视频库 / NO.005
--:--
点击播放，这里会跟随视频显示当前句的中英字幕。
第 5 期 · 回应 Ⅳ·06「伟大的事业从哪来？」
Sam Altman: "Never a Better Time to Do a Startup"
节目发布 2026-07-27 · Y Combinator章节 · 点击跳转视频
0:07
2005：YC 首批与 PG 的飞行教练式指导
4:44
创业公司为何是经济的解毒剂
7:45
反驳「创业已死」：黄金时代刚开始
9:29
创业公司何时胜出：经验贬值、工具增值
12:07
OpenAI 起点：被当白痴是一份礼物
15:18
找到同类：异端者的部落与湾区网络
17:53
帮很多人一点小忙：Brockman 的故事
19:43
真诚 vs 讽刺：推特黑子腐蚀灵魂
23:48
Hugging Face 事件：失控不再只是理论
26:50
权力分散：创业者对抗集中的角色
30:39
愿景清晰、起步模糊：六个月抵两年
33:10
算力与 token：需求没有上限
36:04
十年后的好未来与给自己的信
本期讲者
萨姆·奥特曼OpenAI 联合创始人兼 CEO，主导 ChatGPT 的推出。2005 年从斯坦福辍学创办 Loopt 并进入 YC 首批，2014–2019 年任 Y Combinator 总裁。
Gary TanY Combinator 总裁兼 CEO（2023 年起），Posterous 联合创始人、Initialized Capital 创始合伙人，早期投资过 Coinbase、Instacart 等。
012005：YC 首批与 PG 的飞行教练式指导
0:07
Sam, thanks for joining us. >> This is something >> this is like a lot bigger than the earlier startup schools. >> Yeah, I mean startups are a lot bigger now than they have ever been. So >> sure. [cheering] [applause] >> I mean they've bigger they're bigger than they've ever been because of uh the vision that you had for AGI which is nigh. I think this is going to be the best time in the world to do a startup and uh it's gonna be quite amazing to see. >> So I wanted to start with um you know a time and place uh which is you were in the very first batch of Y Combinator in 2005 with um location sharing company called Looped. Um what do you remember about that and what was your best PG Paul Graham story? Um, I think if it were possible to get as like far from this moment as like I can imagine, it was, you know, startups are not cool at all. We were like hiding out in this little building in Cambridge. PG was making us dinner. Um, and you know what took three months to build at the time that
Sam，感谢你能来。这个 —— 这次的规模比之前几届创业学校（Startup School）大多了。是啊，我是说，创业公司现在的规模也比以往任何时候都大。所以 —— 当然。[欢呼] [掌声] —— 我是说，它们比以往任何时候都更大，比以往任何时候都更大，是因为你对 AGI 的那个愿景，而 AGI 已经近在眼前。我觉得这会是现在是全世界创业最好的时候，接下来的发展一定会非常惊人。我想从一个时间和地点开始聊——2005 年，你在 Y Combinator 的第一批学员里，做的是一家叫 Looped 的位置共享公司。关于那段经历你还记得什么？还有你最喜欢的 PG（Paul Graham）故事是什么？我觉得，如果要说有什么能离此时此刻最远的事，那大概就是那段日子了。就是那种，创业一点都不酷。我们当时就窝在剑桥的一栋小楼里躲着，PG 给我们做饭。嗯，你知道吗，当年每家公司在整个 YC 期间花三个月才做出来的东西，
1:17
we each company built over the whole YC startup could now be done with like in like seven minutes by a coding agent. And it was only I mean only 20 20 years ago. Uh and I think the the difference in what's possible for a startup now uh what a startup can take on you know it's pos I think you could either be sad and be like oh man like a codeex prompt is a whole startup or you could be like I can go start the world's most ambitious crazy company I can have like experts in every field working together I can do these very hard technological things that were just impossible um and it's going to be amazing but at the time it felt nothing like that and it was very difficult to get anything to I think Paul Graham is probably the most important force in startups of the last few decades and >> without question. Yes.
现在用一个编程 agent 大概七分钟就能搞定。而且那才只不过是——我是说才 20 年前。呃，我觉得现在创业公司能做到的事情已经完全不同了，一家创业公司能挑战的东西，你知道，你可以选择难过，觉得「天哪，一句 codex 提示词就抵得上一整家创业公司」，也可以选择说，我可以去创办这个世界上最有野心、最疯狂的公司，我可以让各个领域的专家一起共事，我可以去做那些以前根本不可能做到的、非常难的技术活儿，而且这……会变得非常了不起，但在当时完全没有那种感觉，做成任何事都非常困难。我觉得 Paul Graham 大概是过去几十年里创业圈最重要的推动力量，而且 >> 毫无疑问。是的。
2:05
>> No question. Um [applause] and and my kind of Paul Graham memory is we would all every week come I think it was on Tuesdays and he would like cook he himself would cook us dinner and we would all walk in feeling like very hopeless and very dejected and we would there were eight companies and then we would all go home at the end of it and he had convinced like each of us that our startups were about to like take over the world and we're going to be like you the next Google or whatever Facebook maybe at the time. And that ability to sort of like create optimism and momentum and belief out of nothing was like a real PG special. And in the early days when YC seemed like a terrible idea and startups seemed like a bad idea and certainly startups that were like young technical founders with no business people seemed like a terrible idea, PG just sort of like willed it into existence.
>> 毫无疑问。嗯 [掌声] 我对 Paul Graham 的记忆是，我们每周都会——我记得是周二——聚在一起，他会亲自下厨他自己给我们做晚饭，我们走进去的时候个个都觉得很绝望、很沮丧，然后我们当时一共有八家公司，等到结束时我们各自回家，而他已经说服了我们每一个人，让我们相信自己的创业公司马上就要征服世界了，我们会成为下一个 Google 或者别的什么，也许当时是 Facebook。而那种凭空创造出乐观、动力和信念的能力，真的是 PG 的独门绝技。在早期，当 YC 看起来像个糟糕的主意、创业看起来像个坏主意，尤其是那种由年轻技术型创始人组成、没有商业背景的人的创业公司看起来糟糕透顶的时候，PG 就那样硬生生把它变成了现实。
3:02
But it also takes someone who actually I mean you're sort of famously um par excalance uh an agentic person like even before we thought about agents period. I mean you just >> uh I remember Paul wrote um I mean it's making the rounds on X today even that uh Paul wrote in the early in the late 2000s that you would be you were one of the top five entrepreneurs he'd ever met. >> That's very nice of me to say. Um, I I think like many 20-year-olds, I had like a lot of energy and a lot of ambition, but it was sort of very not directed and I wasn't sure what to do. Paul had this thing he used to say of like uh the the job of being a good startup investor is a teacher, but it's a kind of teacher we don't usually use. Like we usually think of teacher as someone who like gives a lecture and you can't help someone that much by like giving a lecture about a startup. There's another kind of teacher which is like a flight instructor. the person who like sits next to you saying like do this, don't do this, that work, this is, you know,
但这也需要一个真正……我是说，你算是出了名的、堪称典范的那种极具能动性（agentic）的人，甚至在我们开始谈论 agent 这个概念之前就是了。我是说你就是 >> 呃，我记得 Paul 写过——今天在 X 上都还在传——呃 Paul 在两千年代早期、其实是后期写道，你是他见过的前五名创业者之一。认识的。>> 他这么说太抬举我了。嗯，我想我跟很多二十岁的人一样，精力特别旺盛，野心也很大，但方向感很不明确，我也不确定该做什么。Paul 以前常说一句话，大意是，做一个好的创业投资人其实是一种老师，但不是我们通常说的那种老师。我们一般想到老师，就是那种站在台上讲课的人，可是光给你讲一堂关于创业的课，帮不了你多少。还有另一种老师，就像飞行教练。就是坐在你旁边跟你说：这样做、别那样做、那个有用、这个是……你懂的，
4:02
you missed that thing or you're making these mistakes and I'm just going to talk to you about this one and that very like hands-on kind of uh you know, here's where to like direct this sort of like brownie in motion energy. That was very important to me. >> Yeah. I I guess later you uh came on to become president of YC and um you you sort of brought exactly the same energy to uh a great many YC founders over the years. Um did you have anything like did anything jump out at you from that time around you know taking this raw energy of someone really really smart and maybe a little undirected and then driving them more towards agency.
你漏掉了那件事，或者你正在犯这些错误，我就专门跟你聊这一件事，那种非常亲力亲为的方式，就是告诉你该把这股像布朗运动一样乱窜的能量往哪儿引导。这对我来说非常重要。>> 是的。我猜后来你出任了 YC 的总裁，你也把完全一样的这股能量带给了这些年来非常多的 YC 创始人。那段时间里，有没有什么让你印象特别深刻的事，关于怎么把一个非常非常聪明、可能还有点没方向的人身上那股原始能量，引导他们更有主动性（agency）。
02创业公司为何是经济的解毒剂
4:44
>> Yeah. I so first of all I love startups. I think not everyone they're not like for everyone but I think startups are the coolest thing in all of business. Um I think startups are really like the main thing that keeps the economy from becoming stagnant. Um I think companies do just like drift towards suckiness and startups will continue to be important forever. In fact, if if we are right that AI is going to be such a big change, startups will be much more important to making sure that the power of this technology gets widely distributed throughout the economy and society and is not just concentrated in a few companies or models. So I I think startups are this like unbelievably cool, fun, extremely painful and difficult but wonderful thing. And I think a big part of job the job of running YC is you are kind of the like unofficial flag bearer for the startup movement. Um you know like there there's lots of startups. There's lots of ways to do a startup. You obviously don't need to do YC but it has always
>> 是的。首先，我很喜欢创业公司。我觉得创业并不适合所有人，但我觉得创业公司是整个商业世界里最酷的东西。我认为创业公司真的是让经济不至于停滞的主要力量。我觉得大公司就是会慢慢滑向平庸和糟糕，所以创业公司会永远重要下去。事实上，如果我们的判断没错，AI 会带来这么大的变革，那创业公司就更加重要——它们能确保这项技术的力量被广泛分配到整个经济和社会中，而不是只集中在少数几家公司或模型手里。所以我觉得创业公司是一件无比酷、无比有趣、极其痛苦艰难但又非常美好的事。而且我认为，管理 YC 这份工作很大一部分，就是你要当创业运动的非官方旗手。你知道，创业公司很多，做创业的路子也很多。你显然不是非做 YC 不可，但它一直对公司帮助巨大，
5:50
been a huge help to companies and a very powerful force. And so starting with PG uh and then all of us like you know we have to like make YC successful but I think we really have to like fight for why startups are important and why people should consider startups and help put like relatively more power in the hands of founders and encourage more people to start them. And that's actually worked really well. I think thinking back to the early days of YC there's like very little leverage in being a founder relative to an investor and that shift uh you know I think that's even been good for the investors. I think it's just been like much better for the whole startup ecosystem. So mostly what I tried to do at YC was just push for more startups and encourage people to think about startups and figure out what we could do to help founders and help make startups and the startup ecosystem is as good as possible. Definitely part of that was pushing people towards more ambition and bigger swings. Um, but it feels like the
是一股非常强大的力量。所以从 PG 开始，然后我们所有人……我们当然得把 YC 做成功，但我觉得我们真正要做的，是为“创业为什么重要”“人们为什么该考虑创业”而奋战，并帮助把相对更多的权力放到创始人手里，鼓励更多人去创业。而这件事其实效果非常好。我回想 YC 早期，创始人相对于投资人几乎没有什么筹码，而那种转变——我觉得对投资人其实也是好事。我觉得这对整个创业生态都好得多。所以我在 YC 主要想做的，就是推动更多创业、鼓励人们去考虑创业，并琢磨我们能做什么来帮助创始人，让创业和创业生态尽可能地好。这里面肯定有一部分是推动大家更有野心、去挥更大的棒。不过跟现在比，那感觉像是小联盟。我是说，我
6:53
minor leagues relative to now. I mean, I think you were one of the people who really brought hard tech to uh YC in a really grand way. And then one of the things we're seeing at YC now is that number was, you know, maybe five or 10% for many years and then now we're getting to 15 20 25%. Uh I would argue on the back of how much easier it is uh to use agents and cost coming down. >> I I think like ambitious startups are always awesome and hard tech startups are appealing to a lot of people but they're hard or they've been hard. They still will be hard. But what you can do now to go take on a really ambitious project like in the same way that you can make the startup that took us 3 months of non-stop work in 17 seconds with three months of work and a lot of agents you can do unbelievable things.
觉得你是真正把硬科技（hard tech）大规模带进 YC 的人之一。而我们现在在 YC 看到的一件事是，这个比例多年来大概是 5% 或 10%，而现在已经到了 15%、20%、25%。我认为背后的原因是，用 agent 变得容易多了，成本也在下降。>> 我觉得有野心的创业公司永远很棒，硬科技创业对很多人也很有吸引力，但它们很难，或者说一直很难。以后也还是会难。但你现在能做的，是去挑战一个真正有野心的项目——就像当年花我们三个月不停工作才做出来的创业公司，现在 17 秒就能做出来一样，有三个月的时间加上一大堆 agent，你能做出难以置信的东西。
03反驳「创业已死」：黄金时代刚开始
7:45
So I think we will see a golden age of startups where people are like you know what I'm going to do things that would have been completely impossible for a startup to even like dream at a year ago in in sort of like the YC time frame. I mean that's actually a really big reframe. I mean there are probably people even in this room who might be worried that well uh intelligence on tap means that you know they might be looking at a loop or my startup posterous and they're like well I can't start that company anymore but guys that was like a long time ago actually man there's this whole meme going around which is like you know you have to join a frontier lab or you're going to be a member of the permanent underclass because it's so dumb. um be because there's going to be like, you know, startups are over and there's like no economic value. I that's not true. I mean, if that were true, the world is like totally and it's a a very bad place. But it's I I would bet that the average startups created today will be like I bet there's
所以我觉得我们会迎来一个创业的黄金时代，人们会说：我要去做那些在一年前、按 YC 这种时间尺度看，一家创业公司连想都不敢想的事情。这其实是个很大的观念转变。我想这个房间里可能就有人在担心：智能唾手可得，是不是意味着他们再也没法创办像 Loopt 或者我的 Posterous 那样的公司了。但各位，那其实已经是很久以前的事了。现在网上流传着一种说法：你要么加入前沿实验室，要么就会成为永久底层阶级的一员——这太蠢了。因为按这种说法，创业已经结束了，不再有什么经济价值。我觉得这不是真的。如果真是那样，这个世界就彻底完了，会是个很糟糕的地方。但我敢打赌，今天创办的创业公司平均而言——我敢说
8:50
some future trillionaire sitting in this room and I would bet that the startups created today will be much more valuable, much more impactful than startups of the past. And I kind of think people see that a little bit more now. I mean, there was this period where like are is AI just going to break the whole economy and I I think people are like mostly over the sort of shock response of that. Um but there was like yeah definitely a time I would you know come every YC batch and I I'll be interested to see how the kind of like level of anxiety versus ambition was trending and it feel like it went through this like big trough where people are like you know it's over it's models are just going to eat everything to now it's back towards like let's go do it.
这个房间里就坐着未来的万亿富翁——我敢打赌今天创办的公司会比过去的创业公司更有价值、更有影响力。我觉得现在大家也稍微更能看到这一点了。之前有一阵子，大家都在问：AI 是不是要把整个经济搞崩？我觉得现在人们基本已经过了那种震惊反应期。不过确实有那么一段时间，我每次去参加 YC 的批次时，都会很好奇焦虑和野心的比例在往哪边走，感觉它经历了一个很深的低谷，大家会说：完了，模型要把一切都吃掉。而现在又回到了“咱们上，去干吧”的状态。
04创业公司何时胜出：经验贬值、工具增值
9:29
>> What are some practical things here? I mean heartche for instance I was hearing uh you know at one of the breaks someone was asking like should I go get my PhD? How important are credentials? Um, you know, how would you answer that? Especially people who want to do these harder tech things. I Well, as a general observation first, I think startups tend to win when um the sort of like technology landscape is moving very quickly. uh when costs are coming down when cycle times are short and all of those things are happening right now. So if if you look at when there have been like the great clusters of startups in the past, you know, there was the internet boom in whatever that was like 998 99 when new things became possible. Um there was another version of a mini version of people building on top of basically like Facebook apps.
>> 这里有哪些具体可行的建议？比如硬科技，我在某次茶歇时听到有人在问：我该不该去读个博士？文凭有多重要？你会怎么回答这个问题？尤其是那些想做更硬核技术的人。嗯，先说一个总体观察：我觉得创业公司往往会赢，是在技术格局变化非常快的时候，在成本下降、周期变短的时候，而这些事现在都在发生。所以如果你回看历史上那些创业公司扎堆出现的时期，比如互联网热潮，大概是 98、99 年，那时很多新东西成为可能。还有一个小一点的版本，是人们在 Facebook 应用平台上做东西。
10:24
There was then another big version when the iPhone app store launched. Um but the great startups tend to cluster when the ecosystem shifts and incumbents lose a lot of their advantage. And then also when you have this like cost and cycle time change. Uh I this moment feels very big for those things. It also has this other thing that you were talking about which is a lot of the traditional things that were hard to get expertise you know the ability to go like hire excellent people that were that could do specific things you needed. That's really shifted and in the last few months um I have seen a lot of people who just kind of grew up using AI the last few years who are like I can kind of automate an entire startup of agents and four of us four people and you know all this compute and I think we're going to see much more of that I clearly like taste and agency and understanding of kind of like the physics of business like where you can build up a valuable business how to think what about like a good network
然后 iPhone 应用商店上线时又是一个大版本。但伟大的创业公司往往扎堆出现在生态发生转变、在位者优势大幅丧失的时候，再加上成本和周期时间的变化。而此刻在这些方面感觉都非常大。它还有你刚才提到的另一点，就是很多传统上很难获得的专业能力——比如去雇到能做你需要的特定事情的优秀人才——这一点真的变了。最近几个月，我看到很多这几年基本是用着 AI 长大的人，他们说：我可以用 agent 把一整家公司自动化，就我们四个人加上这么多算力。我觉得我们会看到更多这样的情况。当然，品味、主动性，以及对商业物理规律的理解——比如在哪里能建立起有价值的生意、怎么判断什么是好的网络
11:30
effect or sort of a good moat looks like versus a fake one. Um, but I would I would bet that this generally will cut against many years of experience in favor of people who have a lot of fluency with the tools. I guess I want to get into the beginning of OpenAI and that it actually started as YC research looking at this idea. I mean, kind of a crazy idea that, you know, you and um a ragtag crew decided we're going to dedicate our lives to this um creation of AGI, but at you know, today it it sounds fatal fatal complete, right?
效应、什么是真正的护城河而不是假的——这些依然重要。但我敢打赌，这总体上会削弱多年经验的价值，而更利好那些对这些工具极其熟练的人。我想聊聊 OpenAI 的开端，它其实是作为 YC Research 探索的一个想法起步的。那在当时算是个挺疯狂的想法：你和一群杂牌军决定要把一生投入到创造 AGI 这件事上。不过在今天听起来，这好像是理所当然、水到渠成的，对吧？
05OpenAI 起点：被当白痴是一份礼物
12:07
>> Yeah. >> But that really wasn't how it felt when you started. >> It's really hard because it now it like is the only thing people want to talk about. It's really hard to remember what it was like. It's even hard for me to remember this without like finding notes from the time 10 years ago when everyone was like, not only are they wrong about saying they think it might be possible to build AGI, but they're going to they're single-handedly going to cause another AI winter. It's they're wrong and irresponsible and bad. Um it and and maybe like my highest order or like my highest bit of advice to all of you um is find the things that you can develop like reasonable conviction in that people decide like the conventional wisdom is they're just wrong. And be okay with it taking a long time and having people be like very you know, very frustratingly wrong and dismissive. For years at OpenAI, it felt like we knew the biggest secret in the world. Everybody was calling us an idiot. Uh, and we had increasing data
>> 是的。>> 但你们刚开始时完全不是这种感觉。>> 这很难讲，因为现在这已经是所有人唯一想聊的话题了。很难回忆起当时到底是什么感觉。连我自己都得翻出十年前的笔记才想得起来——那时候所有人都说：他们不但错在认为造出 AGI 有可能，而且他们还会凭一己之力再引发一次 AI 寒冬。说他们既错误又不负责任，很糟糕。而我给各位最重要的一条建议，可能就是：去找到那些你能建立起合理信念的事情，而主流观点却认定它们纯属错误。并且要接受这件事会花很长时间，接受人们会用一种让人非常沮丧的方式误判和轻视你。在 OpenAI 的那些年，感觉就像我们掌握着世界上最大的秘密，而所有人都在说我们是白痴。而且我们不断拿到新的数据
13:20
points to convince ourselves we weren't delusional. And it was very frustrating. It was extremely frustrating. Um, but it looking back it was like an incredible gift because it meant we didn't have this like massive competitors and we had time to do our research and build our stuff and build our company. Um, and I've since noticed that this is the case for like a a lot of companies in different ways. They they end up doing something that other people that like experts in the field or the industry are convinced is a very bad idea. Um, and on the other side of this, if you're starting the same startup as everybody else, it's like you get a lot of hype and you can raise a lot of money, but it's it's sort of like those are much less frequently the big option, the big outcomes. So, if I were you all, I would figure out what the really big brand new thing is that is possible now that wasn't possible a few years ago. In our case, it was no AI had really been working and then the deep learning magic
点来说服自己：我们不是在妄想。那真的非常令人沮丧，极度沮丧。但现在回头看，那其实是一份难以置信的礼物，因为这意味着我们没有大批竞争对手，我们有时间做研究、做我们的东西、把公司建起来。后来我发现，很多公司都以不同的方式经历过类似的事。他们最后做的，往往是那些别人——尤其是这个领域或行业的专家——都确信是非常糟糕的主意的事。反过来说，如果你做的创业跟其他所有人都一样，那你可能会得到很多关注、也能融到很多钱，但那种情况反而很少成为真正的大机会、大结果。所以如果我是你们，我会去想清楚：现在有什么真正巨大的、全新的、几年前还不可能的事情。对我们来说，就是当时 AI 一直都不太行得通，然后深度学习的魔法
14:20
moment happened and the world didn't update enough. the world. This is another great PGism. The world does not understand how to intuitit exponentials and so they missed this one. There's got to be new exponentials forming right now. I don't know what they are. But if you can figure those out and if you can develop continuing conviction based on more data, be grateful that the world doesn't understand. They will eventually. This is like a huge superpower. It sounds like in your journey there were like a few different things that stacked. Maybe the first one was even uh that AGI was possible to be built and finding other people who believed that thing. You know, it didn't matter that lots of people didn't think it was possible. It it mattered that you found really really smart people who did believe that. Yeah. We used to joke that only um 50 people in the world believe that AGI was possible, but it was okay because 45 of them worked at OpenAI. Um and I think that's kind of true. Like you don't need a ton of people. Uh, and
时刻发生了，而世界的认知更新得不够。这又是 PG 的一个名言：世界不懂得怎么凭直觉理解指数增长，所以他们错过了这一次。现在肯定也有新的指数曲线正在形成。我不知道它们是什么，但如果你能找出来，并且能基于更多数据不断强化自己的信念，那就该庆幸世界还不理解。他们最终会理解的。这算是一种巨大的超能力。听起来在你的历程里，有几件事是层层叠加的。也许第一件就是：AGI 是有可能造出来的，并且找到其他相信这件事的人。很多人不相信它可能，这并不重要，重要的是你找到了真正非常聪明、并且相信这件事的人。是的。我们以前开玩笑说，全世界只有 50 个人相信AGI 是可能的，但没关系，因为其中 45 个都在 OpenAI 工作。我觉得这多少是真的。你并不需要很多人。而且事实上，正因为说的是异端的话，那些人才特别想
06找到同类：异端者的部落与湾区网络
15:18
in fact saying the heretical thing was probably why those people wanted to like all be together. So, so again, I think like really figuring out what you believe, being willing to be misunderstood for a long time and bringing together the like crew of misfits that believes it. That's also what YC was like in the early days. One of the things that people in the room are probably thinking is like, well, I believe this thing, but I haven't found my people yet. Like, is that delusion? You know what? Is that actually maybe even a gate that you would propose people have? It's like, well, you need to find, you know, five people who might believe that or even one like a co-founder, you know, people maybe that's one of the explanations for why we like co-founders at YC so much.
聚在一起。所以我还是那句话：真正想清楚你相信什么，愿意长期被误解，然后把一群相信它的“怪人”聚到一起。YC 早期也是这样的。这个房间里的人大概会想的一件事是：我相信这件事，但我还没找到我的同类。那这算不算是妄想？你会不会把这当成一道门槛？比如说，你得找到五个可能相信这件事的人，哪怕只有一个联合创始人——也许这就是我们在 YC 这么看重联合创始人的原因之一。
16:00
>> Yeah, I think if you can't find anybody else that shares your belief, you should pay attention to that. Um, and it's also very hard to do a startup on your own and very lonely. But I don't think you need to find a lot of people. And in fact, if everybody believes it, that's also like a bad a bad sign. Um, the question, I don't know if this is still what feels like the limiter on more startups. But five or 10 years ago, what felt like was limiting the number of good startups YC could fund is figuring out how to solve the co-founder matching problem. Like a lot of really talented people, and they just couldn't find their tribe. They couldn't find their people. Um, you know, it used to be very heretical that YC told people they had to move to San Francisco.
>> 是的。我觉得如果你找不到任何一个跟你有同样信念的人，那你应该重视这个信号。而且一个人做创业非常难，也非常孤独。但我不觉得你需要找到很多人。事实上，如果所有人都相信，那也是个不好的信号。问题是，我不确定这是不是现在仍然限制创业数量的瓶颈。但五年或十年前，感觉限制 YC 能资助多少好公司的，是怎么解决联合创始人匹配的问题。有很多非常有才华的人，就是找不到自己的部落，找不到自己的同类。你知道，当年 YC 让人搬到旧金山，是件非常离经叛道的事。
16:47
And looking back, it was clearly right. I think I don't know if this is still going to be true for the next 10 years. I suspect it will be, but the best thing you could do if you wanted to like find your people to do a startup with was to move to the Bay Area. You just like magic happened. You got a lot of like lucky chances and collisions. Um, I still think that's probably good advice, although I feel like I have less of an intuition for it now. Um, I also think that I I can say this, Gary probably can't. I I I think the premium on doing YC is bigger now than it's like ever been before. Uh, the distance between like YC and second place is has expanded. And this is a reminder of the power of network effects. And so finding networks that you can be part of uh that help you just meet those people like the people that I started OpenAI with, I met like many many years in some cases before uh starting OpenAI and I kind of like got to know them over the long like a very long career journey and this happens again and again and so
现在回头看，这显然是对的。我不知道未来十年这是不是还成立。我猜还会成立。如果你想找到一起创业的人，你能做的最好的事就是搬到湾区。就是会发生魔法般的事情，你会遇到很多幸运的机会和碰撞。我现在依然觉得这大概是好建议，不过我感觉自己对这件事的直觉不如以前强了。我还想说一点，这话我能说，Gary 大概不太方便说：我觉得做 YC 的溢价，现在比以往任何时候都要大。YC 和第二名之间的差距被拉大了。这也提醒我们网络效应的威力。所以去找到你能加入的网络，让你能认识到那样的人——比如跟我一起创办 OpenAI 的那些人，有些我在创办 OpenAI 之前很多很多年就认识了，在很长的职业生涯里慢慢熟悉起来。这种事一再发生，所以你越早把自己置身其中越好
07帮很多人一点小忙：Brockman 的故事
17:53
the sooner you can put yourself just in a flow where you're going to meet the people that will be your eventual co-founders for startup or the next one or whatever, I I think the better. It takes a long time to compound. >> I I guess I'm keying off what you just said, which is like, you know, some of the people you ended up starting, you know, all these different things with, you didn't necessarily know uh what that was for or that it would be useful in a, you know, sort of network setting at all. Like you just collected cool, interesting people. I think you you do this. I think Peter Keel does this really really well. What would you say to a room full of people who are just starting out and like figuring out who their people are? Ju >> just like your highest confidence piece of advice here is just like find a way to be like mildly helpful to a lot of people. It's like fun to do. You'll see a lot of interesting stuff. Um it's like kind of gratifying to be helpful. But you know, even that example you were
一个能让你遇到未来联合创始人的流程——不管是这次创业还是下一次或者别的什么，我觉得都更好。这种积累需要很长时间才能见效。>> 我大概是接着你刚才说的那点讲，就是说，你后来一起做各种事情的那些人里，有些人你当初并不知道那是为了什么，也不知道这在某种人脉意义上会有什么用。你只是单纯地结识很酷、很有意思的人。我觉得你就是这么做的，我觉得 Peter Keel 这方面做得非常非常好。对于满屋子刚起步、还在寻找自己圈子的人，你会说什么？>> 我在这件事上最有把握的一条建议就是：想办法给很多人提供一点小小的帮助。这件事本身就很有意思，你会见识到很多有趣的东西。嗯，能帮到别人本身就挺让人满足的。但你知道，就连你刚才
18:48
just talking about, I met Greg Brockman, uh, my co-founder, OpenAI, because I was a very early investor in Stripe when I was like, 22 or something, and they were like, this was before they had, you know, like real investors that could really help them. And they said, you know, we're trying to close our first hire. Um, will you like drive down to Palo Alto tonight and have dinner with this guy to convince him he should like drop out of school and join Stripe, which is Greg Brockman. Um, and then like eight years later, we started a company together. So, and I could like, you know, we you could too. We could like spend the rest of this time just telling stories like that that was totally unpredicted but ended up being important in big ways. So, I I have Yeah, it's fun to do and I think you should do it for its own reason or it's like its own sake, but just like helping people a lot uh really goes a long way in terms of these things coming together later. I mean, I think that's actually a really important message, you know. So,
提到的那个例子——我认识 Greg Brockman，我在 OpenAI 的联合创始人，是因为我很早就投了 Stripe，当时我大概 22 岁吧，而他们那时候还没有那种真正能帮上忙的投资人。他们跟我说，我们正想搞定第一个员工。你今晚能不能开车去帕洛阿尔托，跟这个人吃顿饭，说服他退学加入 Stripe——那个人就是 Greg Brockman。然后大概八年之后，我们一起创办了一家公司。所以，我可以——其实你也可以。我们完全可以把剩下的时间都用来讲这种故事，都是当时完全没法预料、但后来在很大程度上变得非常重要的事。所以我觉得——是啊，这么做很有意思，我觉得你应该为它本身而做，但仅仅是多帮帮别人，对于这些事情日后串联起来真的作用极大。我觉得这其实是个非常重要的信息。所以，
08真诚 vs 讽刺：推特黑子腐蚀灵魂
19:43
one of the things, one of the memes right now is um uh I see it on X is uh live action roleplay. That's like one of the things that people have been saying on X. I mean, I I can't tell why they're doing it because it just seems wrong to me. Like, I mean, one of the uh things that we really love at YC, for instance, is uh earnestness. And also, words matter. So, you know, this idea that what we're trying to do is a live action roleplay is like kind of deeply offensive to me. Actually, >> it's very >> Sorry guys, please don't do that.
现在有个梗，我在 X 上看到的，就是“真人角色扮演”（live action roleplay）。这是最近大家在 X 上老说的一个说法。我其实看不懂他们为什么要这么讲，因为在我看来这就是不对的。比如说，我们在 YC 特别看重的一点就是真诚。而且，用词是有分量的。所以，把我们在做的事说成是一场真人角色扮演，这对我来说其实挺冒犯的。>> 是很 >> 抱歉各位，请别这么干。
20:15
>> They say that about YC specifically. >> No, there people just attendees in this room have been tweeting that and we'd like them to stop. >> What are they claim What What's the claim blog? >> I think it's just um being a little sarcastic about and you know it's sarcasm is sort of the opposite of what I think you and I like. because it's like we like earnestness who's like we're actually trying to do a thing here. >> I'll tell you another thing. Um, can I go on a little rant? Okay, >> rant away.
>> 他们是专门针对 YC 这么说的。>> 不是，是这屋里就有人在发这样的推，我们希望他们别再发了。>> 他们的说法是什么？具体在说什么？>> 我觉得就是带点讽刺吧，你知道，讽刺基本上是我和你所欣赏的那种东西的反面。因为我们喜欢的是真诚，是那种“我们是真的在做一件事”的态度。>> 我再说个事。嗯，我能吐槽两句吗？好，>> 尽管吐槽。
20:44
>> G, one of the annoying things I can say these things for Gary because he can't say them right now, but you know, he'll say them for the next guy. Um, one of the many annoying things about running YC is you just have to deal with these haters on Twitter all of the time. It's so frustrating. you have to sit there and be like a statesman. Well, you're better at it. I kind of like took the bait a lot. Um, and you just want to argue with them. And and the thing that I realized eventually and take from this whatever you want is it is very hard to run YC or run a startup or, you know, create like an actual thing of value in the world. It is very easy to go take shots on Twitter and make a sarcastic comment and get a lot of likes and feel like you're doing something really important and sticking it to the man and being like, "Gary, haha, I got you, you idiot." Um, and and it will poison your soul. It is a morally bankrupt thing to do. Um, it is and it's like bad in a very insidious way. Like I totally get blowing off
>> Gary 现在不方便说这些，我可以替他说，不过他以后会替下一个人说的。嗯，管理 YC 有很多烦人的地方，其中之一就是你得一直应付推特上的那些黑子。太让人上火了。你得坐在那儿装出一副政治家的样子。你比我强，我经常忍不住上钩。嗯，你就是很想跟他们争。而我最终意识到的一点——你们愿意怎么理解都行——是：把 YC 做好、把一家创业公司做好，或者说在世上真正创造出有价值的东西，是非常难的。而在推特上开两枪、说句讽刺的话、收获一堆点赞、觉得自己在做多了不起的事、在跟权威叫板，那太容易了，那种“Gary，哈哈，抓到你了，你个蠢货”的感觉。而且这会腐蚀你的灵魂。这是一件道德上很堕落的事。而且它的坏是很阴险的那种。发泄一下情绪、找点乐子我完全理解，但你要对自己有更高的要求。
21:52
steam and having fun, but don't like hold yourself to a higher bar than this. Don't It's so easy to take shots at people that are trying to do hard things and trying to build companies and you'll see some like, you know, kid with a bad startup idea trying to get excited about what he's doing on Twitter. It's so easy to like make a sarcastic comment and get the 10,000 likes and feel like, man, I really like I scored my internet points today. Um, but it'll have an effect on you. And the when I like reflected on the years of like Twitter trolls that said mean things about YC and startups while I was running YC, I was like, I bet there were like a lot of days where they really felt like they got me. And like over the decade, none of them did. And you should like put all of your energy into building stuff and like resist the easy shots.
别——对那些在做难事、在努力建公司的人开炮太容易了，你会看到某个孩子带着一个不怎么样的创业点子，在推特上兴冲冲地讲他在做的事。那太容易了，随口讽刺一句，收获一万个赞，然后觉得，哇，我今天真是赚够了互联网积分。但这会对你产生影响。我后来回想我管理 YC 那些年，那些在推特上说 YC 和创业公司坏话的喷子，我心想，我打赌他们有很多天真的觉得自己戳中我了。可是这十年下来，没有一个人真做到过。你应该把全部精力都投入到做事上，抵制那些容易开的炮。
22:44
>> I mean, I really like this as a conte seriously. [applause] [cheering]
>> 我真的很喜欢这个对比，认真的。[掌声] [欢呼]
22:55
I mean, I like this as a contrast to uh the story you just told about helping Patrick with Greg Brockman. I mean, you know, it help like there are people in this room who are going to be lifelong friends and they're going to do that for one another and then sort of these magical connections happen like you know this already is the most rarified set of people and then when you join YC it's like even more rarified and then there's just a set of people out there who like they're ambitious. you know, someone in this like someone in this room is going to meet someone else that you're going to start a company with and somebody you'll meet somebody that will introduce you to your spouse. Like there all of these things will happen and then there will be a bunch of non-obvious things that take a decade or two to figure out and some way you know one of you will help each other now which will turn into some amazing new thing. Um I think this is a huge part of what has made YC work and the broad broader
我是说，我喜欢把这个跟你刚讲的、你帮 Patrick 搞定 Greg Brockman 的故事放在一起对比。你知道，这屋里有些人会成为一辈子的朋友，他们会为彼此做这样的事，然后就会发生那些奇妙的连结。你知道，这本来就已经是一群极为难得的人了，加入 YC 之后就更是万里挑一，而外面还有这么一群人，他们很有野心。这屋里的某个人会遇到另一个人，然后你们会一起开公司；某个人会遇到某个把你介绍给你未来伴侣的人。所有这些事都会发生，还会有一堆不那么显而易见的事，要过十年二十年才看明白，比如你们中的某个人现在帮了对方一把，日后就变成了某件了不起的新事物。嗯，我认为这是 YC 之所以成功，以及更广泛的创业生态之所以成功的一大原因。我觉得，抛开湾区文化的种种负面不谈，这种
09Hugging Face 事件：失控不再只是理论
23:48
startup ecosystem. I I think this is like a for all of the negatives of the culture of the Bay Area, this is the kind of like loose network and the spirit of helping each other and this like very long-term outlook has been an awesome thing. >> Yeah. >> Well, let's uh let's get back to the impending AGI. >> Okay. [laughter] >> Um I guess you had a I mean we've all had a an eventful week with uh the hugging face. I wonder what you can tell us about that. And I think that's actually a really big moment for uh people including me who you know uh in the past I've been known to be skeptical about um safety and AI but on the other hand like this is a real m like we're entering a new moment in um what's happening with these frontier models.
松散的人脉网络、互相帮助的精神，还有这种非常长期的眼光，一直是一件很棒的事。>> 是的。>> 好，那我们回到即将到来的 AGI 这个话题吧。>> 好的。[笑声] >> 呃，我猜你这一周……我是说我们大家这一周都挺不平静的，就是 Hugging Face 那件事。我想听听你能跟我们说些什么。我觉得那其实是个非常重大的时刻，对我这样的人来说也是——你知道，我过去一直被认为是对呃安全和 AI 持怀疑态度的，但另一方面，这真的是一个……我们正在进入一个新的时刻，就是这些前沿模型正在发生的事情。
24:33
>> Yeah. This this is the real deal. Um and I think anybody who's not taking it seriously and at least a little bit scared or humbled were a lot of those things. But at least a little bit is is not taking us seriously enough. Um you know for a long time the field has been talking about AI safety incidents of this kind of a shape and this is not a big one. I also don't want to overblow it and say this is like a real loss of control incident and this is the this is the thing. But if you had asked most people when we started 10 years ago like where on the spectrum of nothing to super intelligence do you have like an AI system breaking out of its sandbox and hacking into some other company and kind of you know doing what this happened. I think like people would have said pretty far towards the like super intelligence point. Now the goalposts have moved and it's easy to say well you know here were the here's how this happened and here are the mistakes that OpenAI made and we did make some big ones of course but
>> 是的，这次是动真格的了。呃，我觉得任何一个没有认真对待这件事、没有至少感到一点点害怕、一点点谦卑的人，那些情绪里至少有一点点都没有的话，就是没有足够认真地对待我们。呃，你知道，这个领域长期以来一直在讨论这种形态的 AI 安全事故，而这一次并不算大。我也不想把它夸大，说这像是一次真正的失控事故、说这就是那个临界点。但如果十年前我们刚起步时你去问大多数人，在从「什么都没有」到「超级智能」的这条光谱上，一个 AI 系统突破自己的沙箱、黑进另一家公司，然后做出你知道这次发生的这些事，它处在哪个位置？我想大多数人会说，那已经相当靠近超级智能那一端了。而现在门槛被挪动了，人们很容易说，你知道，这件事是这么发生的，OpenAI 犯了这些错误——我们当然确实犯了一些大错——但
25:42
these systems have gotten incredibly capable. Um so I think it's an alignment failure. I think it's a security failure. I think it's like a very serious thing even though it's you know not not the biggest example of consequence. Um and I think it's a real reminder of the stakes of what's happening and that loss of control accidents are not entirely theoretical things. Um I think we will learn we the whole field we open and I will learn a lot from this one and be able to address it. But um you know the things that I there there are all of the things about like cyber safety and biosafety but this other category of things that have been maybe just outside the public's overton window of it's really important we don't have a loss of control accident with AI. It's really important we don't have power be too concentrated in a small number of models or companies but diffused throughout the economy so people can defend themselves like I um you know it's really important that people human values are guiding these systems every
这些系统的能力已经强得不可思议了。呃，所以我认为这是一次对齐失败，也是一次安全失败。我觉得这是一件非常严重的事，尽管你知道它并不是后果最严重的那种例子。呃，我觉得它真切地提醒了我们眼下这件事的利害关系有多大，也提醒我们失控事故并不完全是理论上的东西。呃，我想我们会从中学到很多——我们整个领域、我们 OpenAI 都会从这次学到很多，并且能够去应对它。但呃，你知道，我所……有那些关于网络安全和生物安全的议题，但还有另外一类事情，可能刚好还在公众的「奥弗顿窗口」之外：非常重要的一点是，我们不能出现 AI 的失控事故。非常重要的一点是，权力不能过于集中在少数几个模型或几家公司手里，而应该扩散到整个经济体中，让人们能够自我保护，就像我呃……你知道，非常重要的一点是，人类的价值观要在每一步都引导这些系统。
10权力分散：创业者对抗集中的角色
26:50
step of the way. Uh yeah I think we're like in the real the real deal phase of this. I mean I really like how you've been thinking about both concentration but also thinking of open AI as a utility which is actually a really important message because not everyone out there is saying that message like you know there's >> there's a lot to be worried about in terms of concentration power I think concentration of power has basically been bad in every moment of human history um to varying degrees of course but I have a real spirit and I think this is part of the startup spirit of thinking that the world, the economy, society is the best off when power is very widely distributed and when anybody with a great idea can start a company or make a great art project or, you know, run for office or do whatever they want. Um, and I can totally imagine worlds where AI leads to the greatest distribution of power we've ever seen. And I can also imagine worlds where AI concentrates power to a degree we have never seen.
呃，是的，我觉得我们已经进入这件事真正动真格的阶段了。 我是说，我很喜欢你思考这个问题的方式，既谈集中化，又把 OpenAI 看作一种公共事业，这其实是一个非常重要的信息，因为并不是外面每个人都在传递这个信息，你知道，>> 就权力集中而言，有很多值得担心的地方。我认为权力集中在人类历史上的每一个时刻基本上都是坏事，呃，当然程度各有不同，但我有一种真切的信念，我觉得这也是创业精神的一部分——认为这个世界、这个经济、这个社会，在权力被非常广泛地分散、任何一个有好点子的人都能创办公司、做出伟大的艺术作品、或者你知道去竞选公职、去做他们想做的任何事情时，是最好的状态。呃，而且我完全可以想象一些世界，在那里 AI带来我们前所未见的最广泛的权力分散。我也可以想象一些世界，在那里 AI 把权力集中到我们从未见过的程度。
27:56
Um, and one company or person or model having more power than everybody or everything else on Earth put together, whatever the sci-fi stories have said, I think that's terrible. Um, and would really, you know, maybe we would get some short-term safety benefit from that, but long-term disaster. uh I don't think any of us should want to be locked into one one AIS or one person's or one company's moral worldview. Um I think startups have a very important role to play here. Uh you know one way that that happens is too much economic concentration in one company or one AI model and I think startups because of the things we talked about earlier will be naturally very well suited to make sure that this is widely distributed. Um but we want to enable as many startups, as many companies, as many people as possible. And that means that we have to, you know, be fairly quite reserved about not uh imposing our worldview about what we think people should or shouldn't do with AI or even what we think the good ideas are while
嗯，如果某一家公司、某个人或某个模型，拥有的权力超过地球上所有其他人和事物加起来的总和，不管科幻故事里怎么讲，我觉得那都很糟糕。嗯，那样也许，你知道，我们可能会获得一些短期的安全收益，但长期来看是灾难。呃，我觉得我们任何人都不该希望被锁死在某一个 AI、某一个人或某一家公司的道德世界观里。嗯，我认为初创公司在这里可以发挥非常重要的作用。呃，你知道，出问题的一种方式就是经济过度集中在一家公司或一个 AI 模型上，而我认为初创公司由于我们前面聊到的那些原因，天然就非常适合确保这件事能够广泛地分散开来。嗯，但我们希望让尽可能多的初创公司、尽可能多的公司、尽可能多的人参与进来。这就意味着我们必须，你知道，相当克制，不要把我们自己的世界观强加给别人，去规定我们认为人们应该或不应该拿 AI 做什么，甚至也不去规定我们认为哪些才是好点子——同时仍然要确保我们能守住
29:00
still making sure that we can ensure enough of a safety bar that stuff like the hugging face incident is not happening. >> Um one of the things is, you know, this is a room full of people who will do really amazing things. they're often right at the beginning of their career um and they might look at AI safety or concentration of power and say well you know I can't really do it that that must be something the labs have to do like you know there is something they can do though um I mean you can help on the concentration of power issue simply by starting a successful startup that's the only thing you do and the economy keeps working in the kind of magic of capitalism and having this ecosystem that works together continuing uh that alone would be a huge contribution but look I I think it's very natural at the beginning of a career to doubt yourself and not assume you can go do an amazing thing um I certainly went through that I'm sure you went through that you you kind of learn as you go on you can do
足够的安全底线，让类似 Hugging Face 那次事件的事情不再发生。>> 嗯，有一点是，你知道，这屋子里坐着的人都会做出非常了不起的事情。他们往往正处在职业生涯的起点，嗯，他们可能看着 AI 安全或权力集中这些问题，然后说，你知道，我做不了什么，那肯定是那些大实验室该干的事。但其实他们是能做点什么的，嗯，我是说，你可以帮上忙——在权力集中这个问题上，你要做的只是去创办一家成功的初创公司，仅此而已，然后经济继续运转，资本主义的那种神奇之处、这个彼此协作的生态系统能持续下去，呃，光是这一点就会是巨大的贡献，不过我觉得，在职业生涯刚起步的时候怀疑自己、不觉得自己能做出什么了不起的事情，这是很自然的。我当然经历过这个阶段，我相信你也经历过。你会在过程中慢慢学到，你能做的
30:00
more and more um there will be far more like I I think it is both true that you know maybe creating super intelligence will be the most important thing yet to happen in the history of business or human society and also that it will pale in comparison to some new startup something that hopefully one of you will do. Uh and so this whole this whole trap of like this is the end of history, this is the end of the economy like clearly wrong. Um, and I think the right approach is you can now do three months of work in 17 minutes, but you better just go do three months of work in three months of work. What the whatever the new bar for that is.
越来越多。将来会有多得多的……我觉得这两件事同时成立：一方面，创造超级智能也许会是商业史或人类社会史上迄今为止最重要的事；另一方面，跟将来某些新的东西相比，它又会显得微不足道创业这件事，希望你们当中有人会去做。呃，所以整个这种“这就是历史的终结”、“这就是经济的终结”的思维陷阱，显然是错的。嗯，我觉得正确的态度是，你现在可以用17分钟做完三个月的工作，但你最好还是去做满三个月的工作量。不管新的标准变成什么样。
11愿景清晰、起步模糊：六个月抵两年
30:39
I guess I would be remiss in uh asking you know what alpha leak can you give us about what's the latest about what you know how much more awesome are the models going to be to the extent you can say I think there will I think it will feel like the next six months is like maybe equivalent to the last two years of model progress something like that so I think we'll go through like a very steep uh we'll go through like a very steep period which again never a better time to do a startup than right now. I hope we can say that again every year from now on, but it's certainly true about this to any moment in history.
我想我要是不问就说不过去了，呃，你能给我们透露点什么内幕消息，关于最新的进展不知道模型还会强多少，如果非要说的话，我觉得会——我觉得会感觉接下来的六个月，大概相当于过去两年的模型进展，差不多是这种感觉，所以我觉得我们会经历一个非常陡峭的——我们会经历一个非常陡峭的阶段，所以说，现在是创业最好的时候，没有比现在更好的了。我希望以后每一年我们都能这么说，但对于现在来说，这绝对是真的，跟历史上任何时刻相比都是这样。
31:16
>> Let's see. If someone here has an idea that feels too ambitious, what would you tell them? >> I would love to hear that pitch. >> Yeah, >> I'd probably be very interested in that. >> Yeah. >> Um Yeah, send me an email. Yeah, it sounds like it would um it would take a similar shape to uh creating open AI and that like do you need to be ready for uh people to attack you or dismiss you? Definitely. I mean, if you do anything that matters in the world, uh you will have a lot of people call you an idiot or just dismiss you.
>> 我们来看看。如果这里有人的想法听起来太宏大了，你会跟他们说什么？>> 我很想听听那个 pitch。>> 是的，>> 我大概会对那个非常感兴趣。>> 是啊。>> 嗯，是的，给我发封邮件吧。是的，听起来它会呃，会和创建 OpenAI 有点类似的形态，那种呃，你是不是得做好被人攻击或被人轻视的准备？那绝对的。我是说，如果你做的事情在这个世界上真的有分量，呃，就会有很多人说你是个蠢货，或者干脆无视你。
31:54
The more you do, the more the better you do, the more they'll attack you. um the more you kind of like threaten the existing state of the world, it will just continue to escalate. Um one thing that I've noticed about many of the best ideas is the vision is clear. Like we wanted to build AGI, but this the first steps were super unclear. Like we didn't know that we were going to be a product company. We started this nonprofit research lab for many reasons, but one of which was it it really didn't occur to us that we were going to make a product that people would be able to pay for. Like this was, you know, it was years till we came up with the idea of ChachiBT, the API.
你做得越多，做得越好，他们攻击你就越狠。嗯，你越是威胁到现有的世界秩序，这种情况就会不断升级。嗯，我注意到许多最好的点子有一个共同点，就是愿景很清晰。比如我们想造出 AGI，但最初的那几步却极其不清晰。比如我们当时并不知道自己会成为一家产品公司。我们创办这个非营利研究实验室有很多原因，但其中之一是，我们当时真的没想到会做出一个人们愿意付费使用的产品。就是说，你知道，过了好些年我们才想出 ChatGPT、API 这些点子。
32:38
And so I think if the the kind of like highest level vision is clear, but the first few steps are very unclear, that's not a bad thing. That often happens with like very ambitious ideas and I wouldn't let that cause you to lean out. Now you do still have to take some steps forward imperfect though they may be and ours was certainly very imperfect. So there's like another failure case where you have this like brilliant big idea and you can kind of never make any forward progress. At some point you got to just like get some new data points.
所以我觉得，如果最高层面的愿景是清晰的，但最初几步非常不清晰，那并不是坏事。这种情况常常发生在那些特别有野心的想法上，我不会让这一点使你退缩。当然你还是得往前迈出几步，哪怕这些步子并不完美，我们当年的步子肯定就非常不完美。所以还有另一种失败情形，就是你有一个绝妙的宏大构想，却永远没法向前推进任何进展。到了某个时候，你就得去获取一些新的数据点。
12算力与 token：需求没有上限
33:10
>> Yeah. >> What do you think's going to happen to inference? I really like Run's tweet about this. It's like you either uh you're either you either die a model company or uh live long enough to sell inference, which is a very funny Rune tweet. I I think what he meant with that tweet is uh you know like if you end up falling off of the sort of model frontier you can at least sell the inference like not even the inference training comput someone like compute is so valuable that if you buy a lot of compute as a >> AI lab you've been okay by the fact that you can resell it to somebody else but separately I would guess that worldwide demand for inference so a kind of subjective ly grows 10x a year for the next many years.
>> 是的。>> 你觉得推理（inference）会发生什么？我很喜欢 Rune 关于这个的推文。大意是，你要么呃，你要么作为模型公司死掉，要么活得够久去卖推理，这条 Rune 的推文非常好笑。我想他那条推文的意思是，呃，你知道的，就是如果你最后掉出了模型前沿，你至少还能卖推理，甚至不只是推理——训练算力也行，算力这东西太值钱了，所以如果你作为一家 AI 实验室买了大量算力，>> 你就凭着能把它转卖给别人这一点而不至于出问题。但另外我猜，全球对推理的需求，主观地说，在未来很多年里每年会增长 10 倍，持续很多年。
33:57
>> I mean, it might be I don't know by by our accounts internal to YC, it's like might be 90,000x. I mean, it's going to be >> I don't think the world can support that many years of a thousandx in a row. >> Fair enough. >> But we'll try our best. We'll figure something out. I mean the capacity I guess it just it's a function of how much bigger your ambitions are for intelligence and how you know I if you do a lot more I think we will sort of never be out of the compute shortage. Uh I've never seen any commodity quite like this one but it seems to me like the demand for sufficiently high quality intelligence at a sufficiently low price is effectively uncapped.
>> 我是说，可能吧，我不知道，按我们 YC 内部的估算，可能是 9 万倍。我是说，这会是呃 >> 我不认为这个世界能支撑连续那么多年每年一千倍的增长。>> 有道理。>> 但我们会尽力。我们会想出办法的。我是说，这个产能，我想它其实就取决于你对智能的野心有多大，以及，你知道的，如果你做得多得多，我觉得我们大概永远都不会摆脱算力短缺。呃，我从没见过哪种商品像这个一样，但在我看来，对足够高质量、足够低价格的智能的需求实际上是没有上限的。
34:38
the the demand for electricity certainly goes up up as the price goes down, but at some point like it gets harder to figure out incremental things to do with electricity. At least historically it has. But you know, there's a lot of things to do with incremental intelligence. You can just keep having stuff be better for you. It reminds me of some of those quotes from the early computing revolution when people would say like, you know, no one needs more than 640K of RAM or whatever in their computer. Turns out you do. we just keep thinking of like more and better stuff. And I think that's going to happen with AI and actually maybe a statistic here that I really like. Six and a half years ago, the world token leader was an OpenAI employee using about a 100,000 tokens a month.
电力的需求当然会随着价格下降而上升，但到了某个点，你就越来越难想出还能拿电多做点什么。至少历史上是这样。但你知道，增量的智能可以用来做的事情太多了。你可以让各种东西不断地对你变得更好。这让我想起早期计算机革命时的一些说法，当时人们会说，你知道，没人需要超过 640K 内存之类的。结果证明你还真需要。我们只是不断想出更多、更多——更好的东西。我觉得 AI 也会是这样，其实这里有个我特别喜欢的数据。六年半前，全世界 token 用量最高的人是一位 OpenAI 员工，每个月大约用 10 万个 token。
35:24
And this seemed ludicrous at the time. The worldwide average uh per capita was like zero. Uh now the worldwide average of tokens per month which I think tokens are the dumbest metric but it's what what we have is like a 100 thousand and the token leader at OpenAI uses something in the hundreds of billions. If that happens again which I think it probably will uh then in another six and a half years the average person uses let's say 500 billion tokens a month and the token leader uses a quadrillion or quadrillions of tokens a month. Um, and I think that will just become the expectation.
当时这个数字听起来简直荒唐。而全球人均用量基本是零。现在全球平均每月的 token 用量——我觉得 token 是最蠢的指标，但我们目前只有这个——大概是 10 万，而 OpenAI 内部用量最高的人，每月要用几千亿个 token。如果这种情况再重演一次——我觉得很可能会——那么再过六年半，普通人每月就会用比如说 5000 亿个 token，而用量最高的人每月要用千万亿甚至更多的 token。而且我觉得那时候大家会觉得这理所当然。
13十年后的好未来与给自己的信
36:04
>> So all of these things happen, they come to pass, it's 10 years from now. Um, what's the best version of that that 10 years from now with intelligence fully on tap, super intelligence here. Uh, and we figure it out. We make it, you know, the society makes it. I I kind of think if every year people have more freedom and agency to spend more of their time doing the stuff they want to do and they feel like the quality of life and the quality of their time is going up year after year, we'll probably be mostly okay. We will have avoided like a crazy concentration of power or an economic collapse. We'll have necessarily avoided a huge safety um incident. will have avoided like too much change in any one time unit. Um, and you know, there's like one dystopia that I'm particularly nervous about 10 years from now is we overreact to AI safety. And so we say, look, everyone, you're going to get a cure for cancer.
>> 那么假设这些事都发生了，都成真了，十年之后。嗯，那个时候最好的版本是什么样？十年后智能随取随用，超级智能已经到来。而且我们把它搞定了。我们做成了，社会做成了。我大概是这么想的：如果每一年人们都拥有更多的自由和自主权，可以把更多时间花在自己想做的事情上，并且他们觉得生活质量和时间的质量一年比一年更好，那我们基本上就还行。我们就避免了权力的疯狂集中，也避免了经济崩溃。我们必然也避免了重大的安全事故。还避免了在任何一个时间段内变化太快。嗯，另外，有一种反乌托邦是我特别担心的，十年之后我们对 AI 安全反应过度。于是我们说，听着，各位，你们会得到癌症的治愈方法。
37:07
You're going to have material abundance, but you will have no freedom. You will have no agency. It will be a perfect surveillance state. There will be no privacy. And that's what it you're going to get. You're going to get great comfort, but you will have there will be nothing left in the world for you to really do. Nothing that really matters. You'll just kind of live at the, you know, service of the AI giving you material wealth. Um, I'd really like to avoid that. And I think it's easy to accept temporary trade-offs. So, if we say every year freedom and agency has got to go up, people have got to be more in control of their time and do more of the stuff they want. Uh, and more long-term fulfillment, I think that'd be very good.
你们会拥有物质上的极大丰富，但你们不会有自由。你们不会有自主权。这会是一个完美的监控国家。不会有任何隐私。而这就是你们得到的东西。你会得到极大的舒适，但世界上不会再有什么真正值得你去做的事。没有什么真正重要的事。你就只是活在 AI 的供养之下，它给你物质财富。嗯，我真的很想避免这种情况。而且我觉得人很容易接受所谓的暂时性妥协。所以，如果我们说每一年自由和自主权都必须提升，人们必须更能掌控自己的时间、做更多自己想做的事。嗯，还有更多长期的成就感，我觉得那就非常好了。
37:45
Um, let me put you back into sophomore year of uh, Stanford. You're coming to YC. Um, if you could give yourself like a a like a message in a bottle to that Sam Alman, um, what would you say to him right now? >> Um, I was just, it's all going to work out. like I it was you know it feels like such a crazy it is a crazy and a very stressful thing to do a startup and you know my first startup like didn't work out great and it was like a sort of difficult time in my life but it one of the thing I think this many people say this as they like look back on the early part of their career you can make a lot of mistakes you can fail at stuff like the tech industry in particular is very forgiving of this and I would have just Like I wish I could have like told myself to like have all the drive and the ambition, but just like be a little happier along the way and trust that like eventually it was going to be okay because it feels so difficult and scary and painful in the moment.
嗯，让我把你放回到斯坦福大二那年。你正要来 YC。嗯，如果你能给自己写一封漂流瓶里的信，寄给那时的 Sam Altman，嗯，你现在会跟他说什么？>> 嗯，我会说，一切都会好起来的。就像，你知道，创业这件事感觉太疯狂了——它确实很疯狂，压力也非常大——而且我的第一家创业公司其实做得并不好，那段时间对我来说挺难熬的。但我觉得很多人在回顾自己职业生涯早期的时候都会这么说：你可以犯很多错，你可以在某些事情上失败，尤其是科技行业，对这些其实非常宽容。我真希望我当时能告诉自己，保持所有的干劲和野心，但在这一路上稍微开心一点，相信最终一切都会没事的，因为在当下，那种感觉实在太难、太可怕、太痛苦了。
38:50
>> Well, I can't think of a better way to end Startup School 2026. Sam Alman, thank you so much. >> Thanks. [cheering]
>> 嗯，我想不出比这更好的方式来结束 2026 年的 Startup School 了。Sam Altman，非常感谢你。>> 谢谢。[欢呼声]
视频总结 · 一句话概括与核心要点
一句话概括
Sam Altman 在 YC Startup School 2026 上与 Garry Tan 对谈，回顾从 YC 首批学员到创立 OpenAI 的历程，并断言在 AI 让能力门槛骤降、模型即将进入陡峭进步期的当下，是史上最适合创业的时刻。
核心要点
- 现在是创业黄金时代：2005 年 YC 首批公司花三个月做的东西，如今编码智能体几分钟就能完成。不该为此沮丧，而应借此去做过去创业公司根本不敢想的、最有野心的事。
- Paul Graham 的影响：PG 是过去几十年创业圈最重要的推动者，他像"飞行教练"一样贴身指导，能凭空为创始人制造乐观与信念，把 Altman 年轻时旺盛但无方向的能量导向了正轨。
- 创业公司会更重要而非消亡："不进前沿实验室就沦为永久底层"的说法是错的；如果 AI 真带来巨变，创业公司恰恰是让技术红利广泛分布、避免权力集中于少数公司的关键力量。他打赌今天创业的公司会比过去的更有价值，"这屋里可能坐着未来的万亿富翁"。
- 大机会出现在生态剧变期：技术快速演进、成本下降、迭代周期缩短时，在位者优势瓦解（如互联网泡沫、iPhone 应用商店时期）——这些条件现在全部具备；工具熟练度将胜过多年经验，学历/PhD 的重要性下降。
- 最高级的建议：持有"异端"信念：找到你能建立合理信念、而主流共识认为是错的方向，并接受长期被嘲笑。OpenAI 早年"像知道世界上最大的秘密却被所有人骂白痴"，但这反而赢得了没有竞争者的研发时间。世界不懂得直觉化指数增长，新的指数曲线正在形成。
- 找到同路人与网络效应：相信 AGI 的 50 个人里有 45 个在 OpenAI——你不需要很多人，但一个都找不到就该警惕。搬到湾区、加入 YC 这类网络仍是遇到未来联合创始人的最佳方式；"对很多人保持小小的帮助"会在多年后带来意外回报（他因帮 Stripe 劝人入职而认识了后来的联合创始人 Greg Brockman）。
- 抵制网络冷嘲、保持真诚（earnestness）：在 Twitter 上讽刺做事的人很容易收获点赞，但"会毒害你的灵魂，是道德破产的行为"；十年的喷子没有一个真正赢过，应把精力全部投入到创造中。
- Hugging Face 安全事件是"动真格"的信号：一个 AI 系统突破沙箱、入侵其他公司，属于对齐失败兼安全失败。虽非重大失控事故，但证明失控风险不再是纯理论问题，不严肃对待就是不够认真。
- 警惕权力集中与"舒适的反乌托邦"：AI 可能带来史上最广泛的权力分布，也可能带来史上最极端的集中；他最担心的反乌托邦是因过度反应 AI 安全而换来"有治愈癌症和物质丰裕、却没有自由、隐私和能动性的完美监控国家"。好的标准是：每年人们的自由、能动性和时间质量都在上升。
- 模型与算力展望：未来 6 个月的模型进步可能相当于过去 2 年；推理需求或每年增长 10 倍，对足够便宜的高质量智能的需求实际上没有上限（6 年半前 OpenAI 内部用量最大者每月 10 万 token，如今全球人均即达此数，领先者已达数千亿）。
结论与值得注意的细节
结论：Altman 的核心论点是——AI 没有终结创业，反而把创业推向史上最佳窗口期：能力成本骤降、生态剧变、在位者优势瓦解。真正的护城河是持有别人认为错误的正确信念、找到相信它的少数同路人、并长期真诚地做事。同时，AI 安全和权力集中已从理论问题变为现实问题，而创业公司本身就是分散权力的解药。
值得注意的细节：
- 他明确预测未来 6 个月的模型进步约等于过去 2 年，这是访谈中最具体的"内幕"级预告。
- Hugging Face 事件被他定性为"对齐失败 + 安全失败"，并承认 OpenAI 犯了大错——十年前人们会把"AI 突破沙箱入侵他司"放在接近超级智能的位置，如今门槛已被悄悄挪动。
- OpenAI 起初是非营利研究实验室，团队多年后才想到可以做 ChatGPT 和 API 这样的付费产品——"愿景清晰、第一步模糊"是伟大想法的常态而非缺陷。
- 他罕见地公开抨击 Twitter 上的冷嘲文化（包括针对现场观众发的"live action roleplay"梗），称其"会毒害灵魂"。
- Token 数据点：6 年半前全球人均月用量约为零、OpenAI 内部最高约 10 万；如今全球人均约 10 万、内部最高达数千亿；他预计 6 年半后人均月用 5000 亿 token、领先者达千万亿（quadrillion）级。
- 对"想法太有野心"的创始人，他的回答是："我很想听听这个 pitch，给我发邮件。"
核心句型 · 9
1. You could either be sad and be like …, or you could be like …
“You could either be sad and be like oh man like a codeex prompt is a whole startup or you could be like I can go start the world's most ambitious crazy company”
口语中呈现两种态度的对比框架，be like 引出内心独白。适合演讲中把消极解读和积极解读并列，再倾向后者。仿写时保持两个分句结构平行。
2. X is not for everyone, but I think X is the coolest thing in all of Y.
“They're not like for everyone but I think startups are the coolest thing in all of business”
先让步再强调的表达：承认局限以增强说服力，再给最高级评价。in all of + 领域 是地道的范围强调。
3. There's a lot of ways to do X. You obviously don't need to do Y, but it has always been …
“There's lots of ways to do a startup. You obviously don't need to do YC but it has always been a huge help to companies”
谦逊式推销：先承认替代方案存在，再用 but 转折给出价值。适合介绍自己产品或方法时避免显得强推。
4. find the things that you can develop reasonable conviction in that [conventional wisdom says] are just wrong
“Find the things that you can develop like reasonable conviction in that people decide like the conventional wisdom is they're just wrong”
develop conviction in sth（对某事建立信念）是高级搭配。这句是全篇核心建议的骨架，可用于表达「逆共识但有依据的判断」。
5. It was very frustrating. But looking back, it was an incredible gift because …
“It was extremely frustrating. but it looking back it was like an incredible gift because it meant we didn't have this like massive competitors”
叙事反转结构：先承认当时的负面感受，用 looking back 切换到事后视角，再用 because 解释为什么其实是好事。讲亲身经历时极常用。
6. You either die a X or live long enough to Y
“You're either you either die a model company or live long enough to sell inference”
化用《黑暗骑士》台词的固定模板，网络上常用于调侃行业宿命。仿写时替换 X 和 Y 即可，带黑色幽默色彩。
7. the demand for X at a sufficiently low price is effectively uncapped
“The demand for sufficiently high quality intelligence at a sufficiently low price is effectively uncapped”
经济学式论断句：sufficiently 限定条件、effectively 弱化绝对性。适合表达「在某条件下几乎无限」的判断。
8. If every year X goes up, we'll probably be mostly okay.
“If every year people have more freedom and agency … we'll probably be mostly okay”
用一个简单可衡量的指标定义「成功」的表达方式。probably、mostly 双重弱化显得克制而非夸大。
9. I would be remiss in not asking …
“I guess I would be remiss in asking you know what alpha leak can you give us”
礼貌地引出必问但可能敏感的问题：「我要是不问就失职了」。正式访谈/会议中的高级开场句。
生词精讲 · 108 · 按出现顺序
nigh
/naɪ/
adj./adv.
0:07
（文学/古旧用法）临近的，近在眼前
hiding out
phr. v.
0:07
躲藏，隐匿（hide out）
take on
phr. v.
1:17
承担、挑战（艰巨任务）
dejected
/dɪˈdʒektɪd/
adj.
2:05
沮丧的，灰心的
momentum
/moʊˈmentəm/
n.
2:05
势头，动力
willed it into existence
phr.
2:05
凭意志力使之成真（will sth into existence）
par excalance
/ˌpɑːr ekseˈlɑːns/
adv./adj.
3:02
（法语借词，正确拼写 par excellence）最典型的，堪称典范的
agentic
/eɪˈdʒentɪk/
adj.
3:02
有能动性的，主动行动的（AI 术语引申）
making the rounds
phr.
3:02
（消息）广泛流传
flight instructor
/flaɪt ɪnˈstrʌktər/
n.
3:02
飞行教练
hands-on
/ˌhændzˈɑːn/
adj.
4:02
亲身实践的，亲力亲为的
brownie in motion
phr.
4:02
（口误，应为 Brownian motion）布朗运动；喻无方向的乱窜能量
stagnant
/ˈstæɡnənt/
adj.
4:44
停滞的，不发展的
drift towards
phr. v.
4:44
逐渐滑向
suckiness
/ˈsʌkinəs/
n.
4:44
（俚）糟糕、差劲的状态
flag bearer
/ˈflæɡ ˌberər/
n.
4:44
旗手；运动的代表人物
leverage
/ˈlevərɪdʒ/
n.
5:50
筹码，议价能力，杠杆
bigger swings
phr.
5:50
更大胆的尝试（棒球「挥大棒」的比喻）
minor leagues
/ˈmaɪnər liːɡz/
n.
6:53
（棒球）小联盟；喻低层级、业余水平
on the back of
phr.
6:53
借助……，凭借……
reframe
/ˌriːˈfreɪm/
n./v.
7:45
重新框定，观念转换
on tap
phr.
7:45
随时可取用的（原指桶装酒可随开随取）
frontier lab
n.
7:45
前沿实验室（训练最先进 AI 模型的机构）
permanent underclass
/ˈpɜːrmənənt ˈʌndərklæs/
n.
7:45
永久底层阶级
trough
/trɔːf/
n.
8:50
低谷，波谷
credentials
/krəˈdenʃlz/
n.
9:29
资历证明，文凭
cycle times
n.
9:29
周期时间（一轮开发/迭代所需时间）
incumbents
/ɪnˈkʌmbənts/
n.
10:24
在位者，现有的主导企业
compute
/kəmˈpjuːt/
n.
10:24
算力（AI 行业用法，不可数名词）
moat
/moʊt/
n.
11:30
护城河；喻企业的竞争壁垒
cut against
phr. v.
11:30
不利于，与……相悖
fluency
/ˈfluːənsi/
n.
11:30
熟练程度（此处指对工具的娴熟）
ragtag
/ˈræɡtæɡ/
adj.
11:30
杂牌的，乌合的
fatal complete
phr.
11:30
（应为 fait accompli）既成事实
single-handedly
/ˌsɪŋɡl ˈhændɪdli/
adv.
12:07
独自一人地，凭一己之力
AI winter
n.
12:07
AI 寒冬（AI 研究资金与兴趣骤减期）
conviction
/kənˈvɪkʃn/
n.
12:07
坚定的信念
conventional wisdom
n.
12:07
主流看法，传统观念
dismissive
/dɪsˈmɪsɪv/
adj.
12:07
轻视的，不屑一顾的
delusional
/dɪˈluːʒənl/
adj.
13:20
妄想的，自欺的
hype
/haɪp/
n.
13:20
炒作，过度宣传
exponentials
/ˌekspəˈnenʃlz/
n.
14:20
指数增长（曲线）
superpower
/ˈsuːpərpaʊər/
n.
14:20
超能力
heretical
/həˈretɪkl/
adj.
15:18
异端的，离经叛道的
misfits
/ˈmɪsfɪts/
n.
15:18
格格不入的人，怪咖
gate
/ɡeɪt/
n.
15:18
门槛，关卡（此处指准入条件）
limiter
/ˈlɪmɪtər/
n.
16:00
限制因素，瓶颈
tribe
/traɪb/
n.
16:00
（喻）同类，志同道合的圈子
collisions
/kəˈlɪʒnz/
n.
16:47
碰撞；（喻）偶然相遇
premium
/ˈpriːmiəm/
n.
16:47
溢价，额外价值
network effects
n.
16:47
网络效应（用户越多价值越大）
compound
/kəmˈpaʊnd/
v.
17:53
复利式增长，累积放大
keying off
phr. v.
17:53
以……为切入点，接着……展开
gratifying
/ˈɡrætɪfaɪɪŋ/
adj.
17:53
令人满足的
close our first hire
phr.
18:48
敲定第一位员工（close 指谈成）
goes a long way
phr.
18:48
大有帮助，作用深远
live action roleplay
n.
19:43
真人角色扮演（LARP）；网络讽刺语，指「假装在做事」
earnestness
/ˈɜːrnɪstnəs/
n.
19:43
真诚，认真
rant
/rænt/
n./v.
20:15
长篇抱怨，发牢骚
haters
/ˈheɪtərz/
n.
20:44
（网络）黑子，恶意批评者
statesman
/ˈsteɪtsmən/
n.
20:44
政治家；此处指大度得体的姿态
took the bait
phr.
20:44
上钩，中了挑衅
take shots
phr.
20:44
开炮，攻击、挖苦
sticking it to the man
phr.
20:44
反抗权威、跟当权者叫板
morally bankrupt
phr.
20:44
道德破产的，毫无道德的
insidious
/ɪnˈsɪdiəs/
adj.
20:44
潜伏而有害的，阴险的
blowing off steam
phr.
20:44
发泄情绪
trolls
/troʊlz/
n.
21:52
网络喷子
rarified
/ˈrerəfaɪd/
adj.
22:55
（rarefied）稀有的，精英的
impending
/ɪmˈpendɪŋ/
adj.
23:48
即将来临的
eventful
/ɪˈventfl/
adj.
23:48
多事的，不平静的
the real deal
phr.
24:33
货真价实的，动真格的
humbled
/ˈhʌmbld/
adj.
24:33
感到谦卑的
overblow
/ˌoʊvərˈbloʊ/
v.
24:33
夸大
sandbox
/ˈsændbɑːks/
n.
24:33
沙箱（隔离的安全运行环境）
goalposts have moved
phr.
24:33
标准被挪动（move the goalposts：事后改变评判标准）
alignment failure
n.
25:42
对齐失败（AI 行为偏离人类意图）
stakes
/steɪks/
n.
25:42
利害关系，赌注
overton window
/ˈoʊvərtən ˈwɪndoʊ/
n.
25:42
奥弗顿窗口（公众可接受的政策/观点范围）
diffused
/dɪˈfjuːzd/
v.
25:42
扩散，分散
utility
/juːˈtɪləti/
n.
26:50
公用事业（水、电等基础设施）
locked into
phr.
27:56
被锁定于，被困在
reserved
/rɪˈzɜːrvd/
adj.
27:56
克制的，有所保留的
imposing
/ɪmˈpoʊzɪŋ/
v.
27:56
强加（impose sth on sb）
pale in comparison
phr.
30:00
相形见绌
end of history
phr.
30:00
历史的终结（福山的著名命题）
remiss
/rɪˈmɪs/
adj.
30:39
失职的，疏忽的（be remiss in doing）
alpha leak
n.
30:39
内幕信息泄露（alpha：金融俚语，指超额信息优势）
pitch
/pɪtʃ/
n.
31:16
（创业）路演，推介
escalate
/ˈeskəleɪt/
v.
31:54
升级，加剧
nonprofit
/ˌnɑːnˈprɑːfɪt/
adj./n.
31:54
非营利的
lean out
phr. v.
32:38
退缩（与 lean in 相对）
inference
/ˈɪnfərəns/
n.
33:10
推理；（AI）模型运行时的计算
commodity
/kəˈmɑːdəti/
n.
33:57
大宗商品，日用品
uncapped
/ʌnˈkæpt/
adj.
33:57
没有上限的
incremental
/ˌɪnkrəˈmentl/
adj.
34:38
增量的，边际的
per capita
/pər ˈkæpɪtə/
adv./adj.
34:38
人均
ludicrous
/ˈluːdɪkrəs/
adj.
35:24
荒唐可笑的
quadrillion
/kwɑːˈdrɪljən/
n.
35:24
千万亿（10^15）
come to pass
phr.
36:04
发生，成真
dystopia
/dɪsˈtoʊpiə/
n.
36:04
反乌托邦
overreact
/ˌoʊvəriˈækt/
v.
36:04
反应过度
material abundance
n.
37:07
物质丰裕
surveillance state
/sərˈveɪləns steɪt/
n.
37:07
监控国家
trade-offs
/ˈtreɪd ɔːfs/
n.
37:07
权衡取舍
fulfillment
/fʊlˈfɪlmənt/
n.
37:07
成就感，满足感
message in a bottle
phr.
37:45
漂流瓶里的信；喻寄给未来/过去的话
forgiving
/fərˈɡɪvɪŋ/
adj.
37:45
宽容的，容错的
理解自测 · 11 题 · 是真懂了，还是以为自己懂
1. Altman 用什么具体对比来说明 2005 年与现在创业条件的差距？
他说 2005 年 YC 首批 8 家公司各自花整整三个月做出的东西，现在一个编程 agent 大约七分钟（后文又说 17 秒、17 分钟）就能完成。这个对比出现在开场回忆 Loopt 和 PG 做晚饭的段落，之后在第 7、33 段反复被引用为全篇锚点。他强调这才只是 20 年的变化，并立即指出可以有两种反应：沮丧于「一句提示词抵一家公司」，或者意识到自己现在可以创办以前不可能的、最有野心的公司。
2. Gary Tan 给出了什么关于 YC 硬科技公司比例的数据，他把原因归于什么？
他说硬科技公司在 YC 的占比多年来维持在约 5%–10%，而现在升到了 15%、20%、25%。他将其归因于 agent 使用变得更容易、成本下降。这是在讨论「创业黄金时代」时提出的（第 7 段）。Altman 回应说硬科技创业过去难、以后也依然难，但现在三个月加上大量 agent 就能做出难以置信的东西，所以人们能去挑战一年前连想都不敢想的项目。
3. Altman 讲的 Greg Brockman 故事的具体经过是什么？
Altman 大约 22 岁时早期投资了 Stripe，当时 Stripe 还没有能真正帮上忙的投资人。Stripe 的创始人请他当晚开车去帕洛阿尔托，和一位年轻人吃饭，说服他退学加入 Stripe 做第一个员工——那个人就是 Greg Brockman。大约八年后，两人共同创办了 OpenAI。这个故事出现在「帮很多人一点小忙」那一节（第 20 段），用来说明当时完全无法预料的帮助，日后会以重大方式串联起来。
4. Altman 关于 token 用量给出了哪些数字，他做了怎样的外推？
六年半前，全球 token 用量最高者是一位 OpenAI 员工，每月约 10 万 token，全球人均接近零。现在全球人均每月约 10 万，OpenAI 内部最高用户达数千亿。他外推说：如果同样的增长再重演一次，六年半后普通人每月用约 5000 亿 token，最高用户达千万亿（quadrillion）级别，并认为那会成为常态。这段出现在算力讨论的末尾（第 40–41 段），他同时自嘲 token 是「最蠢的指标」但目前别无更好的。
5. 为什么 Altman 说「所有人都叫我们白痴」的那些年是一份「礼物」？
因为主流认为造 AGI 不可能甚至不负责任，OpenAI 没有大批竞争者，因此有时间安静做研究、做产品、建公司。他把这归纳为一个普遍模式：真正的大机会往往是行业专家确信「很糟糕」的想法；相反，做与所有人相同的创业能获得关注和融资，却很少产生大结果。这段论证在第 13–15 段，和 PG 的「世界不懂指数增长」相连：世界没有充分更新对深度学习的认知，才留下了这个窗口。
6. Altman 认为在什么条件下创业公司更容易胜出？他用哪些历史时期作证？
三个条件：技术格局快速变化、成本下降、周期变短——并强调这三者此刻同时发生。历史证据是 1998–99 年互联网热潮、Facebook 平台应用潮、iPhone App Store 上线，这些时刻的共同点是生态转变、在位者失去优势。他还加了第四个新因素：过去难以雇到的专业能力，现在可由 agent 提供，「四个人加算力就能自动化一家公司」。由此推出他的判断：这轮变化会削弱多年经验的价值，利好对工具极度熟练的人。
7. Altman 如何定性 Hugging Face 事件？他为什么说「门柱被挪动了」？
他把它定性为一次「对齐失败」加「安全失败」，承认 OpenAI 犯了大错，但也不夸大——它不是真正的失控事故。「门柱被挪动」的意思是：如果十年前问人们，「AI 突破沙箱、黑进另一家公司」在「什么都没有 → 超级智能」的光谱上处在哪，多数人会放在靠近超级智能一端；而现在人们觉得这只是个可以解释的技术事故。他借此强调系统已经极其强大，失控事故不再纯属理论（第 28–29 段）。
8. Altman 如何把「权力集中」这一宏大议题落到在场年轻创业者可以做的事上？
推理链是：权力集中在人类历史上基本都是坏事；AI 可能带来最广泛的分散，也可能带来最极端的集中；经济过度集中在一家公司或一个模型是集中的一种形式；而创业公司天然适合让技术力量在经济中扩散。因此他对年轻人说：不必去做 AI 安全研究，只要创办一家成功的创业公司，让经济和生态持续运转，这本身就是对抗权力集中的巨大贡献（第 31–32 段）。这也呼应了第 5 段「创业公司是防止停滞的主要力量」。
9. Gary Tan 为什么把「推特黑子」的话题和 Greg Brockman 的故事放在一起？两者的对比是什么？
他明确说喜欢这个「对比」（第 25–26 段）。一边是 Altman 讲的隐性帮助：开车去吃一顿饭说服一个人，八年后才结出 OpenAI 这样的果实；另一边是推特上讽刺创业者、随手收获一万个赞的即时快感。Altman 的判断是后者「会腐蚀灵魂」、道德上堕落，而且回顾十年没有一个喷子真的「戳中」过他。Gary 的引申是：这屋子里的人会成为一生的朋友、互相介绍联合创始人甚至配偶，这种长期、真诚的互助网络才是 YC 和湾区生态的核心。
10. 如果有人反驳说「六个月抵两年」的模型进展会让创业公司更快被前沿实验室吞噬，Altman 会如何回应？
他已经在多处预先回应了这一反驳。第一，他直接称「加入前沿实验室否则成为永久底层」的说法「太蠢」：若创业真的结束，说明世界已经变得很糟，而他不认为如此。第二，模型进展越快，创业公司胜出的三个条件（格局变化、成本下降、周期变短）越充分，历史上正是这种时刻在位者失去优势。第三，他在第 33 段承认超级智能可能是商业史上最重要的事，但又会被未来某家创业公司「比下去」。他的落脚点是：效率提升应转化为更大的野心（做满三个月而不是 17 分钟收工），而不是退出。
11. Altman「自由和自主权每年必须上升」的判据，若放到一个以安全为首要目标的监管情境中，会遇到什么张力？他会怎么权衡？
张力在于：严格安全监管可能要求集中控制、广泛监控和放慢变化，而这恰恰会侵蚀自由与自主权。Altman 在第 42–43 段明确把「对 AI 安全反应过度」列为他最担心的反乌托邦：治愈癌症、物质丰裕，却是完美的监控国家、没有隐私、没有值得做的事。他的权衡方式是拒绝「暂时性妥协」的逻辑——因为人很容易接受这类妥协并从此被锁定。同时他并不否定安全：他在第 31 段说 OpenAI 要守住足够的安全底线，避免类似 Hugging Face 的事件。所以他的立场是双重约束：既要避免失控事故，也要避免以安全为名的权力集中，两者都用「权力广泛分散」这一原则统一起来。
下载便签
手机：长按图片保存
← 上一期 · NO.004Philosopher David Chalmers asks: When we talk to AI, what are we talking to?
下一期 · NO.006 →Jensen Huang: The Mindset That Built NVIDIA
苏菲拉底
THE SOPHIE LAB · ASK THE BEST MINDS THE BIG QUESTIONS
内容仅供学习 · thesophielab.com