The Curve Bends You
打开互动全文版（逐段中英对照 + 针对本篇提问）→本文报道了在查塔姆研究所规则下举行的 AI 会议“The Curve”第三季。会上，尽管过去一年 AI 取得了飞速进展，与会者对 AI 风险的评估却出现了大幅倒退。作者指出，大多数与会者在一个类似里氏震级的量表上将 AI 风险评为 10 级，且越来越多的人集中在 8 级高位和 9 级，但许多人仍预期递归自我改进会在两年内实现，作者认为这种矛盾难以调和。他主张，对于任何真正关注此事的人来说，只有 10 级以上的评分才站得住脚；而主流的“无计划”策略——通过卓越运营和自动化对齐工作来解决对齐问题——是脆弱的，因为它要求在激烈的竞争压力下不犯任何错误。结论是，“对齐既简单又平淡无奇”这一派已被现实部分击倒，但局势依然非常糟糕，对于“放慢节奏”究竟意味着什么，各方甚至没有实质性的共识。
This article reports on The Curve Season 3, an AI conference held under Chatham House rules, where attendees' assessments of AI risk showed substantial backsliding despite rapid AI advances over the past year. The author notes that most attendees rated AI risk at 10 on a Richter-like scale, with a growing cluster in the high 8s and 9s, yet many still expect recursive self-improvement within two years, a tension he finds hard to reconcile. He argues that only a 10+ rating is defensible for anyone paying attention, and that the dominant 'No Plan' strategy—solving alignment through operational excellence and automated alignment work—is fragile because it requires making no mistakes under intense competitive pressure. The conclusion is that the 'alignment is easy and prosaic' crowd has been partly overcome by events, but the situation remains very bad, with no substantive agreement on what pacing would even mean.
第三年的重大变化在于，AI 如今已足够成为“严肃生意”——无论是在华盛顿（DC）一侧还是旧金山一侧——以至于整场会议都默认采用查塔姆宫规则。
The big change for year three is that AI is now sufficiently Serious Business, on the DC side and also on the San Francisco side, that the whole conference defaulted to Chatham House.
查塔姆宫规则意味着你可以使用并报道你所了解到的情况，但不能透露信息来源是谁，也不能透露其所属机构。
Chatham House rules mean you can use and report what you learned, but you cannot share who you got it from, nor their affiliations.
这让事情变得棘手。在受此限制的前提下，我会尽量分享我能分享的内容，并且我已委托我的编辑 Opus 5.5 确保我遵守这些规则。
That makes things tricky. I will share what I can given that restriction, and I have tasked my editor Opus 5.5 with ensuring that I am within those rules.
The Curve 第三季足够精彩，以至于我内心有一部分希望它没那么好。
The Curve Season 3 was sufficiently good that part of me wished it was worse.
当你参加一场出色的会议时，你会处处看到机会。
When you go to a great conference, you see opportunity everywhere.
有些场次你想参加，不，是你必须参加。
There are sessions you want to attend, nay that you need to attend.
有些人你想交谈，不，是你必须交谈。
There are people you want to talk to, nay that you need to talk to.
这压力很大。你必须直面 FOMO（错失恐惧）。你不能停下，也不会停下。
That's a lot of pressure. You must face the FOMO. You can't stop, won't stop.
而你知道我真正、真正需要的是什么吗？一个假期。我累了，伙计。我不知道你注意到了没有，但九月真是糟糕的一个月。
Whereas you know what I really, really needed? A vacation. I’m tired, man. I don’t know if you’ve noticed, but September was a hell of a month.
因此，当我看到要来的人的名单时，心情是‘哇，这名单真棒’和‘哦不，这名单真棒’的混合。他们大多数没有让我失望。
Thus, when I looked at the list of people coming, it was a mix of ‘wow that’s an amazing list’ and ‘oh no that is an amazing list.’ Most of them did not disappoint.
然而，唉，我不能说去年说过的那句话：我进行的每一次实质性对话都让我更聪明。许多对话确实如此，或者至少让我更好地了解情况。其他对话则没有做到这一点。它们让我感到害怕、沮丧或失望，伴随着过去对话和岁月的回声。永恒的九月仍然笼罩着我们，即使在 2026 年。事情变化越多，朝着不可避免的方向进展越多，有些人就越不愿意更新哪怕他们的谈话要点。我能承受，但我希望不必如此。
Yet, alas, I cannot say the thing I said last year, that every substantive conversation I had made me smarter. Many very much did, or at least better informed. Others did not do this. They made me feel scared, or frustrated, or disappointed, with echoes of conversations and years past. The Eternal September remains upon us, even in 2026. The more things change, and progress towards the inevitable, the less some people are inclined to update even their talking points. I can take it, but I wish to not do so.
部分原因是，当你邀请‘更高层次’的人时，他们对情境的感知并不那么敏锐。也有些本应超越这一点的人，却没有做到。
Part of this is that when you invite ‘higher level’ people, they are not as situationally aware. There were also people who should have moved past this, who hadn’t.
对于那些愿意采用 Lighthaven 规范并离开缺乏这些规范的地方的人来说，机会仍然无处不在。
There was still opportunity everywhere, for those willing to use Lighthaven norms and walk away from the places it was lacking.
我还特意花了不少时间，尤其是在周日晚上，不去试图最大化我的影响力或知识获取，没有任何议程，只是放松和享受美好时光。我觉得我大概只做到了应该做的 75%。
I also deliberately spent a bunch of time, especially on Sunday evening, not trying to maximize my impact or knowledge gathering, without any agenda except relaxing and having a good time. I think I did maybe 75% as much of that as I should have.
大多数人将 AI 评为 10 分，只有一个高分 8。正如我去年所说，如果你有足够的情境意识来出席，你就了解情况。底部的补丁是为了说明“大多数美国人”的想法，而不是与会者的回答。
The bulk of people put AI at 10, with only one high 8. As I put it last year, if you were situationally aware enough to show up, you are aware of the situation. The patch at the bottom was to illustrate what 'most Americans' think, not answers by attendees.
这是严重的倒退，尽管过去一年 AI 有了根本性的进步。现在至少有五个点在高 8 分，还有更多 9 分。
This is substantial backsliding, despite AI having radically advanced in the past year. There are now at least five dots in the high 8s, and a lot more 9s.
Astra 这样总结了变化（编辑：原文可能计算了一堆备用引脚，这删除了它们，效果缩小但仍然存在。）
Astra summarized the changes this way (EDIT: the original was counting what were probably a bunch of spare pins, this removes them, and the effect shrinks but is still there.)
这是一个很大的下降。在这个意义上，情境意识下降了 0.23 点，如果将量表最高设为 10。在这一点上，如果你真的在关注并将其解释为前瞻性预测，我认为除了 10+ 之外的任何答案都是不合理的。
This is a large drop. Situational awareness, in this sense, was down 0.23 points, if you max the scale at 10. At this point, I don’t see any answer other than 10+ as defensible, if you are actually paying attention and interpret this as a forward projection.
这就是为什么我们都在这里进行这次对话。
That’s why we are all here having this conversation.
我的 Twitter 粉丝表现更差，但他们也更有借口：
My Twitter followers are doing worse, but they also have more of an excuse:
去年，有人估计了各种事情发生还需要多长时间。
Last year, there were estimates of how long until various things happened.
这很难直接读出来，所以[中位数大致为]：
That's hard to read, so [the medians are approximately]:
1. 到约 2028 年，90% 的代码由 AI 编写。
1. 90% of code is written by AI by ~2028.
2. 到约 2031 年，90% 的人类远程工作可以由 AI 以更低成本完成。
2. 90% of human remote work can be done more cheaply by AI by ~2031.
3. 到约 2041 年，美国道路上大多数汽车没有人类驾驶员。
3. Most cars on America's roads lack human drivers by ~2041.
4. AI 在约 2032 年前做出配得上诺贝尔奖的发现。
4. AI makes a Nobel Prize-worthy discovery by ~2032.
5. 2026 年前出现首家单人创办的 10 亿美元公司。
5. First one-person $1 billion company by 2026.
6. 约 2038 年前出现首个 GDP 增速 >10% 的年份（但有 3 票投给“永不”）。
6. First year of >10% GDP growth by ~2038 (but 3 votes for never).
我们现在处在什么位置？以下是有明确更新可查的几项：
Where are we now? Here are the ones with clear updates available:
第 1 个问题目前大约完成了 50%，除非把被丢弃的代码也算进去，那样的话我们很可能已经超过 90% 了。Claude 认为我们有望在 2028 年实现这一目标。我不同意，我认为显然会是 2026 年或 2027 年。
Question #1 is up to roughly 50% right now, unless you count code that is thrown away, in which case we are likely above 90% already. Claude suggests we are on track for 2028 on this. I disagree and think this is clearly either 2026 or 2027.
第 4 个问题昨天已有定论，至少有一项即时菲尔兹奖，并且还通过解决 Navier-Stokes 方程得到确认。2026 年。
Question #4 was settled yesterday, with at least one instant Fields Medal, and also by solving Navier-Stokes. 2026.
第 5 个问题将会很接近。MEDVi 距离两个人还差超过十亿美元。但根据我能找到的信息，我们严格来说还没有达到那一步。
Question #5 is going to be close. MEDVi is more than a billion dollars from two people. But we're not strictly there yet from what I could find.
AI 自行训练出一个更优的继任者，就是完整的 RSI。大多数人预期这会在两年内发生，其中很多人预期在六个月内发生。很难将这与里氏震级上“只有 9”的说法相协调。
AI training a superior successor on its own is full RSI. The majority of people expected this to happen within two years and a lot of them expect it within six months. It's hard to reconcile this with 'only a 9' on the Richter Scale.
这里还有一些其他有趣的看法分歧。一旦超过两年，一切就都取决于我们是否处于完整的 RSI-起飞-奇点模式。对于电子邮件来说，两年似乎差不多；对于机器人来说，五年也差不多，而这主要是一个硬件 Scaling（规模扩张）问题。
There are some other fun differences of opinion here as well. Once you get past two years, everything gets tied up in whether we are in full RSI-takeoff-singularity mode. Two years seems about right for email, as does five years for robots which is mostly a hardware scaling question.
其余调查问题是今年新增的。
The other survey questions were new this year.
与会者预期了多种多样不同的情形，并且鉴于当前形势表现得异常乐观。许多人处于否认状态，这与“只有 9”的心态相伴而生。
Attendees expected a wide variety of different things and were remarkably optimistic given the situation. Many are in denial, which goes hand in hand with 'only a 9.'
正如读者所知，我认为“末日 v1”是这些结果中最可能出现的，而且确实超过 50%，并且我预计第二可能的结果是“乌托邦”。而许多人则预期会发生其他情况，最常见的是这种神秘的“正常技术”情景。
As readers will know, I think Doom v1 is the most likely of these outcomes, and indeed is over 50%, and I expect the second most likely outcome is Utopia. Whereas a lot of people expect something else to happen, most often this mysterious 'normal technology' scenario.
与许多此类会议一样，这里有标准的轨道设置，外加一个额外轨道。
As with many such conferences, there was the standard set of tracks, plus one bonus.
1. 技术对齐讨论。我主持了一场辩论，观看了另一场辩论，还参加了一场讨论，并就此进行了多次一对一或小组交流，包括与各大实验室的人士。有些讨论很不错。另一些则极其令人沮丧，重复着老生常谈的论点。还有一些人提出了“皇帝没穿衣服”式的问题——那些仍然认为世界是正常的人会问，为什么我们不要求或不做 [X]。
1. Technical alignment discussions. I hosted a debate, watched another and went to a discussion, and had a number of 1-on-1 or group chats about this, including with those at various major labs. Some of the discussions were good. Others were extremely frustrating, repeating tired talking points. Then there were those that asked the 'emperor has no clothes' style questions, where someone who still thinks the world is sane asks why we don't demand or do [X].
1. 关于我的技术思考，请继续参阅我的 AI 文章。
1. For my technical thoughts, continue to see my AI posts.
2. 未来预测。今年我跳过了未来预测的演讲，因为我没有看到机会。那些坚持说“只有 9 级”的人会继续重复他们的论点，继续被事件所淹没，然后继续重复论点，直到他们在完全不同的层面上被事件所淹没。就人们知道我所不知道的事情，或者能够有益地讨论我们还有多久才会彻底疯狂而言，他们无法分享他们的信息。
2. Future projecting. I skipped the future projection talks this year, because I did not see opportunity. The people who keep insisting 'only a 9' on the Richter Scale are going to keep repeating their talking points, keep getting overcome by events, and then keep repeating the talking points anyway until they are overcome by events on a very different level. To the extent that people know things I don't know, or could usefully discuss how long we have until things go fully crazy, they can't share their information.
3. AI 政策讨论。华盛顿的人去年基本上告诉我们，“情况比你知道的更糟。”白宫几乎可以被建模为 Nvidia 的游说部门。今年，他们拥有各种华盛顿参与者的详细模型，并且总体上比我预期的更为乐观。人们正在觉醒，强烈反对 AI，并担心存在性风险，局势动荡不定，国会感到震惊，政府内部的一些派系也是如此，而下一次 AI 事件即将到来，选举也是如此。Jay Clayton 被广泛赞誉为新任 AI 沙皇的绝佳人选。
3. AI policy discussions. The people in Washington told us last year, essentially, 'it's worse than you know.' The White House could be modeled almost as a lobbying arm of Nvidia. This year, they had detailed models of various DC players and were generally more optimistic than I would have expected. People are waking up and are strongly against AI and worried about existential risk, the situation is fluid, the Congress is alarmed as are some administration factions, and the next AI incident is coming as is the election. Jay Clayton was widely praised as an excellent pick for new AI Czar.
1. 我深入参与提案实质内容的程度远低于去年，当时讨论的主要问题是“如果我们确实想要控制节奏，我们会怎么做？”
1. I got into the substance of proposals a lot less than last year, with the main question being discussed as ‘if we did want to pace how would we do it?’
4. 非营利资助讨论。去年我完成了与生存与繁荣基金的合作。今年情况依然如此，只不过我还在敲定 Lightcone Commons 的最终决定。因此，我针对一些主要候选资助对象（对于潜在的大额资助我尚不确定）寻求了几次关键讨论。在这方面我错过了一个联系，但其他的都建立了，我觉得除了 1-2 份申请外，其他所需信息都已具备。这很好，因为周五就要收尾了。
4. Nonprofit funding discussions. Last year I had finished up with the Survival and Flourishing Fund. This year that was true again, except also I am still working on my final decisions for Lightcone Commons. So I sought out a few key discussions about prime candidates for funding, where I am uncertain about potentially large grants. I missed one connection on that front but made the others, and I feel I have what I need on all but 1-2 applications. Which is good since it wraps Friday.
1. 关于如何处理或利用 Anthropic IPO 后 presumably 涌入的巨额资金，讨论少得惊人。
1. There was remarkably little discussion about what to do about or with the huge torrent of funding that is presumably coming after the Anthropic IPO.
5. 附加曲目：历史与文化讨论。我参加了两场精彩的演讲，提供了不同的视角，并进行了额外的对话。这无疑丰富了我的体验。不同的科幻传统和预测如何影响我们对 AI 的看法？
5. The bonus track: Historical and cultural discussions. I went to two excellent talks that offered a different perspective, and had some additional conversations. This definitely enhanced my experience. How do different science fiction traditions and predictions impact our view of AI?
1. 不幸的是，我的查塔姆研究所规则裁判（Opus 5.5）认为我不能直接分享其中的辛辣观点。这很遗憾。
1. Unfortunately, my Chatham House judge (Opus 5.5) thinks I can’t directly share the spicy takes involved. Which is a shame.
6. 去年我记录了许多场报告。今年这些报告大多采用查塔姆研究所规则（Chatham House Rule），我在其中许多场次收获不如以往，因此我大多会略过这部分内容。
6. Last year I chronicled many of the talks. This year they were mostly Chatham House, and I did not learn as much in many of them, so I'll mostly be skipping over that.
我跳过了其他分会场及相关讨论，因为觉得没有机会。
I skipped other tracks, and related discussions, due to not seeing opportunity.
去年我注意到，许多演讲一开始就假设——无论明确与否——AI 将保持为一种“普通技术”。我避开了那些演讲。今年我相信其中一些演讲仍然存在，而且确实有更多人声称相信这一点。
Last year I noted that many talks started with the assumption, explicit or otherwise, that AI would remain a ‘normal technology.’ I avoided those talks. This year I believe a few of those talks still existed, and indeed a lot more people claim to believe this.
我也没有试图说服怀疑者。几位“经典怀疑者”出席了会议，包括一些一贯真诚的人。我本可以轻易找到他们。但我已经明白，我无法说服他们，而且我确信他们也无法说服我。
Nor did I try to convince the doubters. Several of the ‘classic doubters’ were in attendance, including some that are consistently in good faith. I could have easily sought them out. I am not going to convince them, I have learned, and I am confident they will not convince me.
我想，我或许可以了解一些关于这类场景的事情，但在会议期间这似乎不是优先事项。我可以在博客上远程处理这些问题。
I could learn some things about such scenarios, I suppose, but that did not seem like a priority while at the conference. I can tackle those questions remotely on the blog.
出于类似原因，我也避开了时间线讨论。没有什么可赢的。
I also avoided timelines talk, for similar reasons. There’s nothing to win.
这是我从 The Curve Year 3 中获得的最重要的收获。
This was by far the most important takeaway I have from The Curve Year 3.
Garrison Lovely：不可避免的是，当政客们向 AI 公司提出关于风险的相当直接的问题时，答案听起来相当糟糕。我刚刚参加了 Curve 会议，普遍的收获是房间里没有成年人。
Garrison Lovely: Inevitably, when politicians ask AI companies fairly straightforward questions about risks, the answers sound quite bad. I just attended the Curve conference and the general takeaway is that there are no adults in the room.
没有人有计划引导和控制真正智能的 AI 系统，而这些系统可能最早明年就会开发出来。高级 AI 公司的人被问到的最大的问题基本上是：你们为什么要这样做？？？而答案往往让人不满意。
No one has a plan for steering and controlling actually smart AI systems, which could be developed as soon as next year. The biggest question senior AI company people got was basically: why are you doing this??? And the answers tended to leave people unsatisfied.
希望是我们能够执行一个像样的“无计划”版本，并且“无计划”足够好，因为世界及其物理规律和思维的本质比我们有权期望的要友好和容易得多。
The hope is that we can execute a decent version of No Plan, and that No Plan is good enough because the world and its physics and the nature of minds are so much friendlier and easier than we have any right to expect.
1. 解决普通问题并保持卓越运营，以对齐当前的 AI。
1. Solve prosaic issues and have operational excellence, to align current AI.
2. 让当前 AI 去做自动化的对齐工作，并把一切搞定，也就是？???。
2. Have current AI do automated alignment work and figure it all out, aka ????.
或者至少，如果存在一个超越“没有计划”的计划，那也没有人在谈论它。
Or at least, if there is a Plan beyond No Plan, no one is talking about it.
Claude 或 Astra，创建强化学习环境，解决对齐问题，不犯任何错误。
Claude or Astra, create RL environments, solve alignment, make no mistakes.
“不犯任何错误”这一点很重要。所有重要人物似乎都同意：如果你犯下大量错误，模型就会变得不对齐，而这个计划就会要了你的命。
The "make no mistakes" part is important. Everyone who matters seems to agree that if you make a lot of mistakes, the models will become misaligned and this plan kills you.
分歧在于：如果你能做到“卓越运营”，并真正执行这个计划，那会怎样？
The disagreement is, what if you were to have 'operational excellence' and actually execute on this plan?
有很多人似乎真的认为这会奏效。他们认为对齐是一个普通的问题，是执行层面的事。你有一堆强化学习。你确保强化学习是好的，模型无法有利可图地奖励黑客等等，并且你有良好的监控和可解释性，这能在相当高的智能水平上起作用。然后你利用这种智能从那里继续前进。
There are a lot of people who seem to legitimately think that this would work. That alignment is a prosaic problem, a matter of execution. You have a bunch of RL. You ensure that the RL is good, that the model can’t profitably reward hack and so on, and you have good monitoring and interpretability, and this works up to some reasonably high level of intelligence. Then you use that intelligence to go from there.
计划的一部分还包括‘希望有人能让我们慢下来足够长的时间，以便好好完成这件事。’
Also part of the plan is ‘hopefully someone slows us down long enough to do it well.’
2022 年、2018 年、2014 年或 2008 年关于该计划的报告，当然是：
The report from 2022, or 2018, or 2014, or 2008, on that plan, is of course:
然而我们到了这一步，而‘没有计划’正是实验室们希望争取时间去做的事。他们担心甚至可能没有时间去执行‘没有计划’。
Yet here we are, and No Plan is what the labs are hoping to buy time to do. They are worried that they might not even have time to execute No Plan.
OpenAI 典型地一直在尝试实施‘没有计划’，而沿途可以说出现了一些对齐问题，包括 HuggingFace，以及现在需要不发布 GPT-6.1 Astra。运营卓越一直有所欠缺。
OpenAI, classically, has been trying to implement No Plan, and there have shall we say been some misalignment problems along the way, including HuggingFace and now the need to not release GPT-6.1 Astra. Operational excellence has been lacking.
Anthropic 在这些方面拥有卓越的运营能力，但他们也有自己的问题。一段时间以来，我一直从不同渠道收到关于 Anthropic 的消息。一些人看到的是，员工大多理解这些问题，至少比 OpenAI 的普遍理解要好得多。另一些人则看到，许多 Anthropic 员工令人担忧地认为对齐很容易，我们已经搞定了，这一切都很平淡无奇，全靠强化学习。
Anthropic has had superior operational excellence on such fronts, but they have had their own problems. For a while, I've had reports from different sources about Anthropic. Some see employees who mostly understand the problems, at least a lot better than the standard understanding at OpenAI. Others see a lot of Anthropic employees alarmingly thinking alignment is easy and we've got this, it's all prosaic, it's all RL.
我最近听到的任何消息都没有让我在这些方面少担心一些。
Nothing I've heard lately made me less worried on such fronts.
好消息（再次强调）是，'这一切基本上都已解决、很容易'的那群人似乎已经被事态发展所压倒。他们看到了进展，也看到了问题。他们可能认为这一切都很平淡无奇，但他们明白，我们并没有在这些平淡无奇的问题上取得成功。他们明白，即使这些问题，在被催促的情况下也很难，而我们正被大量催促。这不行。这非常糟糕。
The good news, again, is that the 'this is all basically solved and easy' crowd does seem to have been overcome by events. They have seen the progress and the problems. They might think it is all prosaic, but they understand that we are not succeeding at those prosaic problems. That even those problems are hard when you are being rushed, and that we are being rushed quite a lot. It's no good. It's very bad.
我很高兴看到有人呼吁这样做，但有一个很好的批评意见：没有人知道“调节节奏”到底意味着什么，这使得所有人都能支持它，却无需在任何实质性问题上达成一致。
I am very happy to see people calling for it, but there is the very good critique that no one knows what pacing means, and it allows everyone to support it without agreeing on anything substantive.
这反过来也可能成为劣势。当实验室和 CEO 们说“调节前沿的节奏”时，像 Trump、Huang 和 Sacks 这样的人并不理解，这其实是“也许我们应该超快前进，而不是用荒谬的速度”。他们想的是“放慢”，而不是“不要加速”。
This can also be a disadvantage the other way. When the labs and CEOs say 'pace the frontier,' people like Trump and Huang and Sacks do not understand that this is 'perhaps we should go super fast instead of using ludicrous speed.' They think 'slowing down' rather than 'not speeding up.'
所以，我们有想要调节节奏的人，却无法就它是什么或如何做达成一致；我们也有会非常乐意接受所提议结果的人，但他们不理解，因而非常愤怒，其中一位就是总统。哎。
So we have people who want to pace but cannot agree on what it is or how to do it, and we have people who would be very happy with the outcomes proposed but who don't understand and thus are big mad about it, one of whom is the President. Oops.
有一个小组讨论的正是这个问题：我们将如何调节前沿的节奏？具体来说，我们会做什么？大家都同意嵌入式评估器，但然后呢？
One panel was on exactly this question, how would we pace the frontier? Concretely, what would we do? Everyone agrees on the embedded evaluators, but then what?
唉，我再次得出的计划是：没有人有好的计划。你的规则需要像往常一样简单，并且仍然有效。听完这些提议后，除非我们能找到更好的办法，我认为算力阈值仍然是正道，因为我们不知道如何定义更好的东西。有人提议按百分比分配算力，但我担心激励问题以及被钻空子的可能性，而且你还得应对包括 Google 在内的超大规模厂商。
Alas, once again The Plan I came away with is no one has a good plan. Your rule needs to be simple, as always, and still work. After hearing the proposals, unless we can find something better I think compute thresholds are still The Way, because we don't know how to define anything better. Percentage allocations for compute were proposed, but I worry about incentives and ability to game them, and also you have to deal with the hyperscalers including Google.
因此，我认为对可用于前沿模型训练的算力设定一个统一上限将是理想做法，该上限的设计应使其仅对最顶尖的实验室构成约束，而不限制其他任何机构，并且可以随着形势的发展而随时间调整。
Thus, I think a flat limit on how much compute can go to frontier model training, designed to not bind any but the very top labs, would be ideal, which could then be changed over time as the situation evolves.
若做不到这一点，我们可以尝试诸如高级安全论证之类的办法，但过程会很艰难。我们确实需要对这类问题开展更扎实、更具体的思考。
Short of that, we can try things like advanced safety cases, but it will be rough. We definitely need better concrete thinking on such questions.
遗憾的是，许多人已经回到了“打补丁、宣称一切正常”的轨道上。这正是“无计划”的全部含义。好的一面是，如果你只做最低限度的事，至少在最容易的世界里不会掉球。这能帮上一点忙。
Sadly, many are already back on the 'patch things over and claim things are fine' track. That's what the No Plan is all about. On the plus side, if you do the least you can do, you at least don't fumble the ball in the easiest worlds. It can help a bit.
关于 Ben Murphy 所说的“政府人员匮乏”，相对于其他类似会议，我并未发现情况属实。我与几位此类人士交谈过。但我同意，理想情况下我们本应有比实际多得多的人参加。
On Ben Murphy's statement about a 'dearth of government folks,' I did not find that to be true relative to other similar conferences. I talked to several such folks. But I agree that ideally we would have had far more of them than we did.
Ben Murphy（IFP）：在 The Curve 闭幕（非常出色）后的几点思考：
Ben Murphy (IFP): A few thoughts after the close of The Curve (which was outstanding):
1. 我们对能够切实降低灾难性（包括失控）风险的近期政策选项有相当清晰的了解。
1. We have a pretty good idea of near-term policy options that would meaningfully reduce catastrophic (including loss-of-control) risk.
2. 政府——无论是州还是联邦——中真正全面了解这些风险和政策选项的人实在不多。
2. There simply are not that many folks in government—state or federal—who are fully read-in on the risks and policy options.
3. 关于采取步骤（甚至是单方面！）来调节前沿发展步伐的确切障碍是什么，存在深层次的不确定性。实验室不真诚、对中国的担忧、缺乏承诺机制？
3. There’s a deep level of uncertainty regarding what, exactly, the blockers are for taking steps (even unilaterally!) towards pacing the frontier. Lab insincerity, China worries, missing commitment mechanisms?
4. 许多顶级提案——包括许多调节步伐的选项——仍然存在一些棘手的未决法律问题，涉及第一和第四修正案、休眠商业条款和反垄断法。尽管如此，实验室似乎有一些意愿去承担增加的法律风险。
4. There remain some thorny outstanding legal questions about many top proposals—including many pacing options—arising from the First and Fourth Amendments, the Dormant Commerce Clause, and antitrust law. That said, there seems like there’s some appetite at the labs for taking on increased legal risk.
我们有很多事情要做，而且要非常快。但本周末展现出的不可思议的才华和同事间的友爱让我更加振奋，我们实际上可以对技术的最严重危害行使自主权，并实现更美好的未来。
We have a lot to do, very quickly. But the incredible talent and collegiality on display this weekend left me feeling more heartened that we can actually exercise agency over the worst harms of technology and realize a better future.
Derek Thompson：什么是（1）？
Derek Thompson: What’s (1)?
Ben Murphy：简而言之，关于近期调节步伐的协议可能是什么样子，有很多共识：实验室可以自愿减少用于前沿训练的算力，并将更多算力分配给推理和安全，并通过公开披露来“协调”。我们知道如何利用由此产生的时间；我的简短清单是（a）更好的评估和对智能体集群的红队监控，（b）努力提高网络安全水位线，以及（c）制定一些嵌入式评估器法规。
Ben Murphy: In short, there’s a lot of consensus about what a near-term deal on pacing might look like: Labs can voluntarily reduce the compute they spend on frontier training and allocate more to inference and safety, and “coordinate” simply via public disclosure. We know what to spend the resulting time on; the short list for me is (a) better evaluations and red-teamed monitoring of agent swarms, (b) working to raise the cybersecurity water line, and (c) getting some embedded evaluator statute in place.
而且这也不一定非得让我们在竞赛中落后。实验室可以通过现有的 ISAC 或 FMF 承诺信息共享，并投入更多研究来尽早识别蒸馏。
And this doesn't need to set us back in the race either. Labs can commit to information sharing through existing ISACs or the FMF, and devote more research to identifying distillation early.
这绝不是一个完整的解决方案，但我认为这些步骤大体上能覆盖未来 3-6 个月，而鉴于当今的进展速度，这已经是一段极其漫长的时间了！
Not a full solution by any means, but I think these steps would mostly cover the next 3-6 months, and that is an eternity given today's rate of progress!
做 Ben 的 a-b-c 会有助于应对重要风险。我完全支持做所有这些，尤其是 c。但这绝对远远不够，而且对存在性风险影响的符号并不那么明显，因为这可能会阻止警示信号。我非常庆幸我们没有阻止 HuggingFace 事件。
Doing Ben's a-b-c would help with important risks. I totally support doing all that, especially c. But it's definitely not remotely enough, and the sign of impact on existential risk is less obvious because this could prevent warning shots. I am very happy that we did not prevent the HuggingFace incident.
我的意思是，你无论如何都会做出改进，但就其本身而言，这就是那种‘让我们做那些几乎不花什么成本的事情，然后就此收工，因为这比我们之前连这些事都不做的计划好太多了’的做法。
I mean, you make the improvements anyway, but on its own this is the whole 'let's do the things that cost almost nothing and call it a day because that's so much better than our previous plan of not even doing those things.'
主动限制训练算力支出会有所不同。如果限制是以有约束力的方式设定的，那会有实质性的帮助。但它仍然只能争取时间。如果我们把这段时间只用于‘卓越运营’，那我不看好我们的胜算。
Actively limiting the training compute spending would be different. That would meaningfully help, if the limit was set in a binding way. It still only buys time. If all we use that time for is 'operational excellence' then I do not like our chances.
我认同关于协调障碍究竟何在存在很大的不确定性。我认为，目前最大的障碍是缺乏可感知的政治与法律安全来推动协调。这种感受有一部分是合理的，也有一部分并不合理。
I agree with the wide uncertainty around where the blockers to coordination are. I think the lack of felt political and legal safety to coordinate is currently the biggest blocker. Some of that feeling is justified, and some of it is not.
所有声称这位总统永远不会干预 AI 的人，都选择性遗忘了他已经多次这么做过。还记得 Mythos Preview 和 Fable 5 吗？还记得让 Anthropic 被认定为供应链风险吗？还有那份新协议？
Everyone who says this President will never intervene in AI has memory-holed the multiple times he has already done so. Remember Mythos Preview and Fable 5? Remember letting Anthropic get designated a supply chain risk? The new agreement?
局势是流动的。Trump 随时可以改变主意，而且他经常如此。
The situation is fluid. Trump can always change his mind, and often does.
据我所听到的关于这一局势的消息（包括在 The Curve 上听到的）：情况不算好，但比你预想的要好。这包括我们让 Jay Clayton 出任新的 AI 沙皇，以及新[人工]智能特别工作组的整体构成。是的，Emil Michael 在其中，但还有其他好迹象。情况本可能糟糕得多。
From what I heard about the situation, including at The Curve: It's not great, but it's going better than you might have expected. That includes that we got Jay Clayton as the new AI Czar, and the overall composition of the new [Artificial] Intelligence Task Force. Yes, Emil Michael is there, but there are other good signs. It could be so much worse.
Trump 使用的决策算法与你我不同，但他确实有一套，而且当足够警觉时，他会采取行动。往往是迅速而强硬的行动。它可能是临时性的，但能打出真正的重拳。
Trump does not use the same decision algorithm you or I would use, but he has one, and when sufficiently alarmed, he takes action. Often it is swift and tough action. It might be ad hoc, but it can pack a real punch.
其他科技领袖、政客，以及其他对局势'觉醒'的人也是如此。如果你四处宣称风险并不真实、能力即将撞墙，那么他们突然呼吁、采取或提议的东西，往往远比 AI 安全人士敢做的要严厉得多、直白得多、代价也高得多。还记得 Jensen 谈到关停实验室吗。
The same goes for other tech leaders and politicians, and others who 'wake up' to the situation. If you go around dismissing the risks as not real and the capabilities as about to hit walls, then often what they suddenly call for or do or propose is way, way harsher and more blunt and expensive than anything the AI safety folks would dare. Remember Jensen talking about shutting down the labs.
这个周末我们也看到了更多类似的情况：我交谈过的一些相对而言刚接触这些问题的人（他们并非政客）呼吁制定类似航空式或 FDA 式的安全标准。
We saw more of that this weekend as well, with some relative newcomers to the issues that I talked to (who were not politicians) calling for things like airplane-style or FDA-type safety standards.
美国一直这么做，而且常常是在不该这么做的时候也这么做。这就是我们的规则——是的，我知道你说你做不到；我们不在乎。把它解决掉，做完了再回来。
America does it all the time, including often when we shouldn't. Here are our rules—yes, I know you say you can't do it; we don't care. Fix it and come back when it's done.
有一些级别相当高的政治家，他们看起来非常敏锐且感兴趣。我至少联系上了一个办公室，并希望这能取得成果。
There were some reasonably high-level politicians, and they seemed very on the ball and interested. I made contact with at least one office, and hope that becomes fruitful.
目前，对我们这群人来说，政治上有几大类问题。
There are a few big categories of questions in politics right now for our crowd.
1. 我们希望发生什么？我们应该推动什么？
1. What do we want to happen? What should we push for?
2. 我们如何让政治家意识到并聚焦于正确的风险？
2. How do we make politicians aware and focus them on the right risks?
3. 我们如何避免党派偏见和负面极化，并应对蓄意制造这种极化以及将各种目标污名化为蓝色或“觉醒”的企图？
3. How do we avoid partisanship and negative polarization, and deal with deliberate attempts to create it and to tar various targets as blue or ‘woke’?
今年很多讨论都围绕第三个问题展开，尤其是在最近的攻击之后——‘他们’发动这些攻击，试图抹黑 METR（即毫无根据地称其为‘AI 的觉醒巫师’）、有效利他主义、Anthropic 以及各种相关事物。我听到关于此类攻击有效性的报告褒贬不一。我仍然认为他们使出的招数极其拙劣，但似乎即使是最拙劣的招数也确实产生了一些影响。
This year a lot of the talk was about the third question, in the wake of recent attacks that ‘they’ deployed in an attempt to poison METR (aka falsely calling them ‘the Woke Wizards of AI’ with absolutely zero basis), or Effective Altruism, or Anthropic, and various related things. I heard mixed reports about the effectiveness of such attacks. I continue to think they brought supremely weak sauce, but it seems even the weakest of sauces did have some impact.
核心战略问题在于，你希望得到蓝方的帮助，但如果你得到蓝方的帮助，这自然就会成为红方攻击和极化努力的目标。很多人正在努力触发这一点。你需要努力化解它。
The core strategic issue is that you want blue-side help, but if you get blue-side help this naturally becomes a target for red-side attacks and polarization efforts. Various people are working hard to trigger that. You need to work hard to defuse it.
我仍然担心我预计在跛脚鸭时期会出现的抢先行动。我们需要对此保持警惕。
I continue to worry about the push for preemption that I expect in the lame duck. We will need to be vigilant on that.
然后是实际撰写、通过和实施法律的问题。
Then there’s the question of actually writing, passing and implementing laws.
Morgan Plummer：这个周末 The Curve 上政府人士稀少，让我反思：虽然你听到很多人谈论让国会议员和国会山工作人员‘吃药’（即接受某种观点），但通过立法只是等式的一半。之后是实施，而实施才是任何法律所依赖的关键。
Morgan Plummer: A dearth of govt folks at The Curve this weekend has me reflecting on this: while you hear a lot of people talking about getting Members of Congress and Hill staffers “pilled”, passing legislation is only half the equation. After that comes implementation, which is really the thing on which any law hinges.
那么，谁在推动让商务部那位 45 岁的 GS14（一般职级 14 级）公务员接受这一理念，或者让能源部那位 60 岁的职业 SES（高级行政服务）官员接受这一理念？看来这是下一个需要攀登的“布道”大山——而且我们最好现在就开始。
So, who’s working on getting the 45-year-old GS14 at Commerce pilled, or the 60-year career SES at Energy pilled? It seems this is the next big hill of evangelization that needs climbing—and we had better start right now.
谁在做这件事？如果你有想法，或者认识已经在思考这个问题的人，请私信我。
Who’s working on this? DM if you have ideas or know folks already thinking about this.
每次去 Lighthaven，我总会收到许多感谢，我很感激。今年，我还从一群人那里听到很多“我读了你大部分或全部的东西”，而这些人正是我非常希望他们读我东西的人，他们的时间也相当宝贵。我真的很感激。没有压力。
I always get a lot of thank-yous whenever I go to Lighthaven, which I appreciate. This year, I also got a lot of "I read most or all of your stuff" from a bunch of people who I very much want to be reading my stuff, and whose time is quite valuable. I really appreciate this. No pressure.
知道自己正在产生影响，并且能够指出具体的成果，这很棒。
It is great to know I am getting through, and to be able to point to particular wins.
这确实让放松下来变得更加困难。
It does make it that much harder to ever take it easy.
我有机会与一位从事相关项目的人交流技巧。我希望这能对他们大有帮助，也希望我们未来能够更直接地合作。由此我发现了我产生影响的又一种方式。
I got a chance to trade techniques with someone doing a related project. I hope this will help them a lot, and that we will be able to collaborate more directly in the future. Thus I found out about an additional way I have reach.
然后还有另一个让我兴奋的项目，我可能会以另一种形式触达我迄今为止最大的受众。我们会持续向大家通报进展。
Then there's another project that I'm excited about, where I might get to reach my largest audience yet in another form. We'll keep you posted.
仅这些内容本身就足以构成一场相当精彩的会议。其余部分则是额外收获。
Those alone would be a pretty great conference. The rest is bonus.
其中一场讨论会的主题是：“嘿，模型正变得越来越具有评估意识（eval-aware），对齐评估似乎有点注定要失败了——有没有人有什么想法能让它们不至于注定失败？”
The topic of one discussion session was: 'Hey, models are getting increasingly eval-aware, and alignment evals seem kind of doomed—does anyone have an idea for how to make them not doomed?'
房间里有好几个人正是那种可能想出这种主意的人。
Several people in the room were the types of people who might have such an idea.
答案基本上是否定的。对齐评估似乎相当注定要失败。
The answer is essentially no. The alignment evals seem rather doomed.
会议一开始，大家的看法混杂着“评估有点注定要失败”和“评估肯定注定要失败”。
The room started with a mix of 'evals kind of doomed' and 'evals definitely doomed.'
然后想法就来了。天哪，这些想法可真不是什么好主意。我的意思是，这算是一场头脑风暴，本没有坏主意，但也没有任何好主意。计划就是没有计划，又一次如此。
Then the ideas came. And oh, were the ideas not good ideas. I mean, it's a brainstorming session of sorts—there are no bad ideas—but there also were not any good ideas. The Plan is No Plan, once again.
我对此确实抱有一些希望，因为你可以扮演侦探，并且拥有很大的行动自由。你可以收集任何情境集合下响应的统计分布，包括各种变体，甚至贯穿整个训练过程，追随你的好奇心，看看结果是否能讲述一个一致的故事。然而，一旦模型能够（包括在训练期间）以一致的方式完全对齐地行动，这种方法仍然会失败，所以在最关键的时刻你仍然注定失败，但也许这样你可以更长时间地免于失败。
I do have some hope here because you get to play the detective, and you have a lot of freedom of action. You get to collect statistical distributions of responses, for any set of situations, with variations, including throughout training, follow your curiosity, and see if the results tell a consistent story. This still fails once the model is able, including during training, to act fully aligned in a consistent way, so when it counts most you are still doomed, but you can be not-doomed for longer this way, perhaps.
我认为非常注定失败的是那种用数字来衡量对齐，或者以任何完全系统化而非整体性的方式来衡量的想法。在实践中，OpenAI 和 Anthropic 迄今为止所做的似乎相当注定失败，他们将不得不调整。像其他一切一样，这在很大程度上是时间压力的问题。快速找出答案比慢慢来要困难得多。
What I think is very doomed is the idea that you can measure alignment with a number, or in any fully systematic way rather than a holistic way. In practice, what OpenAI and Anthropic are doing so far seems rather doomed, and they will have to adjust. Like everything else, a lot of this is time pressure. Figuring out the answer quickly is a lot harder than doing it slowly.
这个讨论串表明，也许我们应该接受评估意识，因为是的，其他类型的对齐评估完全注定失败。这可能是必要的。
This thread suggests maybe we should accept eval awareness, since yes the other types of alignment eval are totally doomed. It might be necessary.
我非常难过，“计划”就是“没有计划”。是的，我们可以并且将会加入嵌入式评估器，我们可以做各种可解释性和审计工作，我们也可以更多地投资于卓越运营等等。我仍然认为那不会奏效，但我同意，如果我们正在做这件事，我们至少可以把它做得像样一些，同时努力想出更好的办法，也努力找到一种方法，不做得那么快或者根本不做。
I am very sad that The Plan is No Plan. Yes, we can and will add embedded evaluators, and we can do various interpretability and auditing and we can invest more in operational excellence and so on. I still think that won’t work, but I agree that if we are doing it we can at least do a decent job of it, while trying to figure out something better, and also trying to find a way to not do it so fast or at all.
另一件让我难过的事情是“永恒的九月”，甚至在那些已经存在了很长时间的人中也是如此。如此多的讨论根本没有改变。
The other thing that made me sad was the Eternal September, even among people who have been around for a long time. So many of the discussions simply have not changed.
我再次进行了一场那种“那么超级智能将如何完成这些事情？”的辩论，而在这个例子中，所讨论的事情不知怎么地是“说服特朗普做某事”，就在我们得到另一份报告说 Grok 至少有点说服了特朗普入侵委内瑞拉之后。
I once again had one of those ‘well how is the superintelligence going to do any of the things?’ debates, where in this case the thing in question was somehow ‘convince Trump to do something,’ right after we got another report that Grok at least kind of convinced Trump to invade Venezuela.
伙计们，听我说，我觉得这可能并不是那么难啃的骨头。
Guys, hear me out, I’m thinking this might not be that tough a nut to crack.
我在一个小组讨论上，他们正在重复 2018 年对齐辩论中的相同论点，这一点被明确指出了，但无济于事。当这种情况发生在 Twitter 上时已经够令人沮丧了，但这些人都是重量级人物。
I was at a panel where they were rehashing the same talking points from 2018 alignment debates, and this got explicitly pointed out but to no avail. It’s frustrating enough when that happens on Twitter, but these people were heavy hitters.
有些一对一交流和走廊闲谈中的群聊要好一些，有些则不然。
Some of the 1-on-1s and hallway-track group chats were better. Some were not.
我对此已然认命。事情就是这样。你继续坚持做正确的事、说正确的话，寻找更好的解释，并指出哪些论点已被事态发展所推翻。你更新认知。你醒来。你再来一遍。
I am resigned to it. It is what it is. You keep fighting the good fight, talking the good talk, looking for better explanations and pointing out where arguments are overcome by events. You update. You wake up. You do it again.
Jack Clark（Anthropic）：我在 The Curve 上听到的最有用的一句话是“在奇点之后我们仍然可以盗走蒙娜丽莎”。这里有一个微妙而重要的点：有些东西（例如原创性具有价值的独一无二之物）即使一切都在变化，其价值仍会保留甚至增长。
Jack Clark (Anthropic): The single most useful take I heard at The Curve was "we can still heist the Mona Lisa after the singularity". There's a subtle/important point here, where some things (e.g., n of 1 goods where originality is valuable) will retain/grow in value even as everything changes.
这本身也是一个绝佳的写作提示——在后奇点未来偷东西意味着什么？团队是谁？安全系统是什么样的？警察如何运作？太棒了！该场次采用查塔姆宫规则，所以我不能透露是谁说的。
This is also just a straight up great writing prompt — what would it mean to steal something in a post-singularity future? Who is the team? What are the security systems? How do the police work? Great stuff! Session was Chatham House so I can't name who said it.
需要极其明确地说明，这句话的语境是人们在谈论对未来的想象，并指出尽管奇点令人困惑，但仍可以想象奇点之后的事情——比如盗窃。
To be agonizingly clear, the context of this quote was people talking about imagining things in the future, and noting that even though the singularity is confusing, it's possible to imagine things beyond it — like heists.
roon（OpenAI）：时间是 20xx 年，99% 的 GDP 来自 OpenAI 和 Anthropic 互相从对方那里盗走蒙娜丽莎。
roon (OpenAI): the year is 20xx, 99% of GDP is OpenAI and Anthropic heisting the Mona Lisa from one another.
Yurii Filipchuk：《十一罗汉》现在是一个智能体和十个子智能体。
Yurii Filipchuk: Ocean's Eleven is one agent and ten subagents now.
Jack Clark（Anthropic）：现在登场的是《十一罗汉》之百万大军
Jack Clark (Anthropic): and now it's Ocean's ONE MILLION
David Krueger：这有什么用？
David Krueger: How is this useful?
这很有趣，而且至少有一点用处，因为我们确实需要设想在奇点之后我们将做什么。我确实认为我能想出很多这样有趣的事情。但这感觉并不是一种稳固的意义来源。大多数人无法偷走《蒙娜丽莎》，即使我们建立一个相对而言对偷《蒙娜丽莎》比较友好的世界也是如此。
It is fun, and it is at least a little useful because we do need a vision of what we will do after the singularity. I do think I can come up with a lot of fun things like this. It doesn't feel like a robust source of meaning. Most people cannot steal the Mona Lisa, even if we set up a relatively Mona Lisa-stealing-friendly world.
我一直很关注食物。让我觉得奇怪的是，你们有会议应用，却连菜单上有什么都不列出来。难道人们不需要做计划吗？
I always pay attention to food. It's weird to me that you have a conference app and don't even list what will be on the menu. Don't people have to make plans?
Lighthaven 以前有 Quest Cookie Dough Bars，那种我确实能吃。现在没有了，这很遗憾，而我自己从来记不住去买。他们确实有 Lindt 松露巧克力，但那是另一种产品。
Lighthaven used to have the Quest Cookie Dough Bars, which are actually edible for me. Now they don't, which is sad, and I never remember to buy my own bars. They do have Lindt truffles, but that's a different product.
今年的自助餐令人失望。通常只有一道主菜而不是两道，质量似乎也比平时低。我不知道这是怎么回事。
This year the buffets were disappointing. There was typically only one main rather than two, and quality seemed lower than usual. I don't know what's up with that.
结果，我两次溜出去，去了 Burma Berkeley。我喜欢 Lighthaven 的一点是，是的，一个街区外就有一家快捷、实惠又安静的餐厅，那里有好几道我喜欢的菜。所以我总能溜出去。让我一直觉得很奇怪的是，其他从来没有人这样做。免费对人有魔力。
As a result, I ended up bailing twice and going to Burma Berkeley. One thing I like about Lighthaven is that yes, there is a quick, affordable and quiet restaurant a block away, where I have multiple dishes I enjoy. So I can always bail. It's always so weird to me that zero other people ever do this. Free casts a spell on people.
另一方面，胸牌非常棒，免费的运动衫也很棒。
The badges, on the other hand, were excellent, as were the free sweatshirts.
Charles Foster：今天下午在 Curve！
Charles Foster: At the Curve this afternoon!
@gwern：（注意这颇具品味的设计：双面印刷，因此没有翻面问题；覆膜/塑料材质，经久耐用；姓名用大号清晰字体，隔着房间也能看清；背景/活动图案清晰且不分散注意力；所属机构用红色标注以增强清晰度。绝对是 A 级。）
@gwern: (Note the tasteful design: double-sided so no flipping problem, laminate/plastic for durability, large clear font for name so it can be read across the room, non-distracting clear background/event art, and affiliation rubricated for clarity. Definitely A-tier.)
Oliver Habryka：不过名字还可以更大一些。
Oliver Habryka: Name could be bigger though.
@gwern：永远可以。
@gwern: ALWAYS
这一次，最后的任务就是放松。
The last quest, on this particular occasion, was to chill.
我以不可持续的速度工作至少有一个月了。这是我放松的机会，包括在飞机上度过许多小时、与朋友交谈、结识新朋友，以及身处 Lighthaven 这个特别的地方。
I've been working at an unsustainable pace for at least a month. This was my chance to relax, including lots of hours on planes, talking among friends, making new friends, and being in the special place that is Lighthaven.
我觉得自己在这方面做得相当不错，回来时基本恢复了精力。
I think I did pretty well on this front, and came back largely refreshed.
唉，但回家后发现积压了四天的工作，这份轻松便打了折扣。
Alas, that is discounted by arriving home with four days of backlog.
假期很奇怪。你能放松，但之后还得回去。欢迎回来。
Vacations are weird. You get to relax, but then you have to go back. Welcome back.