you take those 3Ds—the data friction, the domain knowledge and determinism—the physical AI has to tweak the infrastructure which exists in larger language models to create the outcomes. But the good news is the outcomes are well known. Our customers, the economic value creation point with Ruth made, they are well defined. There are three problems to solve: the asset life, the operational excellence that you run it better, economic value creation, and the skill enhancement of the people.
所以当你采用这三个D——数据摩擦、领域知识和确定性——物理人工智能必须调整存在于更大语言模型中的基础设施来创造结果。但好消息是结果是众所周知的。我们的客户,与Ruth一起提出的经济价值创造点,它们都有明确的定义。有三个问题需要解决:资产寿命、你更好地运营它的卓越运营、经济价值创造,以及人员的技能提升。
So if we can solve the 3Ds, if we can take care of data friction, if you have the right domain knowledge and if you build deterministic outcome, we know what problem to solve for and that's what excites us. The journey here is very exciting, the value creation is real, and I see our customers feeling urgency to drive that, whether customers in Saudi Arabia or rest of the world. So I really feel excited about the opportunity ahead.
所以如果我们能解决这三个D,如果我们能处理数据摩擦,如果你有正确的领域知识,如果你建立确定性结果,我们就知道要解决什么问题,这就是让我们兴奋的地方。这里的旅程非常令人兴奋,价值创造是真实的,我看到我们的客户感到推动这一点的紧迫性,无论是沙特阿拉伯的客户还是世界其他地方的客户。所以我真的对未来的机会感到兴奋。
Saudi Arabia's Ambitious AI Goals | 沙特阿拉伯雄心勃勃的人工智能目标

Moderator: I mean that sense of urgency is no better borne out than here in the kingdom and the work that you are doing. Tareq, Saudi is vying to grow from handling less than 1% of the world's AI workload to 6%. That's the goal. That would be trailing only the United States and China in terms of global leadership. You're looking to be number three. That's the vision. You argue that Humane's competitive edge is power, AI's insatiable demand for energy that is of course abundant in this kingdom. I just have to ask, is that enough to get you there?
主持人:我的意思是,这种紧迫感在王国和你正在做的工作中体现得最好。Tareq,沙特正在争取从处理不到1%的世界人工智能工作负载增长到6%。这就是目标。在全球领导力方面,这将仅次于美国和中国。你希望成为第三名。这就是愿景。你认为Humane的竞争优势是能源,人工智能对能源的贪得无厌的需求,而这在这个王国当然是丰富的。我只是想问,这足以让你到达那里吗?
6%
Target AI Workload
目标人工智能工作负载
From less than 1%
从不到1%
Tareq: Well, I think when you're building a business, there are many ingredients you start with. I had the opportunity to do amazing things in my previous roles in Japan and India and the United States. There is one commonality you need regardless of saying I have power or not. The first and most critical thing is the foundation of people and organization.
Tareq:嗯,我认为当你建立一个企业时,你有很多成分要开始。我有机会在我之前在日本、印度和美国的角色中做了一些了不起的事情。无论你说我有权力还是没有权力,你都需要一个共同点。第一个也是最关键的事情是人员和组织的基础。
So my surprise—I mean I'm new to Saudi Arabia—when I came here my first thing that I thought I have to worry about is the talent and the depth of talent. Like any other leader on this group, I feel blessed every day I go to the office and I have researchers in my team that I learn from every day. I have executors that know how to build infrastructure.
所以我的惊讶——我的意思是我是沙特阿拉伯的新人——当我来到这里时,我认为我必须担心的第一件事是人才和人才的深度。像这个团队中的任何其他领导者一样,我每天去办公室都感到幸运,我的团队中有研究人员,我每天都从他们那里学习。我有知道如何建设基础设施的执行者。
So one aspect is flawless execution and project. I think I agree with Mike. First of all, I really tell you we are completely underestimating the demand. So the sense of urgency Humane has is around the build. 70% of my job I feel like I'm a real estate developer. It's ensuring that you secure the land, power, connectivity.
所以一个方面是完美的执行和项目。我认为我同意Mike。首先,我真的告诉你我们完全低估了需求。所以Humane的紧迫感围绕着建设。我70%的工作感觉像是一个房地产开发商。它确保你确保土地、电力、连接性。
I have publicly stated that I truly believe Saudi Arabia has a potential and opportunity to be the third largest AI infrastructure provider outside of China and the United States.
我公开表示,我真的相信沙特阿拉伯有潜力和机会成为中国和美国以外的第三大人工智能基础设施提供商。
I have publicly stated that I truly believe Saudi Arabia has a potential and opportunity to be the third largest AI infrastructure provider outside of China and the United States. If you talk to Mike and even Ruth, they will tell you very clearly finding power is not a trivial thing. It's not like I want to build a data center and like magic now you have gigawatt of capacity available.
我公开表示,我真的相信沙特阿拉伯有潜力和机会成为中国和美国以外的第三大人工智能基础设施提供商。如果你和Mike甚至Ruth交谈,他们会非常清楚地告诉你,找到电力不是一件微不足道的事情。这不像我想建造一个数据中心,像魔术一样,现在你有千兆瓦的容量可用。
So what has happened in Humane now, and I mentioned to you when we were outside, this is a phenomenal opportunity in which public and private are partnering together under the direction of His Royal Highness, Ministry of Energy. If you go and see a picture where a collection of 16 government entities and Humane are sitting together in one room trying to find and solve for where is available power that I could tap into that doesn't need substation buildout.
所以现在在Humane发生的事情,我在我们在外面时提到过,这是一个非凡的机会,在国王殿下、能源部的指导下,公共和私人正在合作。如果你去看一张照片,16个政府实体和Humane的集合坐在一个房间里,试图找到并解决我可以利用的可用电力在哪里,不需要变电站建设。
I am motivated by time. Time is my worst enemy. And luckily we have solved for power and we don't need to build additional capacity for the power at least up till 2030, and of course post-2030 I'm very confident our large scale campus will be ready with the capacity that we need.
我被时间激励。时间是我最大的敌人。幸运的是,我们已经解决了电力问题,我们不需要至少在2030年之前为电力建造额外的容量,当然在2030年之后,我非常有信心我们的大型校园将准备好我们需要的容量。
So whoever gets to this path where an affordable low-cost power, a reliable secure structure for data governance—it's an ingredient that I feel could present to you as a compelling alternative for AI native companies. And then lastly, we cannot really forget the importance of partnerships. It's really important. So we're not doing this alone.
所以无论谁走上这条道路,在那里有负担得起的低成本电力,一个可靠的安全的数据治理结构——这是我觉得可以作为人工智能原生公司的一个引人注目的替代方案呈现给你的一个成分。最后,我们真的不能忘记合作伙伴关系的重要性。这真的很重要。所以我们不是独自做这件事。
We have a lot of support that we started with, and I mentioned the early discussion. It's been almost a year ago that we talked to Ruth and the team about this crazy idea. It was crazy. When we started this, saying "Hey, what do you think about the idea of hosting non-Saudi workloads in the kingdom?" Just think about those early days conversation and where we are today. But I'm optimistic. I mean I understand the job and the task in front of me. Yes it comes with its stress, but I've done enough infrastructure buildout in my life. This is now an area where we have to flawlessly execute the business offtake.
我们有很多我们开始的支持,我提到了早期的讨论。大约一年前,我们与Ruth和团队谈论了这个疯狂的想法。这很疯狂。当我们开始这个时,说"Hey, what do you think about the idea of hosting non-Saudi workloads in the kingdom?"只是想想那些早期的对话以及我们今天在哪里。但我很乐观。我的意思是我理解我面前的工作和任务。是的,它带来了压力,但我在生活中做了足够的基础设施建设。现在这是一个我们必须完美执行业务承购的领域。
Export Controls and US Relations | 出口管制和美国关系

Moderator: You launched Humane while Donald Trump was here back in May. There are still export controls to a degree on some of the most important assets that you need here in the kingdom. You've diversified away from just, for example, Nvidia chips—those aren't necessarily available to you in the abundance that you need. How do you describe that relationship with the US and the Trump administration? Very specifically, we know that the three pillars of AI were agreed, there were agreements to agree for a US-Saudi partnership around these significant AI pillars. What are you hoping to achieve?
主持人:你在唐纳德·特朗普在这里的五月推出了Humane。在你在王国需要的一些最重要的资产上仍然存在一定程度的出口管制。你已经从仅仅例如英伟达芯片多样化——这些对你来说不一定以你需要的丰度可用。你如何描述与美国和特朗普政府的关系?非常具体地说,我们知道人工智能的三大支柱已经达成一致,有关于围绕这些重要人工智能支柱的美国-沙特合作伙伴关系的协议。你希望实现什么?
150
Countries Served
服务的国家
From Humane data centers today
今天从Humane数据中心
5%
Saudi Traffic
沙特流量
95% from international markets
95%来自国际市场
Tareq: Yeah, I mean a couple of things. Today, not so well known by the way to the world, but in fact today Saudi Arabia is serving 150 countries from Humane data center, in which 5% of the traffic comes from Saudi, the rest is coming from other international markets. We're obsessed about the concept of democratization of infrastructure, so we look at every possible alternative that could deliver to our customer reliable, secure, ultra-low-cost inferencing.
Tareq:是的,我的意思是几件事。今天,顺便说一下,世界上不太为人所知,但实际上今天沙特阿拉伯正在从Humane数据中心为150个国家提供服务,其中5%的流量来自沙特,其余来自其他国际市场。我们痴迷于基础设施民主化的概念,所以我们寻找每一个可能的替代方案,可以为我们的客户提供可靠、安全、超低成本的推理。
And we got approval on the export controls because our cause was easy for them to understand, and we also understand what the US government's concerns and requirements are. So the process on a larger G2G discussion is ongoing. I am feeling extremely optimistic about our approach. It was extremely important for us to explain to the US government what is Humane, who's our customers, how we manage tenants, how we manage security in the data center, and we've addressed in a very robust way the requirements.
我们获得了出口管制的批准,因为我们的理由对他们来说很容易理解,我们也理解美国政府的担忧和要求是什么。所以更大的G2G讨论的过程正在进行中。我对我们的方法感到非常乐观。对我们来说,向美国政府解释Humane是什么、我们的客户是谁、我们如何管理租户、我们如何管理数据中心的安全非常重要,我们以非常有力的方式满足了要求。
The partnerships with hyperscalers, hopefully also with Mike and the team, we'll do something really meaningful with Oracle as well. But I think we have a really good solution architecture that makes it compelling for us to get the necessary approval to support the mission and the vision and the objective of what Humane wants to do.
与超大规模企业的合作伙伴关系,希望也与Mike和团队一起,我们也将与甲骨文做一些真正有意义的事情。但我认为我们有一个非常好的解决方案架构,使我们能够获得必要的批准来支持Humane想要做的使命、愿景和目标。
Regional AI Competition: UAE and Saudi Arabia | 区域人工智能竞争:阿联酋和沙特阿拉伯

Moderator: Talk about working with Mike. Come back to both of you. Mike, I just want to bring you in here because you are working with the UAE on the Stargate UAE project as well. When we talk about who will lead the race in AI, where do you see this region? I mean Tareq wants Saudi to be number three. Well, the UAE is looking at a similar scenario. What do you see as the opportunity in this region?
主持人:谈谈与Mike合作。回到你们两个。Mike,我只是想把你带到这里,因为你也在与阿联酋合作进行Stargate阿联酋项目。当我们谈论谁将领导人工智能竞赛时,你在哪里看到这个地区?我的意思是Tareq希望沙特成为第三名。嗯,阿联酋正在考虑类似的情景。你认为这个地区的机会是什么?
Mike: Yeah, I think that the kingdom is in a wonderful position, as Tareq said, to provide turnkey AI and embedded business services to the rest of the world. In many places in the world, it's not going to be practical to build or source the power quickly. And even if—keeping in mind that if you do it once, that's not enough. You have to do it twice. It has to be fully redundant and sometimes even three times to make it fully dependable.
Mike:是的,我认为正如Tareq所说,王国处于一个很好的位置,可以向世界其他地方提供交钥匙人工智能和嵌入式业务服务。在世界上许多地方,快速建造或采购电力是不切实际的。即使——记住,如果你做一次,那还不够。你必须做两次。它必须完全冗余,有时甚至三次才能使其完全可靠。
So you've got excellent process, excellent government, excellent governance, world-class security, wonderful business applications, and lots of organizations that are surrounding Humane and others to host their business applications on this infrastructure. And I think the idea of sovereignty is going to be—sovereignty doesn't necessarily mean that it has to run inside a country's borders.
所以你有出色的流程、出色的政府、出色的治理、世界级的安全、出色的业务应用程序,以及围绕Humane和其他人的大量组织,在这个基础设施上托管他们的业务应用程序。我认为主权的想法将是——主权不一定意味着它必须在一个国家的边界内运行。
I think of sovereignty in this model like you think of an embassy. You would think of an embassy being here in the kingdom which is a sovereign entity. I think we can achieve the same thing with data centers.
我在这个模型中想到主权就像你想到大使馆一样。你会想到一个大使馆在王国,这是一个主权实体。我认为我们可以用数据中心实现同样的事情。
I think of sovereignty in this model like you think of an embassy. You would think of an embassy being here in the kingdom which is a sovereign entity. I think we can achieve the same thing with data centers, particularly as we focus on small form factor data centers complete with robust AI infrastructure. We can deliver that to other countries at a price point that's highly attractive, and as you said, over 100 countries are certainly possible customers for that, and I think it makes tremendous sense.
我在这个模型中想到主权就像你想到大使馆一样。你会想到一个大使馆在王国,这是一个主权实体。我认为我们可以用数据中心实现同样的事情,特别是当我们专注于配备强大人工智能基础设施的小型数据中心时。我们可以以非常有吸引力的价格点将其交付给其他国家,正如你所说,超过100个国家肯定是可能的客户,我认为这非常有意义。
The ROI Challenge: Bridging the Implementation Gap | 投资回报率挑战:弥合实施差距

Moderator: Let me bring you in Ruth here. It was an MIT study I think last summer that initially sparked these fears of an AI bubble. It found at the time that 95% of businesses saw zero return on investment from their generative AI projects. And again, early days of course. Why do you think organizations are struggling to implement AI and what role can Google have, for example, in bridging that gap?
主持人:让我把你带进来Ruth。我认为去年夏天的麻省理工学院研究最初引发了对人工智能泡沫的这些恐惧。当时发现,95%的企业从其生成式人工智能项目中看到零投资回报。当然,还是早期。你为什么认为组织在实施人工智能方面遇到困难,例如,谷歌可以在弥合这一差距方面发挥什么作用?
95%
Zero ROI
零投资回报率
From generative AI projects (MIT study)
来自生成式人工智能项目(麻省理工学院研究)
Ruth: So I think what's fascinating about this time is when you look at the pace of science and breakthroughs, it's moving very quickly and it's breathtaking. As an example, our colleagues were awarded the Nobel Prize last year for chemistry in something called AlphaFold, which is predicting the 3D protein structure, viewed as the greatest contribution to drug discovery in our lifetime.
Ruth:所以我认为这个时代令人着迷的是,当你看科学和突破的步伐时,它移动得非常快,令人叹为观止。例如,我们的同事去年因化学获得诺贝尔奖,称为AlphaFold,它预测3D蛋白质结构,被视为我们一生中对药物发现的最大贡献。
01
4 Years to Break 50-Year Challenge
4年打破50年挑战
AlphaFold achievement
AlphaFold成就
02
3.5 Million Scientists Using It
350万科学家使用它
Global impact today
今天的全球影响
03
Nobel Prize Winner
诺贝尔奖获得者
Recognition of breakthrough
突破的认可
It took them from four years to deciding to go after this 50-year grand challenge to accomplishing it, to open-sourcing it. Today three and a half million scientists are using it around the globe to advance science, and they won the Nobel Prize. Four years to break a 50-year grand challenge. This year, one of our colleagues won the Nobel for physics, and this was in quantum computing.
从他们决定去追求这个50年的重大挑战到完成它、开源它,花了四年时间。今天,全球有350万科学家正在使用它来推进科学,他们获得了诺贝尔奖。四年时间打破了一个50年的重大挑战。今年,我们的一位同事获得了物理学诺贝尔奖,这是在量子计算方面。
And we just announced last week something stunning. I've been waiting for years—I keep asking our head of quantum AI, when are we going to see something commercial? And this is the application also to health where there's so many exciting things going on in imaging and MRIs. And so we can see within 3 to 5 years there's going to be a commercial application. That is moving at breathtaking speed.
我们上周刚刚宣布了一些惊人的事情。我已经等了很多年了——我一直在问我们的量子人工智能负责人,我们什么时候会看到商业的东西?这也是对健康的应用,在成像和MRI方面有很多令人兴奋的事情正在发生。所以我们可以看到在3到5年内将会有一个商业应用。这正以惊人的速度前进。
I think what's fascinating is that change management—it starts with what's the prioritization and the mental model. I keep coming back to really builds on the life science and health experiences. One because they're here and two because they're personal to all of us. And in particular I go back to cancer because in America 40% of Americans will be diagnosed with cancer in their lifetime. It's not that different sadly elsewhere around the world.
我认为令人着迷的是变革管理——它始于优先级和心智模型。我一直回到真正建立在生命科学和健康经验上。一是因为它们在这里,二是因为它们对我们所有人来说都是个人的。特别是我回到癌症,因为在美国,40%的美国人将在他们的一生中被诊断患有癌症。不幸的是,在世界其他地方也没有太大不同。
And if you look at the three buckets of applications, it provides a framework for people to think about where do I start? And the problem is just dabbling with a chatbot, you're not going to get the ROI. Dabbling with a chatbot, you have to fundamentally rethink your business.
如果你看这三个应用程序类别,它为人们提供了一个框架来思考我从哪里开始?问题是只是涉猎聊天机器人,你不会得到投资回报率。涉猎聊天机器人,你必须从根本上重新思考你的业务。
And if you look at the three buckets of applications, it provides a framework for people to think about where do I start? And the problem is just dabbling with a chatbot, you're not going to get the ROI. Dabbling with a chatbot, you have to fundamentally rethink your business.
如果你看这三个应用程序类别,它为人们提供了一个框架来思考我从哪里开始?问题是只是涉猎聊天机器人,你不会得到投资回报率。涉猎聊天机器人,你必须从根本上重新思考你的业务。
The Three-Bucket Framework for AI ROI | 人工智能投资回报率的三桶框架
Ruth: So first on innovation. I've already said it. We're seeing it in drug discovery and other elements of it with massive breakthroughs. When Demis—our colleague Demis Hassabis who won the Nobel—was asked how did you decide to do that? He said, "Well, with AI I can take on the most intractable problems I've ever seen," which is what each one of us should be thinking about in our business.
Ruth:首先是创新。我已经说过了。我们在药物发现和其他方面看到了巨大的突破。当Demis——我们的同事Demis Hassabis获得诺贝尔奖——被问到你是如何决定这样做的?他说,"嗯,有了人工智能,我可以解决我见过的最棘手的问题",这是我们每个人在我们的业务中应该考虑的。
Innovation
创新
Tackle intractable problems
解决棘手的问题
Risk Management
风险管理
Early detection saves lives
早期发现挽救生命
Operating Leverage
运营杠杆
Reduce administrative tasks
减少管理任务
The second area in healthcare is risk management. Early detection saves lives. Early detection reduces the magnitude of the intervention in cancer. We're seeing that with early detection of cancer—you can see the cells earlier, metastatic cancer forming earlier than otherwise. It is absolutely staggering. And when I've reflected on it, my view is that's finding the proverbial needle in the haystack.
医疗保健的第二个领域是风险管理。早期发现可以挽救生命。早期发现减少了癌症干预的规模。我们看到通过早期发现癌症——你可以更早地看到细胞,转移性癌症比其他情况更早地形成。这绝对令人震惊。当我反思它时,我的观点是这是在大海捞针中找到众所周知的针。
That is what all of us need to do on cybersecurity or fraud detection. It's the same thing. So you think of that as number two. And the third one you've already commented on. It's the operating leverage you get using AI, in particular agentic AI, where you can reduce the administrative task, let nurses focus on what they want—the patient—you can free up 30% of their time.
这是我们所有人在网络安全或欺诈检测方面需要做的。这是同一件事。所以你认为这是第二个。你已经评论过第三个了。这是你使用人工智能获得的运营杠杆,特别是代理人工智能,在那里你可以减少管理任务,让护士专注于他们想要的——病人——你可以释放他们30%的时间。
That mental model is astounding because again everyone in this room has that ability. And then what's exciting is when you can reduce the unit economics in your business, it unlocks innovation opportunities. The best example—the number of people are talking about here—is if you reduce unit economics in financial services and you can cover everybody with the quality that you used to be only able to cover a few. What does that do for wealth management or for asset management? It opens entirely new markets.
这个心智模型令人震惊,因为这个房间里的每个人都有这种能力。然后令人兴奋的是,当你可以降低业务中的单位经济学时,它会解锁创新机会。最好的例子——很多人在这里谈论——是如果你降低金融服务的单位经济学,你可以用你过去只能覆盖少数人的质量覆盖每个人。这对财富管理或资产管理有什么作用?它打开了全新的市场。
So I think the problem we have seen is people need that mental model: innovation, risk management, operating expense. Pick a lane, go hard at it, and then you'll see the upside.
所以我认为我们看到的问题是人们需要这个心智模型:创新、风险管理、运营费用。选择一条车道,努力去做,然后你会看到上升空间。
So I think the problem we have seen is people need that mental model: innovation, risk management, operating expense. Pick a lane, go hard at it, and then you'll see the upside.
所以我认为我们看到的问题是人们需要这个心智模型:创新、风险管理、运营费用。选择一条车道,努力去做,然后你会看到上升空间。
Reimagining Work: Honeywell's Approach | 重新想象工作:霍尼韦尔的方法

Moderator: Vimal, you're nodding away sagely here.
主持人:Vimal,你在这里睿智地点头。
Vimal: Yeah. So, I agree with Ruth. I think one of the things we have been, as a practitioner, wearing my other hat, how do we as a company like us with 100,000 employees adopt AI and where does ROI come from? What lesson we learned is really go back to the basic applying lean principles. You have to reimagine work.
Vimal:是的。所以,我同意Ruth。我认为作为一个实践者,我戴着我的另一顶帽子,我们作为一家像我们这样拥有10万名员工的公司如何采用人工智能,投资回报率从哪里来?我们学到的教训是真正回到应用精益原则的基础。你必须重新想象工作。
Old Way
旧方式
Give tools to people
给人们工具
Save some time
节省一些时间
Same workflow
相同的工作流程
New Way
新方式
Reimagine the process
重新想象过程
Redefine workflow
重新定义工作流程
Lean the work
精益工作
If you give the tool to the people, they will save some time, but that's work being performed the same way. So to me you have to reimagine the process to say if all these agents existed, how the workflow will happen differently. And unless the workflow is not redefined, it's the classical lean manufacturing thinking—you have to lean the work, you have to lean the process, and that's where the benefits will come.
如果你把工具给人们,他们会节省一些时间,但那是以同样的方式执行的工作。所以对我来说,你必须重新想象过程,如果所有这些代理都存在,工作流程将如何不同。除非工作流程没有重新定义,否则这是经典的精益制造思维——你必须精益工作,你必须精益流程,这就是收益的来源。
Real-World Industrial AI Success | 现实世界的工业人工智能成功

Moderator: I mean Honeywell's long-term strategy hinges on digitally transforming industries through AI and data analytics. Your partnerships with hyperscalers are essential in unleashing that data. I just wonder what the challenges are that you are navigating in that effort to build out autonomous self-sufficient systems.
主持人:我的意思是霍尼韦尔的长期战略取决于通过人工智能和数据分析数字化转型行业。你与超大规模企业的合作伙伴关系对于释放该数据至关重要。我只是想知道你在建设自主自给系统的努力中正在应对哪些挑战。
Vimal: So to me the infrastructure availability or capability of large language model is not a limitation. As I said, we have a lot of resources available from the providers. The challenge is to build economic value which can be adopted at scale. And when we are able to prove that, then market adoption goes high.
Vimal:所以对我来说,大型语言模型的基础设施可用性或能力不是限制。正如我所说,我们有很多来自提供商的资源可用。挑战是建立可以大规模采用的经济价值。当我们能够证明这一点时,市场采用率就会很高。
So what we have worked over the last three or four years is the data friction problem. We can connect any asset, industrial asset in a day. I can connect this whole hotel in a day. So you say why would you want to connect this? What's the real problem you are solving? You really want to know the assets here. What's how much money is being spent on the maintenance? How many people are there and do we really need them? But unless you don't map it, you can't really answer the question.
所以我们在过去三四年里所做的是数据摩擦问题。我们可以在一天内连接任何资产、工业资产。我可以在一天内连接整个酒店。所以你说为什么你想连接这个?你正在解决的真正问题是什么?你真的想知道这里的资产。在维护上花费了多少钱?有多少人,我们真的需要他们吗?但除非你不绘制它,否则你无法真正回答这个问题。
40%
Energy Reduction
能源减少
In quick service restaurants
在快餐店
600
Locations Connected
连接的位置
Across the UK
遍布英国
So increasingly as we are deploying our capabilities, we are finding clear use cases which create large economic value. And I'll give an example which many people relate to. We did deployment of our IoT platform in a quick service restaurant chain where all of us eat—600 of them in UK. You won't imagine a quick service restaurant being an opportunity for cost reduction. You say what's there to reduce? What exists there?
所以随着我们部署我们的能力,我们正在发现创造大量经济价值的明确用例。我会给一个很多人都能理解的例子。我们在一个我们都吃的快餐连锁店部署了我们的物联网平台——英国有600家。你不会想象一个快餐店是降低成本的机会。你说有什么可以减少的?那里存在什么?
But if you think they have to keep lights on, you have to maintain temperature, you have a kitchen, there's a lot of fryers there, you have refrigerators. So when you start adding, there's a lot of equipment, there's a lot of energy. Once we connected everything, we were able to find first thing the kitchen equipment temperature was wrong, so your meat was not being fried at the right temperature even though the recipe said. It's a simple thing but customer dissatisfaction.
但如果你认为他们必须保持灯亮,你必须保持温度,你有一个厨房,那里有很多油炸锅,你有冰箱。所以当你开始添加时,有很多设备,有很多能源。一旦我们连接了所有东西,我们能够首先发现厨房设备温度是错误的,所以你的肉没有在正确的温度下油炸,即使食谱说。这是一个简单的事情,但客户不满意。
But since we connected, the energy reduction is around 40%. Why? Because each human was running this equipment individually and there was no oversight. Nobody knows. How do you manage 600 of them spread over the country? Now we can do that and that's economic value creation.
但自从我们连接以来,能源减少约为40%。为什么?因为每个人都在单独运行这个设备,没有监督。没有人知道。你如何管理分布在全国的600个?现在我们可以做到这一点,这就是经济价值创造。
But since we connected, the energy reduction is around 40%. Why? Because each human was running this equipment individually and there was no oversight. Nobody knows. How do you manage 600 of them spread over the country? Now we can do that and that's economic value creation.
但自从我们连接以来,能源减少约为40%。为什么?因为每个人都在单独运行这个设备,没有监督。没有人知道。你如何管理分布在全国的600个?现在我们可以做到这一点,这就是经济价值创造。
Enterprise AI: Data Governance is Key | 企业人工智能:数据治理是关键

Moderator: Mike, what do you see as the key barriers to AI adoption in B2B, in the B2G space? Is it what I've just described? Is it regulatory pressure? Is it data sovereignty? Is it raw compute power? All of the above or something entirely different?
主持人:Mike,你认为在B2B、B2G空间中采用人工智能的主要障碍是什么?是我刚才描述的吗?是监管压力吗?是数据主权吗?是原始计算能力吗?所有这些还是完全不同的东西?
Mike: Yeah. Well, I think all of the above and some of the things are probably out of the hands of a corporation. But if you think about enterprise application of AI, lots of things you can do today with large language models off the shelf—easier to write narratives, easier to write job descriptions, should be easier to hire people and increase process.
Mike:是的。嗯,我认为以上所有这些,有些事情可能超出了公司的控制范围。但如果你考虑企业应用人工智能,今天你可以用现成的大型语言模型做很多事情——更容易编写叙述,更容易编写职位描述,应该更容易雇用人员并提高流程。
But the next big piece of this, as I mentioned earlier, is how do you take all that enterprise data and how do you make that relevant at the time of an action, whether it's a customer interaction whether it's a back office interaction? And for that I think you need a data governance strategy.
但正如我之前提到的,接下来的大部分是你如何获取所有企业数据,以及你如何在行动时使其相关,无论是客户互动还是后台互动?为此,我认为你需要一个数据治理策略。
01
Find All Data
找到所有数据
Where is it? How is it organized?
它在哪里?它是如何组织的?
02
Consolidate Data
整合数据
Get it all in one place
把它全部放在一个地方
03
Vectorize Data
矢量化数据
Make it applicable for inferencing
使其适用于推理
So some experience we have in working with corporations is the very first step is trying to just find where is all the data, where is it, how's it organized, how many different databases do you have, and how do we get all that data? And then how do we get it and vectorize it? Because if we can vectorize it, we can actually make it applicable for inferencing through augmented generation and all these things that need to happen.
所以我们在与企业合作方面的一些经验是,第一步是试图找到所有数据在哪里,它在哪里,它是如何组织的,你有多少个不同的数据库,以及我们如何获得所有这些数据?然后我们如何获得它并对其进行矢量化?因为如果我们可以对其进行矢量化,我们实际上可以通过增强生成和所有这些需要发生的事情使其适用于推理。
So I think just good old data governance is a very good first step as corporations and governments for that matter prepare themselves to take the next step, which is taking generic models and making them very germane, very specific to the business process, the clinical process or the government process, whatever it may be.
所以我认为,作为企业和政府,准备好迈出下一步,将通用模型变得非常相关、非常具体于业务流程、临床流程或政府流程,无论它可能是什么,良好的数据治理是一个非常好的第一步。
So I think just good old data governance is a very good first step as corporations and governments for that matter prepare themselves to take the next step, which is taking generic models and making them very germane, very specific to the business process, the clinical process or the government process, whatever it may be. And really I think get to an outcome which is not measured in just saving time, saving money, but actually changing the way you do something and actually changing outcomes. And I don't think we're that far away from that.
所以我认为,作为企业和政府,准备好迈出下一步,将通用模型变得非常相关、非常具体于业务流程、临床流程或政府流程,无论它可能是什么,良好的数据治理是一个非常好的第一步。我真的认为得到一个不仅仅用节省时间、节省金钱来衡量的结果,而是实际上改变你做某事的方式并实际上改变结果。我认为我们离那不远。
Humane One: The World's First AI Operating System | Humane One:世界上第一个人工智能操作系统

Moderator: Well, you don't think we're far away at all because you made some announcements here on a number of enterprise AI solutions. Just talk me through a couple which are clearly pertinent to the discussion that we're having.
主持人:嗯,你认为我们根本不远,因为你在这里宣布了一些企业人工智能解决方案。只是向我介绍几个显然与我们正在进行的讨论相关的。
Tareq: When we established Humane in the early days, like any other startup we had an option: go to legacy systems, implement processes around things that we have done for decades, or if you're going to tell the world we really are a capable company, innovative, creates IPs, why don't we do something the world has never done with advanced models that Google has done, OpenAI has done?
Tareq:当我们在早期建立Humane时,像任何其他初创公司一样,我们有一个选择:去遗留系统,围绕我们几十年来所做的事情实施流程,或者如果你要告诉世界我们真的是一家有能力的公司,创新的,创建知识产权的,我们为什么不用谷歌所做的、OpenAI所做的先进模型做一些世界从未做过的事情呢?
Old Interface
旧界面
Icons on desktop
桌面上的图标
HR icon, Finance icon, Legal icon
人力资源图标、财务图标、法律图标
Thousands of applications
成千上万的应用程序
New Interface
新界面
Voice or text only
仅语音或文本
One AI operating system
一个人工智能操作系统
Integrated marketplace
集成市场
I pondered, wondered why the interfaces on mobile compute have never changed. Over the last 30 years, if you go all the way to the invention of Windows in 1985, what we have done is collected icons on desktop. If you want to do something in HR, here's your icon. In finance, here's your icon. Legal, here's an icon. By the way, depending on the size of the enterprise, you'll end up with thousands of applications that organizations are built around.
我思考,想知道为什么移动计算上的界面从未改变。在过去的30年里,如果你一直追溯到1985年Windows的发明,我们所做的就是在桌面上收集图标。如果你想在人力资源部门做某事,这是你的图标。在金融领域,这是你的图标。法律,这是一个图标。顺便说一下,根据企业的规模,你最终会有成千上万个组织围绕其建立的应用程序。
And then we wonder why value realization is not delivered. Today at 1:00 we did something the world has not done. We created the world's first AI operating system. And the interface is just voice or text. But this is not knowledge retrieval. This addresses fundamentally HR, finance, legal, deep research.
然后我们想知道为什么价值实现没有交付。今天下午1点,我们做了世界从未做过的事情。我们创建了世界上第一个人工智能操作系统。界面只是语音或文本。但这不是知识检索。这从根本上解决了人力资源、财务、法律、深度研究的问题。
And so if I want to go on vacation, why do I need to go to this icon to say submit my workflow? So what we did is a fundamental transformation on the architecture of the platform.
所以如果我想去度假,为什么我需要去这个图标说提交我的工作流程?所以我们所做的是对平台架构进行根本性的转型。
And so if I want to go on vacation, why do I need to go to this icon to say submit my workflow? So what we did is a fundamental transformation on the architecture of the platform. Why is that important? We believe that as the world continues to evolve and deliver agents—bespoke agents, ones that do things for legal and finance—what is happening today, they are built as applications that are also built into their silos.
所以如果我想去度假,为什么我需要去这个图标说提交我的工作流程?所以我们所做的是对平台架构进行根本性的转型。为什么这很重要?我们相信,随着世界继续发展并提供代理——定制代理,为法律和金融做事的代理——今天正在发生的事情,它们被构建为也被构建到它们的孤岛中的应用程序。
Humane One integrates a marketplace where my requirement to third parties is one thing: please respect the UX. I just want the prompt to be the same. So I'd encourage everybody to see what really this architecture looks like. I am 100% convinced whether Humane does it, Google does it, Apple does it, this is the future UX. We will get into a world where AI agents are integrated into one platform. They are directly integrated with your enterprise system processes.
Humane One集成了一个市场,我对第三方的要求是一件事:请尊重用户体验。我只是想让提示相同。所以我鼓励每个人看看这个架构真正是什么样子的。我100%确信,无论是Humane做,谷歌做,苹果做,这都是未来的用户体验。我们将进入一个人工智能代理集成到一个平台的世界。它们直接与您的企业系统流程集成。
11→0
Payroll Staff
工资单员工
From 11 consultants to zero
从11个顾问到零
100%
Automation
自动化
"Run my payroll" on the 27th
27日"运行我的工资单"
So today, Humane does not have any third-party products and we build agents now that are running my business. So one specific example of value realization: payroll. Before, 11 people, consultants that were doing the job. Now zero, by just declaring a statement on the 27th of the month: "I want you to run my payroll." Now that's value realization.
所以今天,Humane没有任何第三方产品,我们现在构建运行我的业务的代理。所以一个价值实现的具体例子:工资单。以前,11个人,顾问在做这份工作。现在零,只需在每月27日宣布一个声明:"我想让你运行我的工资单。"现在这就是价值实现。
But I'm not doing it just for employee reduction. I needed to take my own employees and offload them from mundane tasks. If you look at the complexity of HR processes, why does it take so many days to hire an employee, to vet an employee, to do performance management? We have optimized our agents to really carry these mundane tasks and make sure that our employees become product managers.
但我这样做不只是为了减少员工。我需要把我自己的员工从平凡的任务中卸载下来。如果你看人力流程的复杂性,为什么雇用员工、审查员工、进行绩效管理需要这么多天?我们已经优化了我们的代理,真正执行这些平凡的任务,并确保我们的员工成为产品经理。
My HR is no longer HR. They're product managers of HR. Their job is to tell this agent what are the requirements that they need. We call it Humane One because we believe you need only one IT product for your enterprise.
我的人力资源部门不再是人力资源部门。他们是人力资源的产品经理。他们的工作是告诉这个代理他们需要什么要求。我们称之为Humane One,因为我们相信你的企业只需要一个IT产品。
My HR is no longer HR. They're product managers of HR. Their job is to tell this agent what are the requirements that they need. We call it Humane One because we believe you need only one IT product for your enterprise.
我的人力资源部门不再是人力资源部门。他们是人力资源的产品经理。他们的工作是告诉这个代理他们需要什么要求。我们称之为Humane One,因为我们相信你的企业只需要一个IT产品。
Google's Innovation Framework | 谷歌的创新框架

Moderator: Ruth, final comment from you on this.
主持人:Ruth,关于这一点,你的最后评论。
Ruth: If I could add back to your prior question, what is it that's needed to unlock all of this? Is it energy infrastructure? I think we and others have built that infrastructure. And so really reflecting on your question, I think there's one other critical element. As innovative as Google is, we have never taken innovation for granted and we are deliberate about it.
Ruth:如果我可以补充你之前的问题,解锁所有这些需要什么?是能源基础设施吗?我认为我们和其他人已经建立了该基础设施。所以真正反思你的问题,我认为还有一个关键要素。尽管谷歌很创新,但我们从未认为创新是理所当然的,我们对此很慎重。
1
20% Project
20%项目
Early days of Google
谷歌的早期
2
Google X
Google X
Moonshot factory
登月工厂
3
Alphabet
Alphabet
Other bets structure
其他赌注结构
And what I mean by that is in the earliest days of Google, we had something called the 20% project which probably people have read about. If you had a great idea, take 20% of your time, pursue it. That turned into Google X. That turned into Alphabet and the creation of these other bets.
我的意思是,在谷歌的早期,我们有一个叫做20%项目的东西,人们可能已经读过。如果你有一个好主意,花20%的时间去追求它。这变成了Google X。这变成了Alphabet以及这些其他赌注的创建。
And in 2016, Sundar Pichai, our CEO said, "We are moving from a mobile-first company to an an AI-first company. It will be full stack: chips, models, applications, research, the whole thing so that we can better deliver for everyone." He set the north star.
2016年,我们的首席执行官Sundar Pichai说:"我们正在从移动优先公司转向人工智能优先公司。它将是全栈的:芯片、模型、应用程序、研究,整个事情,以便我们可以更好地为每个人提供服务。"他设定了北极星。
And in 2016, Sundar Pichai, our CEO said, "We are moving from a mobile-first company to an AI-first company. It will be full stack: chips, models, applications, research, the whole thing so that we can better deliver for everyone." He set the north star. Point number one: CEO needs to make it really clear we are all moving, everything is moving in this direction.
2016年,我们的首席执行官Sundar Pichai说:"我们正在从移动优先公司转向人工智能优先公司。它将是全栈的:芯片、模型、应用程序、研究,整个事情,以便我们可以更好地为每个人提供服务。"他设定了北极星。第一点:首席执行官需要明确表示我们都在移动,一切都在朝这个方向移动。
And then created something called Labs. And Labs is a wonderful group with a young superstar leader where they have scrappy teams of six to 10 people who come up with ideas that can go horizontally, as Tareq was saying. You have to go horizontally to really scale this. They kill things early and they promote them. And I think an organizational structure is another key part, because change management, measurement is about humans and we need to be able to bring people on the journey so they can see the upside from going on this journey.
然后创建了一个叫做实验室的东西。实验室是一个有年轻超级明星领导者的精彩团队,他们有6到10人的勇敢团队,他们提出了可以横向发展的想法,正如Tareq所说。你必须横向才能真正扩展这一点。他们早期杀死事物并推广它们。我认为组织结构是另一个关键部分,因为变革管理、衡量是关于人类的,我们需要能够带领人们踏上旅程,以便他们可以看到踏上这段旅程的上升空间。
Audience Poll: Who Will Win the Global AI Race? | 观众投票:谁将赢得全球人工智能竞赛?

Moderator: Thank you. Can we bring up the poll that our wonderful audience queued on? Who will win the global AI race? US 43%, China 21%, Europe 0%, MENA interesting 17%. And we can't be won around this floor as we close out. We've got 60 seconds I'm being told. Tareq, what do you make of that? What would your answer be? What would you have pressed?
主持人:谢谢。我们能否调出我们精彩的观众排队的民意调查?谁将赢得全球人工智能竞赛?美国43%,中国21%,欧洲0%,中东北非有趣的17%。当我们结束时,我们无法在这个楼层周围获胜。我被告知我们有60秒。Tareq,你怎么看?你的答案是什么?你会按什么?
Tareq
Well, I would have put Europe a little bit more than 0%. So, all right. But look I think in general I really admire the startup ecosystem in the US. I would say US, China and I would not discount what we're doing in the Middle East.
嗯,我会把欧洲放得比0%多一点。所以,好吧。但看,我认为总的来说,我真的很欣赏美国的创业生态系统。我会说美国、中国,我不会低估我们在中东所做的事情。
Mike
Well I think there'll be lots of winners and I think it'll all change very rapidly. But most importantly, I think the human race is going to win because I think what we're doing here is going to have more profound impact on so many of the things that we depend on as humans. So I'm cheering for everybody in that regard.
嗯,我认为会有很多赢家,我认为这一切都会迅速变化。但最重要的是,我认为人类将会获胜,因为我认为我们在这里所做的将对我们作为人类所依赖的许多事情产生更深远的影响。所以在这方面我为每个人加油。
Vimal
I think adoption rates will define who will be the winner. I think we are concluding too early. Infrastructure, we are in our infrastructure stage. So who drives the adoption and multiplies that will determine the winner. We're kind of more or less at par. The difference is not large outside US. So that's my view that the adoption rates will really determine. And some regions are highly digitized already and I think we have to take cue from that—they're likely to adopt AI faster. If you're natively not digital, you're not going to win the AI race.
我认为采用率将决定谁将成为赢家。我认为我们得出结论太早了。基础设施,我们处于我们的基础设施阶段。所以谁推动采用并倍增将决定赢家。我们或多或少处于同等水平。美国以外的差异并不大。所以这是我的观点,采用率将真正决定。一些地区已经高度数字化,我认为我们必须从中获取线索——他们可能会更快地采用人工智能。如果你本身不是数字化的,你不是赢家。
Ruth
Every head of state I meet with says "I must be a part, my country must be a part of the digital transformation" for all the reasons we talked about. And the US is very focused on engaging with our allies to unlock what that upside is. So I think the US has the opportunity because of engagement with some of these other rows.
我遇到的每一位国家元首都说"我必须成为其中的一部分,我的国家必须成为数字化转型的一部分",原因是我们谈到的所有原因。美国非常专注于与我们的盟友合作,以解锁该上升空间是什么。所以我认为美国有机会,因为与其中一些其他行的合作。
Point number one, and I think I completely agree that the most important thing is the application in a substantive way. Reimagine what your business can be. Think about what our Nobel Prize winner said. Take on those things that seem to be intractable because that's where the upside comes from. And I do think it's the unlock that we're going to be able to deliver.
第一点,我认为我完全同意最重要的是以实质性的方式应用。重新想象你的业务可以是什么。想想我们的诺贝尔奖获得者所说的话。承担那些似乎棘手的事情,因为这就是上升空间的来源。我确实认为这是我们将能够提供的解锁。
Closing Remarks | 结束语
Moderator: I'd love to chair this same panel next year and see where we are at a year from now. I really want to see where you're at because we can see where you've got to in what, five, six months. It's been fascinating. Thank you very much indeed. The very best of luck.
主持人:我很想明年主持同一个小组,看看我们一年后在哪里。我真的想看看你在哪里,因为我们可以看到你在五六个月里取得了什么成就。这很迷人。非常感谢。祝你好运。
Panel: Thank you. Thank you.
小组:谢谢。谢谢。
[Applause]
[掌声]
End of Transcript | 转录结束
Thank you for reading this comprehensive discussion on AI leadership featuring Ruth Porat (Google/Alphabet), Vimal Kapur (Honeywell), Mike Sicilia (Oracle), and Tareq Amin (Humane) at FII9.
感谢您阅读这场关于人工智能领导力的全面讨论,由Ruth Porat(谷歌/Alphabet)、Vimal Kapur(霍尼韦尔)、Mike Sicilia(甲骨文)和Tareq Amin(Humane)在FII9上进行。