2018-04-09 OpenAI.Charter

“我们是一家非营利组织”，“我们的产品可能会导致人类文明终结”；

“我们第一轮投资者的回报上限是其投资额的 100 倍”；

“把对 OpenAI Global, LLC 的投资视为捐赠可能是明智的”；

“在后-[通用人工智能] 时代，金钱的作用可能难以确定”；

“我们的责任是对人类，而不是对你”；

“我们的产品太重要了，不能优先考虑投资者的利益”；

“我们的产品可能会让金钱，甚至人类本身，变得无关紧要”。

1、《OpenAI Charter》

Our Charter describes the principles we use to execute on OpenAI’s mission.

我们的章程描述了我们执行OpenAI使命的原则。

This document reflects the strategy we’ve refined over the past two years, including feedback from many people internal and external to OpenAI. The timeline to AGI remains uncertain, but our Charter will guide us in acting in the best interests of humanity throughout its development.

这份文件反映了我们在过去两年中完善的战略，包括来自OpenAI内部和外部许多人的反馈。AGI的发展时间表仍然不确定，但我们的章程将指导我们在其发展过程中始终以人类的最佳利益为出发点。

OpenAI’s mission is to ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity. We will attempt to directly build safe and beneficial AGI, but will also consider our mission fulfilled if our work aids others to achieve this outcome. To that end, we commit to the following principles:

OpenAI的使命是确保人工通用智能（AGI）——即在大多数经济价值高的工作中超越人类的高度自主系统——能够惠及全人类。我们将努力直接构建安全且有益的AGI，但如果我们的工作能够帮助他人实现这一目标，我们也将视使命为完成。为此，我们承诺以下原则：

Broadly distributed benefits 广泛分布的益处

We commit to use any influence we obtain over AGI’s deployment to ensure it is used for the benefit of all, and to avoid enabling uses of AI or AGI that harm humanity or unduly concentrate power.

我们承诺利用我们在AGI部署中获得的任何影响力，确保其用于全人类的利益，并避免使AI或AGI的使用对人类造成伤害或导致权力过度集中。

Our primary fiduciary duty is to humanity. We anticipate needing to marshal substantial resources to fulfill our mission, but will always diligently act to minimize conflicts of interest among our employees and stakeholders that could compromise broad benefit.

我们的首要信托责任是对全人类负责。我们预计需要调动大量资源来完成使命，但始终会谨慎行事，以尽量减少员工和利益相关方之间可能影响广泛利益的利益冲突。

Long-term safety 长期安全

We are committed to doing the research required to make AGI safe, and to driving the broad adoption of such research across the AI community.

我们致力于开展使AGI安全所需的研究，并推动此类研究在整个AI社区中的广泛采用。

We are concerned about late-stage AGI development becoming a competitive race without time for adequate safety precautions. Therefore, if a value-aligned, safety-conscious project comes close to building AGI before we do, we commit to stop competing with and start assisting this project. We will work out specifics in case-by-case agreements, but a typical triggering condition might be “a better-than-even chance of success in the next two years.”

我们担心后期AGI开发可能演变为一场竞争性的竞赛，导致没有足够的时间进行安全预防。因此，如果有一个与我们价值观一致且注重安全的项目比我们更接近构建AGI，我们承诺停止竞争并开始协助该项目。具体细节将通过逐案协议确定，但一个典型的触发条件可能是“在未来两年内成功的可能性超过五成”。

参考：《The Exciting, Perilous Journey Toward AGI | Ilya Sutskever | TED》，Ilya Sutskever在演讲的结尾讲到这个点，在OpenAI的创始团队中，可能只有Ilya Sutskever是真正相信的。

参考：《OpenAI's "Planning For AGI And Beyond"》。

This is more believable when OpenAI says it than when ExxonMobil does. There are real arguments for why an AI company might want to switch from moving fast and breaking things at time t to acting all responsible at time t + 1 . Let’s explore the arguments they make in the document, go over the reasons they’re obviously wrong, then look at the more complicated arguments they might be based off of.

当 OpenAI 说这句话时，比 ExxonMobil 说这句话更令人信服。确实有理由说明为什么一家 AI 公司可能想在时间 t 从快速行动和打破常规转变为在时间 t+1 表现得负责任。让我们探讨他们在文件中提出的论点，回顾他们显然错误的原因，然后看看他们可能基于的更复杂的论点。

OpenAI wants to sound exciting and innovative. If they say “we are exciting and innovative”, this is obvious self-promotion and nobody will buy it. If they say “we’re actually a dangerous and bad company, our products might achieve superintelligence and take over the world”, this makes them sound self-deprecating, while also establishing that they’re exciting and innovative.

OpenAI 想要听起来令人兴奋且富有创新性。如果他们说“我们令人兴奋且富有创新性”，这显然是自我宣传，没有人会相信。如果他们说“我们实际上是一家危险且糟糕的公司，我们的产品可能会实现超级智能并接管世界”，这让他们听起来自嘲，同时也确立了他们令人兴奋且富有创新性的形象。

The big thing all the alignment people were trying to avoid in the early 2010s was an AI race. DeepMind was the first big AI company, so we should just let them to their thing, go slowly, get everything right, and avoid hype. Then Elon Musk founded OpenAI in 2015, murdered that plan, mutilated the corpse, and danced on its grave. Even after Musk left, the remaining team did everything to challenge everyone else to a race short of shooting a gun and waving a checkered flag.

所有对齐人员在 2010 年代初期试图避免的最大问题是 AI 竞赛。DeepMind 是第一家大型 AI 公司，所以我们应该让他们慢慢来，把一切做好，避免炒作。然后埃隆·马斯克在 2015 年创立了 OpenAI，破坏了这个计划，肢解了尸体，并在其坟墓上跳舞。即使在马斯克离开后，剩下的团队也竭尽所能地挑战其他人进行竞赛，几乎到了开枪和挥舞方格旗的地步。

OpenAI still hasn’t given a good explanation of why they did this. Absent anything else, I’m forced to wonder if it’s just “they’re just the kind of people who would do that sort of thing” - in which case basically any level of cynicism would be warranted.

OpenAI 仍然没有给出一个好的解释说明他们为什么这样做。在没有其他信息的情况下，我不得不怀疑这是否只是因为“他们就是那种会做这种事情的人”——在这种情况下，基本上任何程度的愤世嫉俗都是合理的。

Technical leadership 技术领导力

To be effective at addressing AGI’s impact on society, OpenAI must be on the cutting edge of AI capabilities—policy and safety advocacy alone would be insufficient.

为了有效应对AGI对社会的影响，OpenAI必须处于AI能力的最前沿——仅靠政策和安全倡导是不够的。

We believe that AI will have broad societal impact before AGI, and we’ll strive to lead in those areas that are directly aligned with our mission and expertise.

我们相信，在AGI之前，AI将对社会产生广泛影响，我们将努力在那些与我们的使命和专业知识直接相关的领域中引领潮流。

Cooperative orientation 合作导向

We will actively cooperate with other research and policy institutions; we seek to create a global community working together to address AGI’s global challenges.

我们将积极与其他研究和政策机构合作；我们希望创建一个全球性社区，共同应对AGI的全球性挑战。

We are committed to providing public goods that help society navigate the path to AGI. Today this includes publishing most of our AI research, but we expect that safety and security concerns will reduce our traditional publishing in the future, while increasing the importance of sharing safety, policy, and standards research.

我们致力于提供有助于社会走向AGI的公共产品。目前，这包括发布我们的大部分AI研究，但我们预计出于安全和保障的考虑，未来将减少传统的研究发表，同时增加共享安全、政策和标准研究的重要性。

2、《2025-02-13 Matt Levine.OpenAI》

OpenAI

I’m actually a little disappointed that the core of OpenAI’s schtick didn’t work. It feels like it should have. The core of OpenAI’s schtick was that it raised billions of dollars from equity investors at eleven-digit valuations while telling them things like “we are a nonprofit organization” and “our product could bring about the end of human civilization” and “returns for our first round of investors are capped at 100x their investment” and “it would be wise to view any investment in OpenAI Global, LLC in the spirit of a donation” and “it may be difficult to know what role money will play in a post-[artificial general intelligence] world.” It conveys a certain confidence, to go around asking people for money while simultaneously saying “our duty is to humanity, not you” and “our product is too important for us to prioritize the interests of our investors” and “our product might make money, or for that matter humans, obsolete.” It’s a good pitch! “Business negging,” I once called it. The simplest investment heuristic is probably “if someone is desperate for my money, I shouldn’t give it to them,” and nobody has ever seemed less desperate for investor money than OpenAI.

其实我有点失望，OpenAI 的核心套路没能奏效。感觉它本应该奏效的。OpenAI 的核心套路是，它在告诉投资者诸如“我们是一家非营利组织”“我们的产品可能会导致人类文明终结”“我们第一轮投资者的回报上限是其投资额的 100 倍”“把对 OpenAI Global, LLC 的投资视为捐赠可能是明智的”“在后-[通用人工智能] 时代，金钱的作用可能难以确定”等话语的同时，以十一位数的估值从股权投资者那里筹集了数十亿美元。到处向人要钱，同时又说“我们的责任是对人类，而不是对你”“我们的产品太重要了，不能优先考虑投资者的利益”“我们的产品可能会让金钱，甚至人类本身，变得无关紧要”，这传递出一种特定的自信。这是个不错的推销方式！我曾称之为“商业贬低法”。最简单的投资经验法则可能是“如果有人迫切需要我的钱，我就不该给他们”，而从来没有哪家公司比 OpenAI 更不显得迫切需要投资者的钱。

加强版的“ Palantir”，骗子不是不相信自己说的话，骗子的问题是不能自洽，一边相信自己说的，一边在行动上又不能自洽，总是处于矛盾中。

The only problem is that OpenAI is desperate for investor money: It turns out that it costs a ton of money to build large-scale artificial intelligence models, so OpenAI has to keep raising huge piles of money. And at some point the investors want, not weird capped-profit interests in a complex waterfall of cash flows from a for-profit subsidiary of a nonprofit organization, but just normal equity stakes in a normal company that has normal obligations to them to maximize their profits. I kind of don’t know why? My mental model of SoftBank, in particular, is that if you go to Masayoshi Son and say “I am going to wipe out humanity and render money obsolete, give me $40 billion and maybe you’ll get something out of it, no promises,” he will be like “finally I have found my soulmate” and give you all of his money. But, no, apparently even at SoftBank, they don’t like to view their investments in the spirit of a donation.

唯一的问题是，OpenAI 迫切需要投资者的资金：事实证明，构建大规模人工智能模型需要大量资金，因此 OpenAI 必须不断筹集巨额资金。而在某个时候，投资者想要的不是在一个非营利组织的营利性子公司中，通过复杂的资金流动结构获得奇怪的有限利润权益，而是希望获得一家正常公司的正常股权，该公司对他们有正常的义务，即最大化他们的利润。我有点不明白为什么？尤其是我对软银的心理模型是，如果你去找孙正义说：“我要消灭人类，让金钱变得无关紧要，给我 400 亿美元，也许你能从中得到点什么，但不保证。” 他可能会说：“终于找到了我的灵魂伴侣。” 然后把他所有的钱都给你。但是，显然即使在软银，他们也不愿意把投资当作捐赠来看待。

孙正义-》马云-》芒格，相互吸引。

And so last year OpenAI decided to convert to a more normal for-profit company, saying:

因此，去年 OpenAI 决定转变为更正常的营利性公司，并表示：

The hundreds⁠ of⁠ billions of⁠ dollars that major companies are now investing into AI development show what it will really take for OpenAI to continue pursuing the mission. We once again need to raise more capital than we’d imagined. Investors want to back us but, at this scale of capital, need conventional equity and less structural bespokeness.

各大公司现在投入人工智能开发的数千亿美元表明，OpenAI 要继续追求这一使命究竟需要什么。我们再次需要筹集比预想更多的资金。投资者希望支持我们，但在这种规模的资本下，他们需要传统的股权结构和更少的定制化安排。

If I had Sam Altman as my salesperson I would try doubling down on structural bespokeness, but I guess I am wrong.

如果我的销售员是 Sam Altman，我会尝试加倍强调结构化定制，但我想我错了。

Anyway we discussed the odd fallout of this the other day. You can’t really take the assets of a nonprofit and give them to a for-profit company, so OpenAI will essentially have to sell its assets to itself: The nonprofit will have to give up its control over (and residual profit interest in) the for-profit subsidiary to satisfy investors, but to satisfy charity law, it will have to get paid the fair value of what it is giving up. The natural way to do that would be for the nonprofit to get back shares — conventional equity — in the newly independent for-profit company.

无论如何，我们前几天讨论了这件事奇怪的后果。你不能真的把一个非营利组织的资产转让给一个营利性公司，所以 OpenAI 本质上必须把自己的资产卖给自己：非营利组织必须放弃对营利性子公司的控制权（以及剩余利润权益）以满足投资者的需求，但为了符合慈善法，它必须获得相应的公平价值补偿。最自然的方式是让非营利组织获得新独立的营利性公司的股份——传统的股权。

How many shares? Well, intuitively, the nonprofit currently controls the business, so you might think it should get back a controlling stake in the for-profit business. (Like, 51% or more.) On the other hand, the majority of the business’s profits for the foreseeable future have been promised to outside investors (in the form of a structurally bespoke profit waterfall), and OpenAI is looking to raise many billions more dollars before it becomes profitable, which suggests that the nonprofit should get a minority stake. (The number 25% has been thrown around.)

多少股份？直觉上，非营利组织目前控制着这家企业，所以你可能会认为它应该在盈利性业务中重新获得控股权。（比如 51% 或更多。）另一方面，在可预见的未来，该企业的大部分利润已承诺给外部投资者（以结构上定制的利润分配机制的形式），而 OpenAI 计划在实现盈利之前再筹集数十亿美元，这表明非营利组织应该获得少数股权。（25% 这个数字曾被提及。）

In any case, there is some plan for a separation, and some negotiation over how much the nonprofit will get.2 But this plan arguably puts OpenAI in play: If the nonprofit is looking to sell its stake in the for-profit business (to the for-profit business, for stock), then arguably it should entertain higher offers, for instance from Elon Musk. Who lobbed in a $97.4 billion cash bid this week.

无论如何，都有某种分离计划，并就非营利组织将获得多少资金进行了一些谈判。但可以说，这一计划使 OpenAI 进入了交易状态：如果非营利组织希望出售其在盈利性企业中的股份（出售给盈利性企业，以换取股票），那么它理应考虑更高的报价，例如来自埃隆·马斯克的报价。而他本周提出了 974 亿美元的现金收购要约。

“In play” is a term from public-company mergers and acquisitions. There is a notion in M&A that a company is never obligated to sell itself — it can always reject an acquisition offer and decide to stay independent — but, if it does decide to sell itself, it has an obligation to shareholders to take the highest bid.3 I don’t think that’s quite true for a nonprofit like OpenAI, but you can see Musk’s point: If the nonprofit is planning to sell its valuable asset, it shouldn’t take less than fair value, and fair value is set by the market.

“在交易中”是一个来自上市公司并购的术语。在并购中，有一种观念认为公司从来没有义务出售自己——它始终可以拒绝收购要约并决定保持独立——但是，如果它决定出售自己，就有义务为股东选择最高的出价。我认为这对像 OpenAI 这样的非营利组织来说并不完全适用，但你可以理解马斯克的观点：如果非营利组织计划出售其有价值的资产，它不应该接受低于公允价值的价格，而公允价值是由市场决定的。

But if OpenAI is not in play, never mind:

但如果 OpenAI 不在考虑范围内，那就算了

Elon Musk said he would withdraw his $97.4 billion bid for the nonprofit that controls OpenAI if the board stopped its conversion to a for-profit company.

埃隆·马斯克表示，如果董事会停止将 OpenAI 转换为营利性公司，他将撤回自己 974 亿美元的收购要约。

“If [the] OpenAI board is prepared to preserve the charity’s mission and stipulate to take the ‘for sale’ sign off its assets by halting its conversion, Musk will withdraw the bid,” lawyers for Musk said in a court filing Wednesday.

“如果 OpenAI 董事会准备维护该慈善机构的使命，并通过停止转换来规定撤下其资产的‘待售’标志，马斯克将撤回报价，”马斯克的律师在周三的法庭文件中表示。

Sure! In general you can’t just lob in a hostile bid to buy a nonprofit’s assets4; if OpenAI just stayed as a nonprofit, Musk would not have much leverage to buy it. (Though he could — and did — sue it for not being nonprofit enough.)

当然！一般来说，你不能随便发起敌意收购来购买非营利组织的资产；如果 OpenAI 仍然是非营利组织，马斯克就没有太多筹码来收购它。（不过他可以——而且确实——起诉它不够“非营利”。）

You can see the fairly obvious point here. Musk runs an artificial intelligence company, xAI. It started later and is smaller than OpenAI, but Musk has big ambitions. He has raised a lot of money for it. OpenAI is bigger and wants to raise more money, and investors want to give it that money, but only for “conventional equity and less structural bespokeness.” xAI can offer conventional equity; OpenAI can’t, yet, until it converts. If OpenAI fends off Musk by staying a nonprofit, that will cripple its ability to raise money and scale, and will put xAI in a better competitive position.

你可以看到这里相当明显的要点。马斯克经营着一家人工智能公司，xAI。它起步较晚，规模比 OpenAI 小，但马斯克有远大的抱负。他为此筹集了大量资金。OpenAI 规模更大，并希望筹集更多资金，投资者也愿意提供资金，但前提是“传统股权和更少的结构化定制”。xAI 可以提供传统股权，而 OpenAI 目前还不能，除非它转换结构。如果 OpenAI 通过保持非营利性质来抵御马斯克，那将削弱其筹资和扩展能力，并使 xAI 处于更有利的竞争地位。

I mean, that’s clearly how OpenAI sees the problem. I kind of want them to be wrong. Give it a try, you know? Tell Musk “sure, we’ll stay a nonprofit,” and then go back to SoftBank to be like “we still want your $40 billion at a $300 billion valuation but, small hitch, to fend off Elon Musk, you will have to view it in the spirit of a donation.”

我的意思是，OpenAI 显然是这样看待这个问题的。我有点希望他们是错的。试试看吧，你知道吗？告诉马斯克：“当然，我们会保持非营利组织的身份。” 然后再回到软银，说：“我们仍然想要你们的 400 亿美元，估值 3000 亿美元，但是，有个小问题，为了抵挡埃隆·马斯克，你们得把这笔钱当作捐赠来看待。”

2018-04-09 OpenAI.Charter

2018-04-09 OpenAI.Charter

1、《OpenAI Charter》

Broadly distributed benefits 广泛分布的益处

Long-term safety 长期安全

Technical leadership 技术领导力

Cooperative orientation 合作导向

2、《2025-02-13 Matt Levine.OpenAI》

热门主题

Related Articles

2024-05-29 Microsoft Corporation (MSFT) Jefferies Software Conference (Transcript)

2024-10-30 Microsoft Corporation (MSFT) Q1 2025 Earnings Call Transcript

2025-01-29 Microsoft Corporation (MSFT) Q2 2025 Earnings Call Transcript

2023-07-25 Microsoft Corporation (MSFT) Q4 2023 Earnings Call Transcript

2024-04-25 Microsoft Corporation (MSFT) Q3 2024 Earnings Call Transcript