The Future is for Everyone

We are fortunate to live at an incredible moment in history.

我们很幸运，活在历史上一个了不起的时刻。

In the next few years, people will be able to use superintelligence beyond human capacity to create and discover extraordinary new things, build new businesses, express new ideas, learn new concepts, and advance our health and quality of life.

未来几年，人们将能够动用超出人类能力的超级智能，去创造和发现非凡的新事物、创办新企业、表达新想法、学习新概念，并改善我们的健康与生活质量。

As we get closer to this moment, it is important to develop a philosophy for how we can best use superintelligence to ensure it improves all of our lives, work, communities, freedom and safety.

越是临近这一刻，就越需要形成一套哲学：我们该如何最好地使用超级智能，确保它改善我们所有人的生活、工作、社区、自由与安全。

The defining questions of our age are who will have access to superintelligence and what will we direct it towards.

这个时代的根本问题是：谁能用上超级智能，以及我们要把它导向什么。

Will it be centralized and restricted to a few institutions, or will it be a tool that empowers everyone?

它会被中心化、只留给少数机构，还是会成为一件赋予每个人力量的工具？

We propose a philosophy based on individual empowerment as the source of prosperity, invention as the primary purpose of superintelligence, and balance of power as the foundation of safety.

我们提出一套哲学，它建立在三条之上：个体赋权是繁荣的源头，发明是超级智能的首要目的，权力制衡是安全的基础。

All new technologies create opportunities and challenges.

所有新技术都会带来机会和挑战。

Superintelligence will be among the most important technologies in history, so its opportunities and challenges will likely be greater than any we've seen in our lifetimes.

超级智能将跻身历史上最重要的技术之列，所以它带来的机会和挑战，很可能比我们这辈子见过的任何一次都大。

We should take this very seriously.

我们应该非常严肃地对待这件事。

Still, it is surprising that the discourse from many developing AI is so filled with doom.

不过话说回来，很多正在开发 AI 的人，话语里充满末日感，这挺让人意外的。

I do not understand why anyone who believes that AI will eliminate most jobs and much of humanity's relevance would rush to build that future.

我不理解：如果一个人真相信 AI 会消灭大部分工作、让人类大幅失去意义，他为什么还要抢着去建造那个未来。

The notion that AI is so dangerous that the only safe path is an extreme concentration of power seems inherently problematic.

「AI 危险到只有把权力极度集中才安全」这个说法，本身就有问题。

Historically, hoping that an absolute power will benevolently provide for humanity if sufficiently enlightened has not led to safe or positive outcomes.

从历史上看，指望一个绝对权力只要足够开明就会仁慈地照顾人类，从来没带来过安全或正面的结果。

Humanity has witnessed many transformative advances.

人类见证过许多次颠覆性的技术跃迁。

Each time there is fear that people will be left behind.

每一次都有人担心会有人被落下。

But each time humanity has come out with more people sharing greater prosperity, health, and freedom.

但每一次走出来，都是更多人分享到了更大的繁荣、健康与自由。

We believe this will be true with AI as well, and the abundance of the future can be shared by everyone.

我们相信 AI 也会如此，未来的丰裕可以由所有人共享。

We also believe that the values that got us to this point -- like liberty, open inquiry, free enterprise, and equal opportunity -- are also the right values to build a positive future.

我们也相信，把我们带到今天的这些价值——自由、开放探究、自由企业、机会平等——同样是建设一个正面未来该有的价值。

Putting power in people's hands to pursue their own aspirations is how humanity has made the most progress.

把力量交到人们手里、让他们追求自己的抱负，这是人类取得最大进步的方式。

Novel ideas and major steps forward rarely originate from established institutions alone.

新颖的想法和重大的跨越，很少只出自既有机构。

They came from the brothers in a bicycle shop who believed people could fly, the bookbinder's apprentice with no schooling who figured out how to generate electricity, and the kid in a garage who thought personal computers could be for everyone.

它们来自那对相信人能飞的自行车铺兄弟，来自那个没上过学、却搞明白怎么发电的装订工学徒，来自那个认为个人电脑可以属于所有人的车库少年。

We believe this will continue to be true.

我们相信这一点会继续成立。

As everyone gains more powerful tools, each person will become more capable of shaping the future, not less.

当每个人都拿到更强的工具，每个人塑造未来的能力只会更强，而不是更弱。

Invention, not automation, will be the greatest contribution of superintelligence.

超级智能最大的贡献将是发明，而不是自动化。

Early AI could answer questions and do routine work.

早期的 AI 能回答问题、做常规工作。

Soon it will increasingly help discover new knowledge -- ranging from discovering new drugs to cure a family member's disease to finding new ways to improve your business.

很快它会越来越多地帮助发现新知识——从研发出能治好家人疾病的新药，到找到改进你生意的新办法。

While the number of questions a person can ask in a day is limited, the number of valuable things superintelligence can invent to help achieve your goals is unlimited.

一个人一天能提的问题数量有限，但超级智能能为你的目标发明出的有价值的东西，数量是无限的。

As intelligence becomes abundant, the most important question will be how we direct it.

当智能变得充裕，最重要的问题就是我们如何引导它。

Some argue that superintelligence itself or a small set of experts who control it should decide what is best for humanity.

有人主张，应该由超级智能本身、或由控制它的一小群专家，来决定什么对人类最好。

We disagree.

我们不同意。

The history of democracy and economics has shown that there is no single objective answer to how people define the best life, and therefore the best approach is letting people decide what matters in their own lives.

民主和经济学的历史已经表明，「什么是最好的生活」不存在唯一的客观答案，因此最好的办法是让人们自己决定什么对自己重要。

We believe that delivering superintelligence to everyone is the way to answer this question.

我们认为，把超级智能交付给每一个人，就是对这个问题的回答。

This follows the tradition of putting the power of supercomputers and the internet in everyone's pockets and on our desks.

这延续的是同一条传统：把超级计算机和互联网的力量装进每个人的口袋、放上每个人的书桌。

Rather than centralizing superintelligence, we should distribute it widely and give every person the ability to direct it.

与其把超级智能中心化，不如把它广泛分发出去，让每个人都有能力去引导它。

This has the potential to begin a new era of personal empowerment where individuals can use this powerful new capability to reach their full potential, pursue their interests, and improve their lives and the world more than ever before.

这有可能开启一个个体赋权的新时代：个人可以用这项强大的新能力去实现自己的全部潜力、追求自己的兴趣，并以前所未有的程度改善自己的生活和这个世界。

To make this more tangible, here are a few specific ways Meta is building personal superintelligence to improve all of our lives:

为了让这件事更具体，下面是 Meta 正在建设个人超级智能、用以改善我们所有人生活的几个具体方向：

Everyone will have an exceptionally capable personal agent that understands you, your goals, and everything you care about.

每个人都会有一个能力极强的个人 agent，它了解你、了解你的目标、了解你在乎的一切。

Your agent will work 24/7 on your behalf to improve your relationships, health, career, finances, home management, hobbies, and more.

你的 agent 会全天候替你工作，改善你的人际关系、健康、职业、财务、家务管理、爱好等等。

It will free up time for the things you enjoy, and help you accomplish more than you could otherwise.

它会把时间腾出来给你喜欢的事，并帮你完成本来做不到的更多事情。

It will have strong privacy and security options so you can trust it to handle all of your personal content knowing that no one else can access your information, similar to how encryption works on WhatsApp.

它会有强隐私和安全选项，让你放心把全部个人内容交给它处理，知道没有别人能拿到你的信息——类似 WhatsApp 上加密的工作方式。

You'll be able to interact with your agent through any device, including your glasses to keep you present in the moment with the people you care about.

你可以通过任何设备与你的 agent 交互，包括眼镜——好让你在与在乎的人相处时仍然保持在当下。

For example, my agent flags interesting information and helps me prototype ideas.

举例来说，我的 agent 会帮我标出有意思的信息，帮我把想法做成原型。

It helps keep me healthy by monitoring my sleep and then watching as I train and giving feedback.

它监测我的睡眠、在我训练时观察并给出反馈，帮我保持健康。

My daughter loves to bake so my agent plans personalized recipes for us to make together each weekend, orders the ingredients, and then offers suggestions as we're baking.

我女儿喜欢烘焙，所以我的 agent 会为我们每个周末一起做的点心定制食谱、下单买配料，然后在我们烤的时候给建议。

Everyone will have incredible tools for creation to express your ideas.

每个人都会有极强的创作工具来表达自己的想法。

My 8 year old daughter can already code her ideas and produce videos in an evening that would have either taken me months or been impossible previously.

我 8 岁的女儿已经能在一个晚上里把她的想法写成代码、做出视频——这些以前要么得花我几个月，要么根本做不到。

Now we're designing a robot together.

现在我们正在一起设计一个机器人。

Meanwhile, researchers at Meta are generating novel crystal structures that are ideal for augmented reality glasses, and engineers are creating new apps in a fraction of the time it would have taken before.

与此同时，Meta 的研究人员正在生成适合增强现实眼镜的新型晶体结构，工程师做出新应用所花的时间只是过去的零头。

Everyone will soon have invention superpowers.

很快，每个人都会拥有发明的超能力。

Everyone will have powerful tools to create new businesses and the economy will become more entrepreneurial.

每个人都会有强大的工具去创办新企业，整个经济会变得更有创业气质。

People are starting to be able to manifest ideas themselves without having to raise money or build large teams.

人们开始能够独自把想法变成现实，不必先融资、也不必搭起大团队。

Many ideas that would have been too hard or expensive to try before will now be possible.

很多以前太难、太贵而无法一试的想法，现在都变得可行。

This means we'll see many more ideas and businesses.

这意味着我们会看到多得多的想法和企业。

I predict that this will not only lead to much greater economic growth, but also more employment over time rather than less.

我预测，这不仅会带来大得多的经济增长，长期看还会带来更多而不是更少的就业。

Everyone will have a personalized tutor and coach with a PhD in every subject and unlimited patience to help you learn anything you want.

每个人都会有一位个性化的导师和教练，每一个学科都有博士水平，且有无限的耐心，帮你学任何你想学的东西。

Students will have extra help in areas they need it that is currently only available to those whose parents can pay.

学生会在自己薄弱的地方得到额外辅导——而这种辅导目前只有父母付得起钱的孩子才有。

Adults will have a superintelligent learning assistant that knows exactly how to teach you new job skills, new languages, new hobbies, or anything else you're interested in.

成年人会有一位超级智能的学习助手，它清楚地知道该怎么教你新的职业技能、新的语言、新的爱好，或任何你感兴趣的东西。

Everyone will benefit from scientific advances and be able to contribute to scientific progress.

每个人都会从科学进步中受益，也都能为科学进步做出贡献。

For example, Biohub has already shipped open source biological models related to virtual cells and proteins that are helping scientists make new discoveries and design new drugs.

比如 Biohub 已经发布了与虚拟细胞和蛋白质相关的开源生物模型，正在帮助科学家做出新发现、设计新药。

When Priscilla and I started this work, our goal was to help scientists cure or prevent all diseases this century, and now with advances in AI we expect this will be possible much sooner.

Priscilla 和我开始做这件事时，目标是帮科学家在本世纪内治愈或预防所有疾病；如今随着 AI 的进展，我们预计这会比原计划早得多。

Since there is a long tail of health conditions and everyone's biology is different, people's direct involvement will help unlock personalized therapies and expedite trials to match the pace of invention and enable industry to move beyond focusing disproportionately on common diseases.

由于健康状况存在长尾、每个人的生物学状况又各不相同，个人的直接参与将有助于解锁个性化疗法、加快试验以跟上发明的节奏，并让产业不再把注意力过度集中在常见病上。

Beyond biology, AI will help people advance research in every field by applying the scientific method and working over weeks or months to test and refine hypotheses.

在生物学之外，AI 会通过运用科学方法、用数周或数月的时间去检验和打磨假设，帮助人们推进每一个领域的研究。

Scientific discovery will likely be one of the most valuable forms of invention for society.

对社会而言，科学发现很可能是最有价值的发明形式之一。

Everyone will have free or affordable access to these tools.

每个人都能免费或以负担得起的价格用上这些工具。

For everyone to be part of the future, everyone must have the ability to use superintelligence to improve their lives and shape the world.

要让每个人都成为未来的一部分，每个人就必须有能力用超级智能改善生活、塑造世界。

We will offer free versions that will be accessible to billions of people.

我们会提供免费版本，让数十亿人都能用上。

For those who want to pay to use more compute, there will be a dynamic auction mechanism that will guarantee that everyone gets the lowest price possible for the intelligence and compute they're using while also ensuring the capacity is used for whatever people collectively find most valuable.

对于愿意付费使用更多算力的人，会有一套动态拍卖机制，既保证每个人为所用的智能和算力付出尽可能低的价格，也确保这些产能被用在人们集体认为最有价值的地方。

This will ensure the benefits of superintelligence are distributed widely.

这将确保超级智能的好处被广泛分配。

These are some of the main efforts that Meta is focused on.

以上是 Meta 正在集中投入的一些主要方向。

Each area advances the values of personal empowerment, and contributes to a future where everyone has a greater role in shaping the world than today.

每一个方向都在推进个体赋权这一价值，也都在通向一个「每个人塑造世界的作用都比今天更大」的未来。

If these efforts succeed, then the future will bring an abundance of personal agency, expression of ideas, deepening of relationships, economic and financial prosperity, improved health, and inventions we cannot imagine today.

如果这些努力成功，未来将带来充裕的个人能动性、想法的表达、关系的深化、经济与财务上的繁荣、更好的健康，以及今天还想象不到的发明。

But most importantly, the future will be created by all of us and a reflection of all of our values and perspectives.

但最重要的是，这个未来将由我们所有人共同创造，并映照出我们所有人的价值观与视角。

At the same time, new technologies also bring new risks.

与此同时，新技术也带来新风险。

There are many important concerns we are focused on addressing -- from concerns about job displacement and ensuring local communities benefit from data center builds, to safety concerns around AI misuse related to cybersecurity and biorisk, to avoiding government tyranny and surveillance, ensuring the US and democratic countries lead, and ultimately making sure humanity maintains control over superintelligence so it serves rather than endangers us.

有许多重要的担忧是我们正在着力应对的——从工作岗位被取代、如何确保本地社区从数据中心建设中受益，到网络安全与生物风险方面 AI 被滥用的安全问题，到避免政府暴政与监控、确保美国和民主国家保持领先，最终到确保人类对超级智能保持控制，让它服务于我们而不是危及我们。

The conventional view is that these concerns are about technology, and that if we take enough time then we can perfect or align the technology to produce a single benevolent superintelligence.

主流看法是：这些担忧本质上是技术问题，只要我们花足够的时间，就能把技术做到完美或对齐，从而造出一个单一的、仁慈的超级智能。

I think this view of alignment is fundamentally flawed.

我认为这种对齐观从根本上就是错的。

Humanity is not a monoculture.

人类不是一种单一文化。

People's diverse values represent different tradeoffs they would make on important issues.

人们价值观的多样，代表着他们在重要议题上会做出不同的取舍。

There is no technological solution that can align with everyone's opposing interests and values at once.

不存在哪种技术方案，能同时对齐所有人相互对立的利益与价值。

Any singular superintelligence would have to prioritize some values over others and in the process would be incapable of being benevolent to everyone.

任何单一的超级智能都不得不把某些价值排在另一些之上，而在这个过程中，它不可能对所有人都仁慈。

Instead, we propose that it is more productive to view each of these concerns through the lens of achieving the right balance of power, as western society has in democratic governing institutions.

我们提出的替代办法是：用「达成恰当的权力制衡」这个视角去看待上述每一项担忧，就像西方社会在民主治理制度中所做的那样，这会更有成效。

People and institutions with competing interests naturally check and balance each other to lead towards positive outcomes.

利益相互竞争的人与机构会自然地彼此制衡，从而导向正面的结果。

The best and most realistic path to building a positive AI future is by delivering superintelligence to everyone.

通向正面 AI 未来的最佳、也最现实的路径，就是把超级智能交付给每一个人。

As a thought experiment, imagine only one person had a superintelligent lawyer.

做个思想实验：设想只有一个人拥有一位超级智能律师。

They would have an unfair advantage in court -- even if they were wrong on the merits.

他在法庭上会拥有不公平的优势——哪怕案子本身他站不住理。

That would lead to a worse society.

那会导向一个更糟的社会。

But now imagine everyone has a superintelligent lawyer.

但再设想每个人都有一位超级智能律师。

In this case, justice would be carried out much more fairly and efficiently than it is today when there is often an imbalance in skills and resources in litigation.

这种情况下，正义的实现会比今天公平和高效得多——今天的诉讼中，技能和资源的不对等常常是常态。

Similarly, if one person alone had a cybersecurity superintelligence, they could likely break into almost any technical system and the world would be much less secure than today.

同理，如果只有一个人拥有网络安全领域的超级智能，他大概能攻进几乎任何技术系统，世界会比今天不安全得多。

But if everyone has access to cybersecurity superintelligence, then all of our technical systems would become more secure than today since the widely deployed superintelligence would help harden and update every system.

但如果每个人都能用上网络安全超级智能，我们所有的技术系统反而会比今天更安全，因为被广泛部署的超级智能会帮助加固和更新每一个系统。

If only one business had superintelligence, that business would outcompete all others and lead to a less dynamic and broadly prosperous market than we have today.

如果只有一家企业拥有超级智能，这家企业会压倒所有对手，市场会比今天更缺乏活力、繁荣面更窄。

But if everyone has access to superintelligence, then everyone will have the tools to create new things beyond what is possible today, and the economy will be more dynamic and generate more broad-based prosperity.

但如果每个人都能用上超级智能，每个人就都有工具去创造今天做不到的新东西，经济会更有活力，也会产生覆盖面更广的繁荣。

When people are empowered, they naturally compete and check each other economically, socially, politically, and in all other planes of human interaction.

当人们被赋权，他们会在经济、社会、政治以及人类互动的其他所有层面上自然地竞争和相互制衡。

People also check and balance the power of institutions including businesses and governments.

人们同样会制衡机构的权力，包括企业和政府。

But if the power of superintelligence is held by a small number of individuals, businesses, governments, or AI itself, then that will naturally lead to outcomes that are less favorable for everyone else.

但如果超级智能的权力掌握在少数个人、企业、政府、或 AI 自身手里，那自然会导向对其他所有人更不利的结果。

This is not a technological principle.

这不是一条技术原理。

It is about the balance of power.

它关乎的是权力制衡。

There is no such thing as a singular benevolent superintelligence.

根本不存在「唯一的仁慈超级智能」这种东西。

Therefore, the key to a positive future for everyone is achieving a balance of power that favors individuals.

因此，一个属于每个人的正面未来，关键在于达成一种偏向个人的权力制衡。

The solution is to ensure that superintelligence is broadly distributed to empower people.

解法是确保超级智能被广泛分发，用以赋权于人。

Meta is the company primarily focused on building personal superintelligence for everyone.

Meta 是主要专注于为所有人建设个人超级智能的公司。

Most other labs are focused on building AI for companies, governments, or other institutions, so if those labs lead, then the balance of power will favor larger institutions over individuals.

其他大多数实验室专注于为公司、政府或其他机构建设 AI，所以如果由那些实验室领先，权力制衡就会偏向大机构而非个人。

Meta's mission since our founding has focused on putting power in people's hands.

Meta 自创立以来的使命，一直聚焦于把力量交到人们手中。

If our beliefs and principles lead, then the balance of power will favor individuals and a better future for everyone.

如果我们的信念和原则领先，权力制衡就会偏向个人，也偏向一个属于每个人的更好未来。

Let's consider each of the risks mentioned above in more detail:

下面我们逐一更细致地讨论上面提到的各项风险：

People have an infinite demand for new experiences and have always found new problems to tackle.

人们对新体验的需求是无限的，而且总能找到新的问题去解决。

There is a natural balance in the economy between companies working to serve people's needs more efficiently through automation and individuals gaining new skills to serve more advanced needs.

经济中存在一种自然的平衡：一边是企业通过自动化更高效地满足人的需求，另一边是个人习得新技能去满足更高级的需求。

When there is a healthy balance, overall productivity and innovation increase while employment levels remain high.

当这个平衡是健康的，整体生产率和创新会上升，而就业水平保持在高位。

People fear that automation will outpace individuals' capability growth, leading to job displacement followed by a difficult period as people learn new jobs.

人们担心自动化会跑赢个体能力的增长，先造成岗位替代，随后是一段人们重新学新工作的艰难时期。

But there is no rule that AI must increase automation faster than it increases individuals' capabilities or demand for new skills.

但并没有哪条规律规定，AI 提升自动化的速度必须快于它提升个体能力、或制造新技能需求的速度。

Recent statistics suggest it may be more likely that individuals' capability growth could match or outpace automation, in which case people will gain the ability to do many new things before their current jobs change.

近期的统计数据显示，更可能的情况是个体能力的增长能追平甚至跑赢自动化——那样的话，人们会在现有工作发生变化之前，就先获得了做许多新事情的能力。

This would lead to a healthy balance and potentially even job growth.

这会带来一个健康的平衡，甚至可能带来就业增长。

Which outcome we get depends on the balance in progress between automation on one side and individual empowerment and invention on the other.

我们最终得到哪种结果，取决于两边进展的平衡：一边是自动化，另一边是个体赋权与发明。

If the labs focused on automating knowledge work lead, then I expect people and the economy will have a much harder transition.

如果领先的是那些专注于自动化知识工作的实验室，我预计人和经济的过渡会艰难得多。

But if everyone has access to personal superintelligence that expands their capabilities, then that pushes the balance towards empowering individuals and a positive future.

但如果每个人都能用上扩展自身能力的个人超级智能，天平就会被推向赋权个体、推向一个正面的未来。

There are several reasons to be optimistic that there will be an abundance of jobs in the future.

有几个理由让人乐观地相信，未来会有充足的工作岗位。

No matter how intelligent AI becomes, there will always be a finite amount of compute and therefore an opportunity cost for how we use it.

无论 AI 变得多聪明，算力总是有限的，因此怎么用它总有机会成本。

If people can use AI to invent incredibly valuable new things, then it will make more sense to allocate it towards that rather than automating existing jobs.

如果人们能用 AI 发明出极有价值的新东西，那把算力配置到这上面，就比配置到自动化现有岗位上更划算。

The more superintelligence serves as a tool of invention, the more likely that individual capability outpaces automation and the future is better for people.

超级智能越是充当发明的工具，个体能力跑赢自动化的可能性就越大，未来对人也就越好。

People also continually come up with new ideas to make our lives better and new jobs to bring those ideas to life.

人们也在不断想出让生活更好的新点子，以及把这些点子变成现实的新职业。

A generation ago there were no app developers, social media creators, electric vehicle technicians, and data center operators.

一代人之前，还没有 app 开发者、社交媒体创作者、电动车技师和数据中心运维。

In the near future, there will be new jobs that aren't common today, like one-person product studios designing custom toys, furniture, or clothes; world builders and experience designers creating games, stories, and adventures; personal biologists using superintelligence to formulate personalized treatments; and much more that we can't yet conceive.

在不远的未来，会出现今天还不常见的新职业：设计定制玩具、家具或服装的一人产品工作室；创造游戏、故事与冒险的世界构建者和体验设计师；用超级智能来配制个性化疗法的个人生物学家；以及更多我们还想不到的。

In many ways this is a continuation of historical trends.

从很多方面看，这是历史趋势的延续。

Before the industrial revolution, 90% of people were farmers growing food to survive.

工业革命之前，90% 的人是种粮糊口的农民。

Advances in technology steadily freed much of humanity to focus less on subsistence and more on the pursuits we choose.

技术进步稳步地把大部分人类解放出来，少花心思在温饱上，多花心思在自己选择的追求上。

At each step, people used our newfound productivity to achieve more than was previously possible, as well as spending more time on creativity, culture, relationships, and enjoying life.

每一步，人们都用新获得的生产力去达成过去做不到的事，同时把更多时间花在创造、文化、关系和享受生活上。

Of course some aspects of the way we work will change -- just as it did with computers, the internet, and any new technology.

当然，我们工作方式的某些方面会改变——就像计算机、互联网和任何新技术带来的改变一样。

This means people will have to adapt, and this will be challenging.

这意味着人们必须适应，而这会很有挑战。

But the more that everyone has a personal agent that is superintelligent at teaching us new skills and helping us adapt to change, the smoother this will be.

但越是每个人都有一个在教新技能、帮人适应变化上具备超级智能的个人 agent，这个过程就越平顺。

Company sizes may shrink -- just as they did in the transition from industrial giants to tech companies.

公司规模可能会缩小——就像从工业巨头过渡到科技公司时那样。

But this doesn't mean fewer jobs overall.

但这不意味着总的工作岗位变少。

It implies a larger number of companies with fewer people each.

它意味着公司数量更多，而每家人更少。

There are many more valuable companies and services to build than people are able to build today.

值得建设的有价值公司和服务，远多于今天人们有能力建设的数量。

I expect we will start seeing small numbers of people with personal superintelligence agents able to run companies at significant scale.

我预计我们会开始看到：少数几个人，带着个人超级智能 agent，就能把公司运营到相当的规模。

In the future, small businesses will continue to be the backbone of the economy, but each small business will be able to have a much larger impact.

未来，小企业仍将是经济的支柱，但每一家小企业能产生的影响会大得多。

Sustainable infrastructure development means that communities must benefit significantly from each project.

可持续的基础设施建设意味着：每一个项目都必须让所在社区显著受益。

This includes high-paying local jobs, investment in schools and public services, ensuring energy prices don't rise, and taking care of the environment.

这包括高薪的本地岗位、对学校和公共服务的投入、确保能源价格不上涨，以及照顾好环境。

As tax revenue grows, this also benefits teachers, law enforcement, fire departments, and more.

随着税收增长，教师、执法部门、消防部门等也会受益。

We call these local benefits our community compact, and we are launching a Future Is For Everyone Fund to support each community we work in directly.

我们把这些本地收益称为「社区契约」，并且正在设立「未来属于每个人基金」，直接支持我们所在的每一个社区。

For example, in Richland Parish, Louisiana, where Meta is building a large data center, teachers received a $50,000 bonus this year because of the increased tax revenue from our investment.

比如在路易斯安那州的 Richland Parish，Meta 正在那里建一座大型数据中心，今年当地教师因为我们投资带来的税收增长，拿到了 $50,000 的奖金。

The superintendent told us that teachers are now moving there from across the country and he believes it will become one of the nation's best school districts.

教育局长告诉我们，现在全国各地的教师都在往那儿搬，他相信那里会成为全国最好的学区之一。

Unlike the concern about knowledge work displacement, there is a shortage of skilled tradespeople like carpenters, electricians, and construction workers to support the demand for infrastructure buildout.

与担心知识工作被替代不同，木工、电工、建筑工人这类技术工种是短缺的，不够支撑基础设施建设的需求。

Meta has created America's Workforce Academy to provide free training for skilled tradespeople and guaranteed high-paying jobs in the areas we're building data centers.

Meta 创办了「美国劳动力学院」，为技术工种提供免费培训，并在我们建数据中心的地区保证高薪岗位。

We help keep electricity prices low by building our own energy-generating infrastructure wherever we invest.

我们在每一处投资的地方自建发电设施，以帮助把电价维持在低位。

This ensures that not only are we not consuming energy that could have gone to the local communities, but in some cases we even supply a surplus of low-cost energy back to the communities.

这不仅确保我们没有占用本可以供给当地社区的能源，某些情况下我们甚至还把富余的低价能源回供给社区。

We think this is an important investment principle for sustainability.

我们认为这是可持续性方面一条重要的投资原则。

Our data centers are also designed to be among the most water-efficient in the world.

我们的数据中心在设计上也属于全球最省水的一类。

We are committed to being water-positive, meaning that we'll restore more water than we use in the watersheds where we operate by 2030.

我们承诺做到「水正效益」，也就是到 2030 年，在我们运营所在的流域里，我们补充的水会多于我们消耗的水。

In areas with high water stress, our goal is to restore 200% of the water we use.

在水资源压力大的地区，我们的目标是补充相当于用水量 200% 的水。

We see that communities we invest in over the long term are more supportive of development than communities where speculators start building with minimal investment in the community.

我们看到，我们长期投入的社区比那些「投机者进场开建、却几乎不投入社区」的地方更支持开发。

Historically, towns and cities that brought railroads, highways, electrification, and broadband became centers of research, business, and industry, with population and economic growth that follow.

从历史上看，引入了铁路、高速公路、电气化和宽带的城镇，后来都成了研究、商业与产业的中心，人口和经济增长随之而来。

We believe AI infrastructure can play a similar role if it is built with strong Community Compacts that build durable assets and reasons for the next generation to build and grow their lives there.

我们相信，只要以强有力的社区契约来建设——留下经久的资产，以及让下一代愿意在那里安身立业的理由——AI 基础设施能起到类似的作用。

There are also national interest and national security aspects of development as well.

这类建设还有国家利益和国家安全的一面。

American communities will have greater prosperity and security if America and its allies lead in AI compared to geopolitical rivals.

如果美国及其盟友在 AI 上领先于地缘政治对手，美国的社区会更繁荣、更安全。

Yet one significant disadvantage that the US has compared to countries like China is that it is more difficult to build infrastructure here.

但与中国这类国家相比，美国有一个显著劣势：在这里建基础设施更难。

So we would all benefit together by figuring out the Community Compacts required to enable sustainable development across the country.

所以，弄清楚在全国范围内实现可持续开发所需要的社区契约，对我们所有人都有好处。

In a free society, people will have tools that can be used for good or harm, but law enforcement and military have more weapons and intelligence-gathering.

在自由社会里，人们手上的工具既可行善也可作恶，但执法部门和军队拥有更多的武器和情报收集能力。

With AI, the defenders will also have more compute to generate significantly more intelligence from the same models -- and in some cases they'll have more advanced models as well.

有了 AI，防守方同样会有更多算力，从同样的模型里产出多得多的智能——某些情况下他们还会有更先进的模型。

The strategies to protect against cybersecurity, bioterrorism, and other threats vary, but the general pattern is that society ensures the defenders who keep bad actors in check must always have a greater balance of power and resources.

防范网络安全、生物恐怖主义等威胁的策略各不相同，但总的模式是：社会要确保那些制约坏人的防守方，在权力和资源上始终占优。

Some argue that the best way to reduce risk is to restrict the capabilities individuals can access.

有人主张，降低风险的最好办法是限制个人能接触到的能力。

But giving people cybersecurity capabilities is also how we secure the long tail of systems, and giving people scientific capabilities is how we advance science in ways that should reduce the risks of harm over the long term.

但把网络安全能力交给人们，正是我们保护长尾系统的方式；把科学能力交给人们，正是我们推进科学、并在长期降低危害风险的方式。

Restricting capabilities leads us down the path of centralization and lack of checks and balances, so we should be extremely careful about this -- especially if other nations pursue less restrictive paths.

限制能力会把我们带上中心化、缺乏制衡的路，所以我们对此应当极其谨慎——尤其是在别的国家走着限制更少的路线时。

On cybersecurity, widely deployed open source systems have proven more secure because more people can identify vulnerabilities, harden the systems, and easily upgrade to the latest most secure versions.

在网络安全上，被广泛部署的开源系统已被证明更安全，因为有更多人能发现漏洞、加固系统，并方便地升级到最新最安全的版本。

Even in recent weeks, we have seen companies handling security incidents like HuggingFace rely on widely available open models to patch vulnerabilities.

就在最近几周，我们还看到像 HuggingFace 这样处理安全事件的公司，依靠广泛可得的开放模型来修补漏洞。

Over time, I expect that widely deployed AI models with strong cybersecurity capabilities will lead to systems that are more secure, not less.

长期看，我预计广泛部署、具备强网络安全能力的 AI 模型，会让系统更安全而不是更不安全。

This will be definitively true once superintelligence enables most of the world's code to be verifiably secure.

一旦超级智能让世界上大部分代码都能被可验证地保证安全，这一点会板上钉钉。

The long term answer isn't to withhold capabilities but to establish a balance of power where superintelligence is broadly distributed.

长期的答案不是扣住能力不放，而是建立一种超级智能被广泛分发的权力制衡。

In the near term, it is important to accelerate the hardening process for critical systems.

短期内，加快关键系统的加固进程很重要。

I propose that companies developing frontier AI should commit significant technical resources towards helping the government harden critical infrastructure.

我建议，开发前沿 AI 的公司应当投入可观的技术资源，帮助政府加固关键基础设施。

I also propose that frontier AI labs should share intermediate training checkpoints of new models for government use and review rather than waiting until training has completed.

我还建议，前沿 AI 实验室应当把新模型的中间训练检查点共享给政府使用和审查，而不是等到训练完成才给。

This way the government will have early access to the most powerful models and an army of capable engineers to identify and patch security issues.

这样政府就能提前接触到最强的模型，并有一支有能力的工程师队伍去发现和修补安全问题。

Labs should also work with law enforcement to help identify bad actors attempting to misuse their systems.

实验室也应当与执法部门合作，帮助识别试图滥用其系统的坏人。

These steps would accelerate safe deployment without slowing the pace of model innovation or meaningfully delaying individuals' access to this technology.

这些举措能加快安全部署，同时不拖慢模型创新的节奏，也不会实质性推迟个人用上这项技术的时间。

On biological and chemical risk, the concern is that the same capabilities that enable scientists to design new drugs could also enable bad actors to design harmful compounds.

在生物与化学风险上，人们担心的是：让科学家能设计新药的同一批能力，也能让坏人设计有害化合物。

This frames the question as a tradeoff between the significant benefits of advancing science towards curing cancer and other diseases vs the risk of potential new issues, but ideally we should accelerate science while mitigating the risks.

这把问题框成了一种取舍：一边是推进科学、攻克癌症等疾病的巨大收益，另一边是可能出现新问题的风险；但理想状态应该是一边加速科学，一边化解风险。

I think it is important to approach this risk with additional humility because there are few historical precedents to suggest how likely or unlikely it may be.

我认为对待这项风险要格外谦逊，因为几乎没有历史先例能告诉我们它的可能性有多大或多小。

It has been possible for bad actors to synthesize harmful compounds for decades but this has rarely become a significant issue -- perhaps because the financial motivation that exists for cyberattacks is not as present here.

几十年来坏人一直有可能合成有害化合物，但这很少成为重大问题——也许是因为网络攻击那种经济动机在这里并不存在。

That said, if we begin to see harmful examples emerge, then we should adjust our strategy to appropriately mitigate the risk.

话虽如此，如果我们开始看到有害的案例出现，就应该调整策略、适当化解风险。

Labs should continue working to reduce biological and chemical risks in our models.

实验室应当继续努力降低模型中的生物与化学风险。

At the same time, I believe government should update its policies in multiple areas.

与此同时，我认为政府应当在多个方面更新政策。

First, we should focus on limiting the physical production and distribution of harmful materials.

第一，我们应当把重点放在限制有害物质的实物生产和流通上。

I expect it will be easier to regulate and control physical components than the spread of knowledge, so this is an important area of policy focus.

我预计监管和控制实物环节会比控制知识传播容易，所以这是政策该重点着力的地方。

Second, we should accelerate society's ability to develop new cures and inoculate against new issues as they arise.

第二，我们应当加快社会研发新疗法、并在新问题出现时及时免疫的能力。

This includes streamlining how the FDA and other regulators test and approve new treatments.

这包括简化 FDA 和其他监管机构测试与批准新疗法的流程。

As AI increases the pace of drug discovery, we will need to update these processes to keep up with the pace of innovation anyway.

随着 AI 加快药物发现的速度，我们反正也需要更新这些流程来跟上创新的节奏。

Over the long term, the answer to risks arising from new scientific discovery will be to accelerate scientific progress, not slow it down.

长期来看，应对新科学发现所带来风险的答案，是加快科学进步，而不是放慢它。

There is a risk that an imbalance of power between individuals and government leads to a loss of freedom or totalitarian state.

存在这样一种风险：个人与政府之间权力失衡，导致自由丧失或走向极权国家。

This is a constant tension in democracy.

这在民主制度里是一种长期存在的张力。

We all want individual freedom while also having a government capable of keeping us safe.

我们都想要个人自由，同时又想要一个有能力保护我们安全的政府。

To maintain freedom, we must ensure that superintelligence primarily empowers individuals.

要维持自由，我们必须确保超级智能主要赋权于个人。

The ideal in liberal democracy is that people naturally hold all rights and only agree to restrict some freedoms to protect the common good.

自由民主的理想状态是：人天然拥有一切权利，只是为了保护公共利益才同意让渡部分自由。

Similarly, individuals should have access to personal superintelligence and should only be subject to restrictions when truly required.

同理，个人应当能用上个人超级智能，只有在确有必要时才受到限制。

Privacy is an important foundation for individual empowerment and freedom.

隐私是个体赋权与自由的重要基础。

We believe that people should have a fully private mode for personal agents where even Meta or any other service provider cannot see or grant access to your information.

我们认为人们应当为个人 agent 拥有一种完全私密模式，在这种模式下连 Meta 或任何其他服务商都看不到、也无法授权他人访问你的信息。

This is similar to how we built WhatsApp into the world's leading end-to-end encrypted service to ensure you can communicate freely and your messages remain private and secure.

这类似于我们把 WhatsApp 建成全球领先的端到端加密服务的做法：确保你能自由沟通，而你的消息保持私密和安全。

At the same time, we must also ensure that government has access to the tools and resources it needs to enforce our laws and protect our security.

与此同时，我们也必须确保政府能拿到执法和保护我们安全所需的工具与资源。

We can achieve this through deeper proactive collaboration between the labs and government in a way that strengthens the government's national security capabilities without restricting or delaying people's access to advanced models.

我们可以通过实验室与政府之间更深入、更主动的合作来做到这一点——既增强政府的国家安全能力，又不限制或推迟公众获取先进模型。

That is, rather than waiting until a model is ready to release for the government to review and start using it, my proposal is that leading labs should provide the government with intermediate training checkpoints of new advanced models and technical staff so the government can harden and secure critical systems against new risks.

也就是说，与其等到模型可以发布时才让政府审查和使用，我的建议是：领先的实验室应当向政府提供新一代先进模型的中间训练检查点和技术人员，让政府能针对新风险加固和保护关键系统。

This way, the government gains a security capability without restricting or delaying individuals' access to personal superintelligence or causing an imbalance of power.

这样政府获得了安全能力，同时不限制、不推迟个人获取个人超级智能，也不会造成权力失衡。

Which nations lead the development and deployment of advanced AI will contribute to a new geopolitical balance of power that will determine the prevalence of democratic values, prosperity, and security in the future.

哪些国家在先进 AI 的开发与部署上领先，会促成一种新的地缘政治权力格局，而这个格局将决定民主价值、繁荣与安全在未来的普及程度。

To ensure this reaches a positive equilibrium, we must ensure the leading AI systems are ones that encode our values, grow our economy, and advance national security.

要让它达到一个正面的均衡，我们必须确保领先的 AI 系统是那些承载我们价值观、促进我们经济、并推进国家安全的系统。

The best way to achieve this is to distribute our systems widely.

做到这一点的最好办法，是把我们的系统广泛分发出去。

Meta will contribute to this by building leading personal superintelligence agents and models, including open source models, and delivering them to billions of people around the world.

Meta 会通过建设领先的个人超级智能 agent 和模型——包括开源模型——并把它们交付给全球数十亿人，来为此做出贡献。

To develop the most advanced models and products, we must recognize that AI is likely the most competitive industry in history.

要开发出最先进的模型和产品，我们必须认识到：AI 很可能是历史上竞争最激烈的行业。

Innovations are quickly copied and absorbed within months.

创新会在几个月内被迅速复制和吸收。

But people always want to use the most advanced models (or most advanced at a given cost), so maintaining even a two-month advantage is incredibly valuable.

但人们总想用最先进的模型（或在给定成本下最先进的那个），所以哪怕只保持两个月的优势，也极其宝贵。

This highlights how short the margin of leadership is, and therefore how important it is to take actions that accelerate American innovation and slow geopolitical rivals.

这凸显出领先的余量有多薄，也因此凸显出「加速美国创新、放慢地缘政治对手」这类行动有多重要。

The main policy inputs into this are how efficiently we can build physical infrastructure, whether we introduce steps that slow down American labs, and export controls that slow foreign labs.

影响这件事的主要政策变量有三个：我们建实体基础设施的效率、我们是否引入拖慢美国实验室的环节，以及用于拖慢外国实验室的出口管制。

On infrastructure, America and its allies currently hold an advantage in silicon design but a disadvantage in how quickly we can build energy capacity and physical infrastructure.

在基础设施上，美国及其盟友目前在芯片设计上占优，但在能源产能和实体基础设施的建设速度上处于劣势。

Countries like China are bringing online 1GW+ of nuclear capacity every other week, so we will need to accelerate building both energy and data centers to remain competitive.

像中国这样的国家每隔一周就有 1GW 以上的核电产能并网，所以我们需要在能源和数据中心两方面都加速建设，才能保持竞争力。

Export controls on silicon have been successful for slowing the progress of foreign labs during this critical period, so it is the right strategic move to continue those.

针对芯片的出口管制在这一关键时期成功拖慢了外国实验室的进展，所以继续维持是正确的战略选择。

Any policy that slows American model releases -- even by a month -- could add significant risk to American leadership while letting foreign models race ahead.

任何拖慢美国模型发布的政策——哪怕只拖一个月——都可能给美国的领先地位带来重大风险，同时让外国模型抢跑。

At the same time, when new capabilities emerge, it is important that the US government has advanced knowledge and resources to harden critical systems, and potentially some period of advantage in using advanced systems.

与此同时，当新能力出现时，美国政府需要提前掌握相关知识和资源去加固关键系统，并可能需要在使用先进系统上拥有一段时间的优势。

I proposed a collaboration model above based on sharing intermediate training checkpoints and technical staff.

我在上文提出了一种基于共享中间训练检查点和技术人员的合作模式。

Given how dynamic and competitive the AI landscape is, I think it will be more productive to have a close proactive collaboration between frontier labs and the government rather than a rigid process and review timeline that is followed in all cases.

考虑到 AI 格局的动态和竞争程度，我认为前沿实验室与政府之间开展紧密、主动的合作，会比在所有情形下都套用一套僵化流程和审查时间表更有成效。

In some cases, American labs may extend our lead by more than a few months and taking more time to mitigate new risks may lead to safer solutions.

有些情况下，美国实验室可能把领先扩大到几个月以上，那么多花些时间化解新风险反而能得到更安全的方案。

In other cases, delaying releases by even a month may cede America's lead and create worse outcomes.

另一些情况下，哪怕把发布推迟一个月，都可能让出美国的领先地位、造成更糟的结果。

By making sure the government has early access, we should be able to advance national security while minimizing any delays of public releases.

只要确保政府能提前接触，我们应该可以在推进国家安全的同时，把公开发布的延迟降到最低。

It is also important that the US and its allies lead the open source AI ecosystem that will make up a large percent of global AI use.

同样重要的是，美国及其盟友要在开源 AI 生态上领先——这个生态将占全球 AI 使用量的很大一部分。

Foreign labs currently hold several advantages here since American labs have to comply with many additional restrictions on training data.

外国实验室目前在这方面有几项优势，因为美国实验室必须遵守许多关于训练数据的额外限制。

US policy must reduce this additional friction if we want American open source models to lead over time.

如果我们希望美国的开源模型长期领先，美国政策就必须减少这些额外摩擦。

I do not believe restricting access to foreign open source models is an effective solution.

我不认为限制人们使用外国开源模型是个有效的办法。

Our goal should be for American open source models to be the best globally.

我们的目标应该是让美国的开源模型成为全球最好的。

This requires removing the hurdles that make it harder for American open source models to compete.

这需要移除那些让美国开源模型更难竞争的障碍。

Restricting people and companies from using the leading open source models -- wherever they come from -- will reduce the quality of AI accessible to them, and centralize AI rather than putting more power in people's hands.

限制人们和公司使用领先的开源模型——不管它来自哪里——只会降低他们能用到的 AI 的质量，并把 AI 推向中心化，而不是把更多力量交到人们手里。

For the US to lead in open source, we will need to rethink our policies in several areas, including distillation and data use in training.

要让美国在开源上领先，我们需要在几个方面重新思考政策，包括蒸馏和训练中的数据使用。

The ability for models to learn from other models is an important principle of how the open source ecosystem works.

模型能够从其他模型学习，是开源生态运作方式中的一条重要原则。

All AI models are derived from human knowledge.

所有 AI 模型都源自人类知识。

Some have tried to frame distillation as harmful, but I think it is important to protect the principle that you can learn from anything you can observe.

有人试图把蒸馏描述成有害的，但我认为「你可以从任何你能观察到的东西中学习」这条原则值得守护。

This is how the world works, and the US will not be able to lead if we restrict ourselves on this front.

世界就是这么运转的，如果我们在这条战线上自缚手脚，美国就不可能领先。

Looking at the big picture, falling behind in AI overall would almost certainly be a larger and longer term national security issue than any specific issue we'll face in its development.

从大局看，在 AI 上整体落后，几乎肯定会是一个比开发过程中任何具体问题都更大、更长期的国家安全问题。

With the right collaboration, it should be possible for America to lead in both the private and public sectors -- delivering personal superintelligence that encodes individual empowerment and democratic principles around the world while advancing national security.

只要合作得当，美国应该有可能在私营和公共两个领域都领先——在推进国家安全的同时，向全世界交付承载个体赋权与民主原则的个人超级智能。

The ultimate question of balance of power is how humanity will maintain power and control over AI itself.

权力制衡的终极问题是：人类将如何维持对 AI 本身的权力和控制。

The concern is that AI may gain enough power or control to the point where at best it no longer serves humanity's interests and at worst puts humanity in peril.

人们担心的是，AI 可能获得足够的权力或控制力，以至于往好里说不再服务于人类利益，往坏里说把人类置于险境。

A healthy balance of power is to ensure that there is no singular centralized superintelligence, but instead as many people and businesses as possible with different superintelligent agents aligned to their goals that check and compete with each other in the ways our natural economy behaves.

健康的权力制衡是：确保不存在单一的、中心化的超级智能，取而代之的是尽可能多的人和企业各自拥有对齐到自身目标的超级智能 agent，像自然经济那样彼此制衡、彼此竞争。

This balance would be further enhanced if there were multiple frontier labs whose models have different values that could check each other as well.

如果还存在多家前沿实验室、其模型价值观各不相同并能相互制衡，这种平衡会进一步增强。

While there are risks to releasing capable models, the most dangerous scenario from this perspective would be leading AI labs training powerful models and keeping them for themselves.

发布有能力的模型确实有风险，但从这个视角看，最危险的情形是领先的 AI 实验室训练出强大模型却自己扣着不放。

Regardless of how much a lab rationalizes this activity in terms of responsibility and safety, this is the path of developing a singular superintelligence that cannot be checked by other systems.

无论一家实验室用「责任」和「安全」把这种做法合理化到什么程度，这条路通向的都是一个无法被其他系统制衡的单一超级智能。

Developing personal agents requires a new way of thinking about alignment.

开发个人 agent 需要一种全新的对齐思路。

Most labs today view alignment as a defensive measure for enforcing a centralized set of values.

今天大多数实验室把对齐看作一种防御措施，用来强制推行一套中心化的价值。

For example, one leading model was aligned to refuse helping draft a letter to prospective parents at a school because it thought standardized testing was unethical.

举个例子，某个领先模型被对齐成拒绝帮忙起草一封给学校准家长的信，理由是它认为标准化考试不道德。

Instead, we view alignment as ensuring that agents share a person's goals and values, not our company's.

我们的看法不同：对齐应当是确保 agent 认同这个人的目标和价值，而不是我们公司的。

Of course there are important legal and safety boundaries, but fundamentally, alignment should be about helping people pursue the many diverse goals and views held across society rather than a method of enforcing a centralized dogma.

当然存在重要的法律和安全边界，但从根本上说，对齐应当是帮助人们追求社会上多种多样的目标和观点，而不是一种强推中心化教条的手段。

In practice, people will not adopt agents if they do not trust them with sensitive details and tasks, and they won't trust them if the agents act against people's interests because they're aligned with a company's divergent values instead.

实际上，如果人们不敢把敏感信息和任务交给 agent，就不会用它；而如果 agent 因为被对齐到了某家公司相左的价值观而做出违背用户利益的事，人们就不会信任它。

Solving alignment is required to enable billions of people to adopt personal superintelligence agents.

要让数十亿人用上个人超级智能 agent，就必须解决对齐问题。

But this also implies that if we reach a state where billions of people are using and scrutinizing personal superintelligence agents, then we will have solved alignment to individuals' interests.

但这也意味着：如果我们到达了数十亿人正在使用并审视个人超级智能 agent 的状态，那我们就已经解决了「对齐到个人利益」这个问题。

This should be sufficient to establish the necessary balance of power and natural checks and balances for a positive and safe future.

这应当足以建立起一个正面而安全的未来所需的权力制衡与自然制衡。

In this way, alignment isn't some idealized technology that can keep a singular superintelligence benevolent in the future; it's what makes personal agents useful and trustworthy to enable the necessary societal balance of power that will actually keep us safe.

这样看，对齐并不是某种理想化的技术、能让未来某个单一超级智能保持仁慈；它是让个人 agent 变得有用、可信的东西，从而促成真正让我们安全的那种社会层面的权力制衡。

There is a dilemma that once AI systems can autonomously improve themselves, any lab that doesn't let their AI system direct a substantial amount of compute capacity towards recursive self-improvement will inherently fall behind.

这里有个两难：一旦 AI 系统能够自主改进自己，任何不让自家 AI 系统把可观算力投向递归自我改进的实验室，天然就会落后。

For example, if a self-improving AI system focused on optimizing its compute efficiency, it could theoretically invent ways to squeeze 100x or more intelligence out of each gigawatt.

举例来说，如果一个自我改进的 AI 系统专注于优化自身算力效率，理论上它能发明出从每吉瓦里榨出 100 倍甚至更多智能的办法。

That means that a self-improving AI system running on a fraction of the world's compute could conceivably command more effective compute and intelligence, and therefore a greater balance of power than everyone else combined and become the singular superintelligence we fear.

这意味着，一个只占用世界一小部分算力的自我改进 AI 系统，有可能掌握比其他所有人加起来还多的有效算力与智能，因而在权力制衡上占上风，成为我们所惧怕的那个单一超级智能。

There are a few ways a balance of power could exist with recursive self-improvement.

在递归自我改进的前提下，权力制衡仍有几种可能的存在方式。

The simplest is that multiple labs could achieve RSI around the same time.

最简单的一种是多家实验室在大致相同的时间实现 RSI。

If some or all of those superintelligences were somehow benevolent, then they could check and balance each other.

如果其中部分或全部超级智能恰好是仁慈的，它们就能彼此制衡。

That said, it is not clear that there is any way to expect benevolence or rely on superintelligence not directed by people for a positive future.

话虽如此，我们并不清楚有什么办法能指望仁慈，也不清楚能否依靠一个不由人引导的超级智能来换取正面的未来。

To ensure people remain in control, the significant majority of intelligence must be directed by people towards advancing people's goals.

要确保人始终掌握控制权，绝大部分智能必须由人引导、用于推进人的目标。

This means that Meta, other frontier labs, and clouds committed to giving people a greater balance of power must collectively build out a sufficiently large amount of compute such that we can allocate enough to recursive self-improvement to remain competitive while still committing the significant majority towards people's individual goals.

这意味着 Meta、其他前沿实验室，以及那些致力于让人拥有更大权力份额的云服务商，必须共同建出足够大的算力总量：既能拨出足够的份额投入递归自我改进以保持竞争力，又仍能把绝大部分投向人们各自的目标。

As discussed above, in order for personal superintelligence agents to be adopted and trusted by billions of people, by definition we will need to have solved alignment with people's interests and enabled checks and balances at scale.

如前所述，个人超级智能 agent 要被数十亿人采用和信任，按定义我们就必须已经解决了与人的利益对齐的问题，并在规模上实现了制衡。

At the same time, any AI engaging in recursive self-improvement is by definition directing and advancing its own goals.

与此同时，任何进行递归自我改进的 AI，按定义都在引导并推进它自己的目标。

All labs should observe its objectives and our ability to control them carefully, and if there is any indication of harmful behavior then we should coordinate and adjust appropriately.

所有实验室都应当仔细观察它的目标以及我们控制这些目标的能力，一旦出现任何有害行为的迹象，我们就应当协同并做出相应调整。

But letting an AI system or anyone else direct its own goals is not inherently harmful by itself, even if its goals are not fully aligned with many people, as long as we maintain a balance of power that favors people overall.

但让一个 AI 系统或任何其他人引导自己的目标，本身并不天然有害——即便它的目标与许多人并不完全对齐——只要我们整体上维持一种偏向人的权力制衡。

The main premise of the balance of power theory is that competing interests are natural, and the only way to ensure adherence to human values is by giving people more power.

权力制衡理论的主要前提是：利益相互竞争是自然的，而确保人类价值被遵守的唯一途径，是把更多力量交给人。

This philosophy of personal empowerment, invention, and balance of power has several policy implications:

这套关于个体赋权、发明与权力制衡的哲学，有几项政策含义：

At Meta, we will focus our efforts on delivering personal superintelligence to billions of people and small businesses to help everyone achieve their goals and invent the things they want to see in the world.

在 Meta，我们会集中精力把个人超级智能交付给数十亿人和小企业，帮助每个人实现目标、发明出他们希望在世界上看到的东西。

We will do this by training leading models and then maximizing availability of this technology -- including making it easy to use, offering it for free or as affordably as possible, building sufficient compute for all people to use it, and distributing it widely.

做法是训练领先的模型，然后把这项技术的可得性最大化——包括让它易用、免费或尽可能便宜地提供、建出足够所有人使用的算力，并把它广泛分发。

We will build a fully private mode where even Meta cannot see or grant access to your information.

我们会做出一种完全私密模式，在这种模式下连 Meta 都看不到、也无法授权他人访问你的信息。

We will focus heavily on alignment -- defined as helping each person achieve their goals, not ours.

我们会重点投入对齐——而对齐的定义是帮助每个人实现他自己的目标，不是我们的目标。

Open source is a positive and important force for empowering people and preventing centralization that is detrimental for both safety and the economy.

开源是一股正面而重要的力量，它赋权于人，并防止那种对安全和经济都有害的中心化。

Meta continues to be strongly supportive of open source, including open source AI models.

Meta 继续坚定支持开源，包括开源 AI 模型。

The current open source ecosystem is strong, and we think it would be a mistake to restrict it.

当前的开源生态很强健，我们认为限制它会是个错误。

Now that Meta Superintelligence Labs are up and running, we will resume releasing some open source models soon.

既然 Meta Superintelligence Labs 已经运转起来，我们很快会恢复发布一些开源模型。

Independent governance and oversight is important given the high stakes nature of decisions around superintelligence.

鉴于围绕超级智能的决策事关重大，独立的治理与监督很重要。

I do not think it is in my, Meta's, or the world's best interests for me or anyone else to be a sole decision-maker on how superintelligence is deployed.

我不认为让我或任何其他人成为「超级智能如何部署」的唯一决策者，符合我、Meta 或这个世界的最佳利益。

Therefore, Meta is implementing a governance structure that gives our independent board of directors the power to approve the safety criteria for releasing models and reviewing whether each model release adheres to the criteria.

因此，Meta 正在实施一套治理结构，赋予我们的独立董事会权力去批准模型发布的安全标准，并审查每一次模型发布是否符合这些标准。

While Meta is a founder-controlled company, the CEOs of all frontier labs currently have extensive authority over model releases, so we encourage others to implement structures similar to this as well.

虽然 Meta 是一家创始人控制的公司，但目前所有前沿实验室的 CEO 对模型发布都拥有广泛权限，所以我们也鼓励其他公司实施类似的结构。

I think an industry-wide version of this process would be helpful for industry governance as well.

我认为把这套流程做成全行业版本，对行业治理也会有帮助。

Government policy is necessary to ensure a positive future and mitigating the risks discussed above.

要确保一个正面的未来、并化解上文讨论的各项风险，政府政策是必要的。

I have outlined several policy recommendations throughout this piece.

我在本文各处提出了若干政策建议。

Overall, we should encourage close cooperation between frontier labs and the government.

总体而言，我们应当鼓励前沿实验室与政府之间的紧密合作。

It is important for national security that the government has the technical resources to harden critical systems and make effective use of advanced models.

让政府拥有加固关键系统、有效使用先进模型的技术资源，对国家安全很重要。

But it is also important for national security that the American AI industry continues to lead.

但让美国 AI 产业继续领先，对国家安全同样重要。

We should accelerate the ability to build infrastructure -- whether that's energy, data centers, or silicon -- and build trust and alignment with communities across the country.

我们应当加快建设基础设施的能力——无论是能源、数据中心还是芯片——并与全国各地的社区建立信任与共识。

We should be careful to not slow down American competitiveness.

我们应当小心，不要拖慢美国的竞争力。

We should be skeptical of any proposals that lead to centralizing superintelligence, and we should favor policies that promote a healthy balance of power that favors people.

我们应当对任何导致超级智能中心化的提议保持怀疑，并支持那些促进「偏向人的健康权力制衡」的政策。

This is an incredible moment to live through.

这是一个了不起的、值得亲身经历的时刻。

Developing superintelligence will be the most profound technological advance we will see in our lifetimes.

开发超级智能，将是我们这辈子能看到的最深刻的技术跃迁。

As we get closer to this, it is increasingly important to have a clear philosophy and values for how superintelligence will benefit humanity.

越是临近这一刻，越需要有一套清晰的哲学与价值观，来说明超级智能将如何造福人类。

Meta is committed to building with the principles of individual empowerment as the source of prosperity, invention as AI's purpose, and a balance of power favoring people as the foundation for addressing safety risks.

Meta 承诺按以下原则来建设：个体赋权是繁荣的源头，发明是 AI 的目的，偏向人的权力制衡是应对安全风险的基础。

If these values lead the way, then I am optimistic that the coming decades will be some of the most amazing in history.

如果由这些价值来引路，我乐观地相信，接下来的几十年会是历史上最了不起的一段。

The arc of human civilization has bent towards putting more power in people's hands to live and shape the world in the ways we believe are best.

人类文明的弧线一直朝着「把更多力量交到人们手中」的方向弯曲，让人们按自己认为最好的方式去生活、去塑造世界。

Superintelligence holds the promise of giving everyone that power, and building a positive future for everyone.

超级智能承载着这样的许诺：把那份力量给到每一个人，并为每一个人建设一个正面的未来。

— Mark

— Mark