Sam Altman思想档案馆
Sam Altman 肖像

文献 访谈 alt-025 机器译文 · 未审校

Lex Fridman 播客 #419 — Sam Altman:OpenAI、GPT-5、Sora、董事会风波、Elon Musk、Ilya、权力与 AGI

Lex Fridman Podcast #419 — Sam Altman: OpenAI, GPT-5, Sora, Board Saga, Elon Musk, Ilya, Power and AGI

来源 Lex Fridman · 英文原文出处

STATE2 中文可读 · 机器译文,未经人工审校 证据等级 B
Lex Fridman 播客 #419 — Sam Altman:OpenAI、GPT-5、Sora、董事会风波、Elon Musk、Ilya、权力与 AGI 配图
YouTube 官方视频封面 · 来源页
中文为默认阅读视图

机器译文 本页中文为机器逐句翻译(qwen3.8-flash),未经人工审校,仅供阅读参考;引用与考据请以英文原文为准。 其中 9 段含 AI 修订(仍非人工审校)。 逐句机译与英文原文 1:1 存档(1319 段);未经人工审校。

Sam Altman 我认为,算力将成为未来的货币。 I think compute is going to be the currency of the future.

Sam Altman 我认为它也许会成为世界上最珍贵的商品。 I think it’ll be maybe the most precious commodity in the world.

Sam Altman 我预计,到本十年末,甚至可能更早一些,我们会拥有相当强大的系统,让我们看了会说:「哇,这真的太了不起了。」通往 AGI 的道路将是一场巨大的权力斗争。 I expect that by the end of this decade, and possibly somewhat sooner than that, we will have quite capable systems that we look at and say, “Wow, that’s really remarkable.” The road to AGI should be a giant power struggle.

Sam Altman 我预计情况会是这样。 I expect that to be the case.

Lex Fridman 谁先造出 AGI,谁就获得巨大的权力。 Whoever builds AGI first gets a lot of power.

Lex Fridman 你相信自己能驾驭这么大的权力吗? Do you trust yourself with that much power?

Lex Fridman 以下是我与 Sam Altman 的对话,这是他第二次来到这档播客。 The following is a conversation with Sam Altman, his second time on the podcast.

Lex Fridman 他是 OpenAI 的 CEO,这家公司做出了 GPT-4、ChatGPT、Sora,也许终有一天,它会成为打造出 AGI 的那家公司。 He is the CEO of OpenAI, the company behind GPT-4, ChaTGPT, Sora, and perhaps one day the very company that will build AGI.

Interviewer 这里是 Lex Fridman 播客。 This is The Lex Fridman Podcast.

Lex Fridman 若要支持我们,请查看简介中的赞助商。 To support it, please check out our sponsors in the description.

Lex Fridman 朋友们,下面有请 Sam Altman。 And now, dear friends, here’s Sam Altman.

Lex Fridman 给我讲讲 OpenAI 董事会那场风波吧,它始于 11 月 16 日星期四,也许对你来说是 11 月 17 日星期五。 Take me through the OpenAI board saga that started on Thursday, November 16th, maybe Friday, November 17th for you.

Sam Altman 那绝对是我职业生涯里最痛苦的一段经历,混乱、屈辱、令人心碎,还有一堆别的负面感受。 That was definitely the most painful professional experience of my life, and chaotic and shameful and upsetting and a bunch of other negative things.

Sam Altman 但它也有美好的地方,只可惜当时肾上腺素飙升,我没能停下来好好体会那些美好。 There were great things about it too, and I wish it had not been in such an adrenaline rush that I wasn’t able to stop and appreciate them at the time.

Sam Altman 后来我看到了自己那时发的一条旧推文。 But I came across this old tweet of mine or this tweet of mine from that time period.

Sam Altman 那感觉就像亲眼看着自己的悼词被念出来——听大家说关于你的种种好话,还有我所爱与在乎的人给予的难以置信的支持。 It was like going your own eulogy, watching people say all these great things about you, and just unbelievable support from people I love and care about.

Sam Altman 那真的很好,真的很好。 That was really nice, really nice.

Sam Altman 整个那个周末,除了一个很大的例外,我感受到的多是爱,很少是恨,尽管当时我觉得完全不知道发生了什么、接下来会发生什么,感觉糟透了。 That whole weekend, with one big exception, I felt like a great deal of love and very little hate, even though it felt like I have no idea what’s happening and what’s going to happen here and this feels really bad.

Sam Altman 而且确实有些时刻,我以为这会是 AI 安全史上最糟糕的事情之一。 And there were definitely times I thought it was going to be one of the worst things to ever happen for AI safety.

Sam Altman 嗯,我也觉得,它发生得相对早,反而是好事。 Well, I also think I’m happy that it happened relatively early.

Sam Altman 我一直觉得,从 OpenAI 创立到我们造出 AGI 之间,总会发生某件疯狂的、爆炸性的事,不过也许还有更疯狂、更爆炸性的事在后面。 I thought at some point between when OpenAI started and when we created AGI, there was going to be something crazy and explosive that happened, but there may be more crazy and explosive things still to happen.

Sam Altman 我认为它仍然帮我们建立起了一些韧性,让我们能从容面对未来更多的挑战。 It still, I think, helped us build up some resilience and be ready for more challenges in the future.

Lex Fridman 但你预感自己会经历的那种事,是某种权力斗争? But the thing you had a sense that you would experience is some kind of power struggle?

Sam Altman 通往 AGI 的道路将是一场巨大的权力斗争。 The road to AGI should be a giant power struggle.

Sam Altman 这个世界应该……嗯,不是「应该」。 The world should… Well, not should.

Sam Altman 我预计情况会是这样。 I expect that to be the case.

Lex Fridman 所以你必须经历这些。就像你说的,尽可能高频地迭代:如何设计董事会结构,如何组织公司,如何选择与你共事的人,如何就这一切进行沟通,从而尽可能化解这场权力斗争。 And so you have to go through that, like you said, iterate as often as possible in figuring out how to have a board structure, how to have organization, how to have the kind of people that you’re working with, how to communicate all that in order to deescalate the power struggle as much as possible.

Sam Altman 对。 Yeah.

Lex Fridman 让它平息下来。 Pacify it.

Sam Altman 但到了现在,它感觉已经是过去的事了——非常不愉快、非常艰难、非常痛苦,但我们已经回到工作中,事情太多、节奏太紧张,我没花太多时间去想它。 But at this point, it feels like something that was in the past that was really unpleasant and really difficult and painful, but we’re back to work and things are so busy and so intense that I don’t spend a lot of time thinking about it.

Sam Altman 事后有过一段时间,在那之后的大约一个月、也许 45 天里,我处在一种恍惚状态,就这么浑浑噩噩地一天天漂着。 There was a time after, there was this fugue state for the month after, maybe 45 days after, that I was just drifting through the days.

Sam Altman 我整个人都是懵的。 I was so out of it.

Sam Altman 情绪非常低落。 I was feeling so down.

Lex Fridman 纯粹是个人心理层面的? Just on a personal, psychological level?

Sam Altman 对。 Yeah.

Sam Altman 真的很痛苦,而且在这种状态下还得继续经营 OpenAI,太难了。 Really painful, and hard to have to keep running OpenAI in the middle of that.

Sam Altman 我当时只想钻进一个山洞,安静地恢复一阵子。 I just wanted to crawl into a cave and recover for a while.

Sam Altman 但现在我们就是重新投入使命的工作中了。 But now it’s like we’re just back to working on the mission.

Lex Fridman 不过,回头复盘依然是有用的——反思董事会结构、权力格局、公司是如何运营的,以及研究与产品开发、资金之间的张力,诸如此类。这样一来,你这个最有可能造出 AGI 的人,未来就能以更有条理、更少戏剧性的方式做成这件事。 Well, it’s still useful to go back there and reflect on board structures, on power dynamics, on how companies are run, the tension between research and product development and money and all this kind of stuff so that you, who have a very high potential of building AGI, would do so in a slightly more organized, less dramatic way in the future.

Lex Fridman 所以这背后有价值:既有你作为领导者的个人心理层面,也有董事会结构和这一团乱麻的种种问题。 So there’s value there to go, both the personal psychological aspects of you as a leader, and also just the board structure and all this messy stuff.

Sam Altman 我确实从中学到了很多关于结构和激励机制的知识,学到我们需要什么样的董事会。 I definitely learned a lot about structure and incentives and what we need out of a board.

Sam Altman 我认为,从某种意义上说,这件事发生在现在是有价值的。 And I think that it is valuable that this happened now in some sense.

Sam Altman 我觉得这大概不会是 OpenAI 最后一次高压时刻,但那确实是一个高压到极点的时刻。 I think this is probably not the last high-stress moment of OpenAI, but it was quite a high-stress moment.

Sam Altman 我的公司差点就被摧毁了。 My company very nearly got destroyed.

Sam Altman 对于 AGI 还需要我们把哪些事情做对,我们思考了很多很多别的方面;但如何打造一个有韧性的组织,如何搭建一个能承受巨大外部压力的结构——随着我们越来越近,我预期这种压力会越来越多——我认为这超级重要。 And we think a lot about many of the other things we’ve got to get right for AGI, but thinking about how to build a resilient org and how to build a structure that will stand up to a lot of pressure in the world, which I expect more and more as we get closer, I think that’s super important.

Lex Fridman 你知道董事会的审议过程有多深入、多严谨吗? Do you have a sense of how deep and rigorous the deliberation process by the board was?

Lex Fridman 你能不能讲讲这类局面中涉及的人性 dynamics(人的互动 dynamics)? Can you shine some light on just human dynamics involved in situations like this?

Lex Fridman 是不是就聊了几次,局面突然升级,然后就变成「我们为什么不开掉 Sam」那种? Was it just a few conversations and all of a sudden it escalates and why don’t we fire Sam kind of thing?

Sam Altman 我认为董事会成员总体上都是善意的。我相信,在让人倍感时间压力的紧张局面下,人会理解偏差,做出次优的决策。 I think the board members are well-meaning people on the whole, and I believe that in stressful situations where people feel time pressure or whatever, people understand and make suboptimal decisions.

Sam Altman 我觉得 OpenAI 未来的一个挑战是:我们必须拥有一个善于在压力下运转的董事会和团队。 And I think one of the challenges for OpenAI will be we’re going to have to have a board and a team that are good at operating under pressure.

Lex Fridman 你觉得董事会的权力太大了吗? Do you think the board had too much power?

Sam Altman 我认为董事会本来就应该拥有很大的权力。但我们看到的一点是,在大多数公司结构里,董事会通常要对股东负责。 I think boards are supposed to have a lot of power, but one of the things that we did see is in most corporate structures, boards are usually answerable to shareholders.

Sam Altman 有时人们持有超级投票权股份之类的东西。 Sometimes people have super voting shares or whatever.

Sam Altman 在这种情况下——我觉得我们的结构中本该多考虑一点的是:非营利组织的董事会,除非你设置其他规则,否则拥有相当大的权力。 In this case, and I think one of the things with our structure that we maybe should have thought about more than we did is that the board of a nonprofit has, unless you put other rules in place, quite a lot of power.

Sam Altman 他们实际上只对自己负责,不对任何人负责。 They don’t really answer to anyone but themselves.

Sam Altman 这有好的一面,但我们真正希望的,是让 OpenAI 的董事会尽可能对整个全世界负责——只要这在现实中行得通。 And there’s ways in which that’s good, but what we’d really like is for the board of OpenAI to answer to the world as a whole, as much as that’s a practical thing.

Lex Fridman 所以新董事会已经公布了。 So there’s a new board announced.

Sam Altman 对。 Yeah.

Lex Fridman 我猜先是一个更小的新董事会,现在又有一个最终的董事会? There’s I guess a new smaller board at first, and now there’s a new final board?

Sam Altman 还不是最终董事会。 Not a final board yet.

Sam Altman 我们增加了一些成员。 We’ve added some.

Sam Altman 还会再增加。 We’ll add more.

Lex Fridman 增加了一些。 Added some.

Lex Fridman 好的。 Okay.

Lex Fridman 新的董事会修复了旧董事会哪里的问题? What is fixed in the new one that was perhaps broken in the previous one?

Sam Altman 旧董事会在大约一年时间里不断缩小。 The old board got smaller over the course of about a year.

Sam Altman 从九人减到六人,然后我们在该增补谁的问题上达不成一致。 It was nine and then it went down to six, and then we couldn’t agree on who to add.

Sam Altman 我还认为旧董事会里缺少有经验的董事,而 OpenAI 的新董事很多都拥有丰富的董事经验。 And the board also I think didn’t have a lot of experienced board members, and a lot of the new board members at OpenAI have just have more experience as board members.

Sam Altman 我觉得这会有帮助。 I think that’ll help.

Lex Fridman 对新董事会增补的一些人,外界有批评声音。 It’s been criticized, some of the people that are added to the board.

Lex Fridman 我听到很多人批评增补 Larry Summers 这个决定。 I heard a lot of people criticizing the addition of Larry Summers, for example.

Lex Fridman 挑选董事的流程是怎样的? What’s the process of selecting the board?

Lex Fridman 里面都涉及什么? What’s involved in that?

Sam Altman Brett 和 Larry 是在那个极其紧张的周末里、在风口浪尖上定下来的。那个周末真像坐过山车。 So Brett and Larry were decided in the heat of the moment over this very tense weekend, and that weekend was a real rollercoaster.

Sam Altman 大起大落。 It was a lot of ups and downs.

Sam Altman 我们试图找到新的董事会成员,既要让高管团队认可,也要让旧董事会成员觉得合理。 And we were trying to agree on new board members that both the executive team here and the old board members felt would be reasonable.

Sam Altman Larry 其实是旧董事会成员提出的建议之一。 Larry was actually one of their suggestions, the old board members.

Sam Altman Brett 我记得我甚至在那个周末之前就提议过,但他当时很忙不愿意来;后来我们在 [inaudible 00:09:22] 时段真的很需要帮助。 Brett, I think I had even previous to that weekend suggested, but he was busy and didn’t want to do it, and then we really needed help in [inaudible 00:09:22].

Sam Altman 我们也谈了很多其他人,但我觉得如果我要回来,就需要新的董事会成员。 We talked about a lot of other people too, but I felt like if I was going to come back, I needed new board members.

Sam Altman 我不认为能以同样的配置再和旧董事会共事。不过我们后来达成一致——我很感激 Adam 选择留下——我们考虑过各种配置,最终决定先组成一个三人董事会,于是必须在很短的时间里找到两位新董事。 I didn’t think I could work with the old board again in the same configuration, although we then decided, and I’m grateful that Adam would stay, but we considered various configurations, decided we wanted to get to a board of three and had to find two new board members over the course of a short period of time.

Sam Altman 所以那些是在战火中仓促定下的……你在战场上做决定。 So those were decided honestly without… You do that on the battlefield.

Sam Altman 那时你没有时间去设计一套严谨的流程。 You don’t have time to design a rigorous process then.

Sam Altman 至于那之后以及未来将要增补的新董事,我们有一套认为重要的董事会标准,希望董事会具备不同的专长。 For new board members since, and new board members we’ll add going forward, we have some criteria that we think are important for the board to have, different expertise that we want the board to have.

Sam Altman 这与招聘高管不同——高管只需要把一个角色做好,而董事会要把治理与深思熟虑这整个职能做好。Brett 有句话我很喜欢:我们要成批次地(in slates,成批次地)选聘董事,而不是一个一个地单独招。 Unlike hiring an executive where you need them to do one role well, the board needs to do a whole role of governance and thoughtfulness well, and so, one thing that Brett says which I really like is that we want to hire board members in slates, not as individuals one at a time.

Sam Altman 我们要找的是一批能带来非营利组织经验、经营公司的经验、良好法律与治理经验的人,这就是我们努力优化的方向。 And thinking about a group of people that will bring nonprofit expertise, expertise at running companies, good legal and governance expertise, that’s what we’ve tried to optimize for.

Lex Fridman 那么技术素养对董事个人重要吗? So is technical savvy important for the individual board members?

Sam Altman 不是每位董事都需要,但其中某些席位必须有。 Not for every board member, but for certainly some you need that.

Sam Altman 这是董事会职能的一部分。 That’s part of what the board needs to do.

Lex Fridman 关于 OpenAI,人们大概不了解、我自己也不完全了解的一点,是经营这家公司的所有细节。 The interesting thing that people probably don’t understand about OpenAI, I certainly don’t, is all the details of running the business.

Lex Fridman 当他们想到董事会,鉴于这场戏剧性风波,他们想到的是你。 When they think about the board, given the drama, they think about you.

Lex Fridman 他们想的是:如果你们抵达了 AGI,或者做出了某些影响力惊人的产品并加以部署,那时与董事会的对话会是怎样的? They think about if you reach AGI or you reach some of these incredibly impactful products and you build them and deploy them, what’s the conversation with the board like?

Lex Fridman 他们会想:好,在那种局面下,什么样的阵容才适合做审议? And they think, all right, what’s the right squad to have in that kind of situation to deliberate?

Sam Altman 你看,我认为董事会里一定要有若干技术专家。 Look, I think you definitely need some technical experts there.

Sam Altman 然后你需要有人来问:「我们怎样才能以最大程度帮助世界各地人们的方式来部署它?」还需要有完全不同视角的人。 And then you need some people who are like, “How can we deploy this in a way that will help people in the world the most?” And people who have a very different perspective.

Sam Altman 我觉得你或我都可能犯的一个错误,是以为只有技术理解才重要。那当然是你希望董事会具备的对话的一部分,但还有太多关于它将如何影响社会和人们生活的层面,你也希望在那里得到充分的代表。 I think a mistake that you or I might make is to think that only the technical understanding matters, and that’s definitely part of the conversation you want that board to have, but there’s a lot more about how that’s going to just impact society and people’s lives that you really want represented in there too.

Lex Fridman 你们是看候选人的履历,还是只靠交谈? Are you looking at the track record of people or you’re just having conversations?

Sam Altman 履历是非常重要的。 Track record is a big deal.

Sam Altman 你当然会有很多谈话,但有些角色我会完全不看履历,只看斜率(slope,斜率),忽略截距。 You of course have a lot of conversations, but there are some roles where I totally ignore track record and just look at slope, ignore the Y-intercept.

Lex Fridman 谢谢你。 Thank you.

Lex Fridman 感谢你把这件事讲得这么数学化,为了观众。 Thank you for making it mathematical for the audience.

Sam Altman 但对董事这个位置,我更看重截距(Y-intercept,Y 轴截距)。 For a board member, I do care much more about the Y-intercept.

Sam Altman 我觉得履历在这里确实说明一些很深的东西,而经验是非常难以替代的。 I think there is something deep to say about track record there, and experience is something’s very hard to replace.

Lex Fridman 你是想对履历拟合一个多项式函数还是指数函数? Do you try to fit a polynomial function or exponential one to the track record?

Sam Altman 这个类比就没法推得那么远了。 That analogy doesn’t carry that far.

Lex Fridman 好吧。 All right.

Lex Fridman 你提到了那个周末的一些低谷。 You mentioned some of the low points that weekend.

Lex Fridman 对你来说,心理上的低谷有哪些? What were some of the low points psychologically for you?

Lex Fridman 你有没有想过跑去亚马逊丛林,嗑死藤水(ayahuasca,死藤水),然后从此人间蒸发? Did you consider going to the Amazon jungle and just taking ayahuasca and disappearing forever?

Sam Altman 那是一段非常糟糕的时期。 It was a very bad period of time.

Sam Altman 但也有非常美好的高光时刻。 There were great high points too.

Sam Altman 我的手机一直响个不停,全是善意的消息——有每天共事的人,也有十年没联系过的人。 My phone was just nonstop blowing up with nice messages from people I worked with every day, people I hadn’t talked to in a decade.

Sam Altman 我没能像应有的那样去珍惜这些,因为我正深陷这场救火之战,但那真的很好。 I didn’t get to appreciate that as much as I should have because I was just in the middle of this firefight, but that was really nice.

Sam Altman 但总体而言,那是非常痛苦的一个周末。 But on the whole, it was a very painful weekend.

Sam Altman 那场斗争在相当出乎意料的程度上是公开进行的,这让我极度疲惫,远超我的预期。 It was like a battle fought in public to a surprising degree, and that was extremely exhausting to me, much more than I expected.

Sam Altman 我觉得争斗通常都让人疲惫,但这一次真的是。 I think fights are generally exhausting, but this one really was.

Sam Altman 董事会是在周五下午做的这件事。 The board did this Friday afternoon.

Sam Altman 我真的几乎得不到任何解释,但我也就想:好吧,董事会有权这么做,那我花点时间想想自己想怎么做,但我试着在这里苦中寻福,寻找这件坏事里的转机(blessing in disguise,隐藏的福气)。 I really couldn’t get much in the way of answers, but I also was just like, well, the board gets to do this, so I’m going to think for a little bit about what I want to do, but I’ll try to find the blessing in disguise here.

Sam Altman 我当时想,我在 OpenAI 现在的工作,或者说原本的工作,是经营一家到这个阶段已有一定规模的公司。 And I was like, well, my current job at OpenAI is, or it was, to run a decently sized company at this point.

Sam Altman 而我最喜欢的事,一直是和研究人员一起工作。 And the thing I’d always liked the most was just getting to work with the researchers.

Sam Altman 我就想:好啊,我可以纯粹去做一个非常聚焦的 AGI 研究项目。 And I was like, yeah, I can just go do a very focused AGI research effort.

Interviewer 我对此兴奋起来。 And I got excited about that.

Sam Altman 当时我压根没想过,这一切还有可能被逆转。 Didn’t even occur to me at the time possibly that this was all going to get undone.

Interviewer 那可是周五下午。 This was Friday afternoon.

Lex Fridman 所以你接受了这个孩子的死亡—— So you’ve accepted the death of this-

Sam Altman 非常快。 Very quickly.

Sam Altman 非常快。 Very quickly.

Sam Altman 我经历了一小段困惑和愤怒的时期,但很快就过去了,非常快。 I went through a little period of confusion and rage, but very quickly, quickly.

Sam Altman 到周五晚上,我已经在和人聊下一步该做什么,而且我对此感到兴奋。 And by Friday night, I was talking to people about what was going to be next, and I was excited about that.

Sam Altman 我想是周五傍晚我第一次听到这里的高管团队表态:「嘿,我们要抗争到底。」然后我去睡觉,心里仍然是:好,很兴奋。 I think it was Friday evening for the first time that I heard from the exec team here, which is like, “Hey, we’re going to fight this.” and then I went to bed just still being like, okay, excited.

Sam Altman 向前走。 Onward.

Lex Fridman 你睡着了吗? Were you able to sleep?

Sam Altman 没睡多少。 Not a lot.

Sam Altman 奇怪的事情之一是:有整整四天半,我几乎没怎么睡,也没怎么吃,却仍然有惊人的精力。 One of the weird things was there was this period of four and a half days where I didn’t sleep much, didn’t eat much, and still had a surprising amount of energy.

Sam Altman 你在战争时期会学到关于肾上腺素的奇怪一课。 You learn a weird thing about adrenaline in wartime.

Lex Fridman 所以你接受了 OpenAI 这个孩子的死亡。 So you accepted the death of this baby, OpenAI.

Sam Altman 而我对新事物感到兴奋。 And I was excited for the new thing.

Sam Altman 我就是想:「好吧,这很疯狂,但那又怎样。」 I was just like, “Okay, this was crazy, but whatever.”

Lex Fridman 这是很好的心理防御机制。 It’s a very good coping mechanism.

Sam Altman 然后周六早上,两位董事会成员打来电话说:「嘿,我们无意让局面动荡。 And then Saturday morning, two of the board members called and said, “Hey, we didn’t mean to destabilize things.

Sam Altman 我们不想毁掉这里的太多价值。 We don’t want to store a lot of value here.

Sam Altman 我们能谈谈你回来的事吗?」我本能地不想答应,但多想了想,我觉得:我真的很在乎这里的人,合伙人、股东们。 Can we talk about you coming back?” And I immediately didn’t want to do that, but I thought a little more and I was like, well, I really care about the people here, the partners, shareholders.

Sam Altman 我爱这家公司。 I love this company.

Sam Altman 于是我考虑了一下,说:「好吧,但这些是我需要的条件。」然后,最痛苦的阶段来了——整个那个周末,我一直在想、也不断被告知,不光是我,这里整个团队都一直在想:我们拼命想让 OpenAI 保持稳定,而全世界都在试图把它撕碎,有人到处挖人,诸如此类。 And so I thought about it and I was like, “Well, okay, but here’s the stuff I would need.” And then the most painful time of all was over the course of that weekend, I kept thinking and being told, and not just me, the whole team here kept thinking, well, we were trying to keep OpenAI stabilized while the whole world was trying to break it apart, people trying to recruit whatever.

Sam Altman 我们不断被告知:好了,快谈妥了。 We kept being told, all right, we’re almost done.

Interviewer 快谈妥了。 We’re almost done.

Sam Altman 再需要一点时间。 We just need a little bit more time.

Sam Altman 那是一种非常混乱的状态。 And it was this very confusing state.

Sam Altman 然后到周日晚上——同样地,我每隔几个小时就以为要谈完了,我们快要找到让我回归、让一切恢复原样的办法了—— And then Sunday evening when, again, every few hours I expected that we were going to be done and we’re going to figure out a way for me to return and things to go back to how they were.

Sam Altman 董事会却任命了一位新的临时 CEO。我当时就想:这感觉糟透了。 The board then appointed a new interim CEO, and then I was like, that feels really bad.

Sam Altman 那是整件事的最低谷。 That was the low point of the whole thing.

Interviewer 我告诉你一件事。 I’ll tell you something.

Sam Altman 那很痛苦,但整个周末我感受到很多爱。 It felt very painful, but I felt a lot of love that whole weekend.

Sam Altman 除了周日晚上那一刻,我不会把我的情绪描述为愤怒或仇恨,我从人们身上、并对人们感受到了很多爱。 Other than that one moment Sunday night, I would not characterize my emotions as anger or hate, but I felt a lot of love from people, towards people.

Sam Altman 很痛苦,但那个周末的主导情绪是爱,不是恨。 It was painful, but the dominant emotion of the weekend was love, not hate.

Lex Fridman 你高度评价过 Mira Murati,说她在关键时刻、尤其如你在推文中所说、在那些安静但要紧的时刻帮了大忙。 You’ve spoken highly of Mira Murati, that she helped especially, as you put in the tweet, in the quiet moments when it counts.

Lex Fridman 我们也许可以稍微岔开话题。 Perhaps we could take a bit of a tangent.

Lex Fridman 你欣赏 Mira 什么? What do you admire about Mira?

Sam Altman 她在一片混乱中的那个周末表现得非常出色,但人们往往是在危机时刻看到领导者,无论好坏。 Well, she did a great job during that weekend in a lot of chaos, but people often see leaders in the crisis moments, good or bad.

Sam Altman 而我在领导者身上非常看重的一点是:在一个无聊的周二早上 9 点 46 分,在日常枯燥的琐碎中,人们如何表现。 But a thing I really value in leaders is how people act on a boring Tuesday at 9:46 in the morning and in just the normal drudgery of the day-to-day.

Sam Altman 一个人如何出席一场会议,他做的决策质量如何。 How someone shows up in a meeting, the quality of the decisions they make.

Sam Altman 这就是我说的「安静的时刻」的意思。 That was what I meant about the quiet moments.

Lex Fridman 意思是,大部分工作是在一天一天、一场会议接一场会议中完成的。 Meaning most of the work is done on a day-by-day, in meeting-by-meeting.

Lex Fridman 就在那儿,做出好的决策。 Just be present and make great decisions.

Sam Altman 是啊。 Yeah.

Sam Altman 你看,大家——我理解——过去 20 分钟一直想聊的,是这个戏剧性极强的周末,但那并不是 OpenAI 的真正意义。 Look, what you have wanted to spend the last 20 minutes about, and I understand, is this one very dramatic weekend, but that’s not really what OpenAI is about.

Sam Altman OpenAI 真正的意义,是另外那七年。 OpenAI is really about the other seven years.

Lex Fridman 嗯,是啊。 Well, yeah.

Lex Fridman 人类文明并不是关于纳粹德国入侵苏联的,但人们仍然聚焦于此。 Human civilization is not about the invasion of the Soviet Union by Nazi Germany, but still that’s something people focus on.

Sam Altman 这非常可以理解。 Very understandable.

Lex Fridman 它让我们洞察人性,洞察人性的极端,而人类文明的一些破坏与一些胜利或许就发生在那些时刻,所以它有启示意义。 It gives us an insight into human nature, the extremes of human nature, and perhaps some of the damage in some of the triumphs of human civilization can happen in those moments, so it’s illustrative.

Lex Fridman 让我问问你关于 Ilya 的事。 Let me ask you about Ilya.

Lex Fridman 他是不是被扣押在某个秘密核设施里? Is he being held hostage in a secret nuclear facility?

Sam Altman 没有。 No.

Lex Fridman 那普通的秘密设施呢? What about a regular secret facility?

Sam Altman 也没有。 No.

Lex Fridman 那不秘密的核设施呢? What about a nuclear non-secret facility?

Sam Altman 那也没有。 Neither.

Sam Altman 也不是那个。 Not that either.

Lex Fridman 这早晚要成为一个 meme。 This is becoming a meme at some point.

Lex Fridman 你认识 Ilya 很久了。 You’ve known Ilya for a long time.

Lex Fridman 他显然也是这场董事会风波的一部分。 He was obviously part of this drama with the board and all that kind of stuff.

Lex Fridman 你现在和他的关系如何? What’s your relationship with him now?

Sam Altman 我爱 Ilya。 I love Ilya.

Sam Altman 我非常尊重 Ilya。 I have tremendous respect for Ilya.

Sam Altman 关于他的计划,我现在没有什么可以透露的。 I don’t have anything I can say about his plans right now.

Sam Altman 这个问题得问他本人,但我真心希望我们能一起共事——至少在我职业生涯的剩余时间里。 That’s a question for him, but I really hope we work together for certainly the rest of my career.

Sam Altman 他比我小一点。 He’s a little bit younger than me.

Sam Altman 也许他会工作得更久一点。 Maybe he works a little bit longer.

Lex Fridman 网上有个梗,说他看见了什么东西——也许他看见了 AGI——这让他内心充满了忧虑。 There’s a meme that he saw something, like he maybe saw AGI and that gave him a lot of worry internally.

Lex Fridman Ilya 看见了什么? What did Ilya see?

Sam Altman Ilya 并没有看见 AGI。 Ilya has not seen AGI.

Sam Altman 我们谁都没见过 AGI。 None of us have seen AGI.

Sam Altman 我们还没有造出 AGI。 We’ve not built AGI.

Sam Altman 我确实认为,Ilya 身上让我非常欣赏的众多品质之一,是他非常严肃地对待 AGI 和安全问题——广义上的安全问题,包括这件事将给社会带来的影响。 I do think one of the many things that I really love about Ilya is he takes AGI and the safety concerns, broadly speaking, including things like the impact this is going to have on society, very seriously.

Sam Altman 随着我们不断取得重大进展,Ilya 是我过去几年里花时间最多探讨「这一切意味着什么、我们需要做什么才能把它做对、如何让使命成功」的人之一。 And as we continue to make significant progress, Ilya is one of the people that I’ve spent the most time over the last couple of years talking about what this is going to mean, what we need to do to ensure we get it right, to ensure that we succeed at the mission.

Sam Altman 所以 Ilya 没有见过 AGI,但在他为「确保我们做对这件事」所付出的思考和忧虑方面,他是人类的骄傲。 So Ilya did not see AGI, but Ilya is a credit to humanity in terms of how much he thinks and worries about making sure we get this right.

Lex Fridman 我过去和他有过很多对话。 I’ve had a bunch of conversation with him in the past.

Lex Fridman 我觉得他谈技术时,总是在做那种长远思考。 I think when he talks about technology, he’s always doing this long-term thinking type of thing.

Lex Fridman 他不考虑一年后这会变成什么样。 So he is not thinking about what this is going to be in a year.

Lex Fridman 他想的是十年后,从第一性原理出发思考:「好,如果这个能规模扩展,这里的根本规律是什么?」 He’s thinking about in 10 years, just thinking from first principles like, “Okay, if this scales, what are the fundamentals here?

Lex Fridman 「这会走向哪里?」而这就是他们思考其他所有安全问题之类事情的基础,这让他成为一个非常引人入胜的交谈对象。 Where’s this going?” And so that’s a foundation for them thinking about all the other safety concerns and all that kind of stuff, which makes him a really fascinating human to talk with.

Lex Fridman 你知道他为什么一直保持沉默吗? Do you have any idea why he’s been quiet?

Lex Fridman 是他正在进行某种自我反省吗? Is it he’s just doing some soul-searching?

Sam Altman 同样,我不想替 Ilya 发言。 Again, I don’t want to speak for Ilya.

Sam Altman 我觉得你应该直接去问他。 I think that you should ask him that.

Sam Altman 他绝对是一个善于思考的人。 He’s definitely a thoughtful guy.

Sam Altman 我觉得 Ilya 一直处在一种非常好的自我探寻状态。 I think Ilya is always on a soul search in a really good way.

Lex Fridman 是的。 Yes.

Lex Fridman 嗯。 Yeah.

Lex Fridman 而且,他懂得沉默的力量。 Also, he appreciates the power of silence.

Lex Fridman 另外,我听说他可以很逗比,但我从没见过他那一面。 Also, I’m told he can be a silly guy, which I’ve never seen that side of him.

Sam Altman 当那种情况出现时,其实很可爱。 It’s very sweet when that happens.

Lex Fridman 我从未见识过逗比的 Ilya,但我也很期待看到。 I’ve never witnessed a silly Ilya, but I look forward to that as well.

Sam Altman 最近我和他一起参加一个晚宴,他在跟一只小奶狗玩,那时候他特别逗、特别可爱。 I was at a dinner party with him recently and he was playing with a puppy and he was in a very silly mood, very endearing.

Sam Altman 我当时就在想,天哪,这并不是世人最常看到的那一面 Ilya。 And I was thinking, oh man, this is not the side of Ilya that the world sees the most.

Lex Fridman 那我们把这段整个 saga 收个尾——你对董事会的结构感到满意吗? So just to wrap up this whole saga, are you feeling good about the board structure-

Sam Altman 满意。 Yes.

Lex Fridman ……对这一切以及它的走向感觉如何? … about all of this and where it’s moving?

Sam Altman 我对新董事会感觉非常好。 I feel great about the new board.

Sam Altman 就 OpenAI 的结构而言,董事会的任务之一就是审视它,看看我们可以在哪里让它更加稳健。 In terms of the structure of OpenAI, one of the board’s tasks is to look at that and see where we can make it more robust.

Sam Altman 我们想先安排好新的董事会成员,但在这整个过程中,我们显然吸取了关于结构的教训。 We wanted to get new board members in place first, but we clearly learned a lesson about structure throughout this process.

Sam Altman 我觉得我没有特别高深的东西可说。 I don’t have, I think, super deep things to say.

Sam Altman 那是一段疯狂且非常痛苦的经历。 It was a crazy, very painful experience.

Sam Altman 我觉得那是一场怪事叠加的完美风暴。 I think it was a perfect storm of weirdness.

Sam Altman 对我来说,这预演了当赌注越来越高时会发生什么,以及我们为什么需要稳健的治理结构、流程和人员。 It was a preview for me of what’s going to happen as the stakes get higher and higher and the need that we have robust governance structures and processes and people.

Sam Altman 我很高兴它发生在那个时间点,但经历这一切是令人震惊的痛苦。 I’m happy it happened when it did, but it was a shockingly painful thing to go through.

Lex Fridman 这让你对人更难以信任了吗? Did it make you be more hesitant in trusting people?

Sam Altman 是的。 Yes.

Lex Fridman 只是个人层面? Just on a personal level?

Sam Altman 是的。 Yes.

Sam Altman 我觉得我是那种极其信任别人的人。 I think I’m like an extremely trusting person.

Sam Altman 我的人生哲学一直是:不要为那些多疑的事烦恼。 I’ve always had a life philosophy of don’t worry about all of the paranoia.

Sam Altman 不要为边缘案例担心。 Don’t worry about the edge cases.

Sam Altman 你偶尔会吃点亏,以此换取放松警惕地生活。 You get a little bit screwed in exchange for getting to live with your guard down.

Sam Altman 而这次真的让我很震惊。 And this was so shocking to me.

Sam Altman 我毫无防备地被冲击了,这确实改变了我,我真的不喜欢这样,它确实改变了我的默认信任方式和为坏情况做规划的方式。 I was so caught off guard that it has definitely changed, and I really don’t like this, it’s definitely changed how I think about just default trust of people and planning for the bad scenarios.

Lex Fridman 你对此变得谨慎了。 You got to be careful with that.

Lex Fridman 你会担心自己变得太愤世嫉俗吗? Are you worried about becoming a little too cynical?

Sam Altman 我不担心自己变得愤世嫉俗。 I’m not worried about becoming too cynical.

Sam Altman 我觉得我恰恰是一个愤世嫉俗的反面,但我担心的不是变得少了一些默认信任——是变得不再默认信任别人。 I think I’m the extreme opposite of a cynical person, but I’m worried about just becoming less of a default trusting person.

Lex Fridman 其实我不确定以什么方式运作最好——对于一个正在开发 AGI 的人来说,是信任还是不信任。 I’m actually not sure which mode is best to operate in for a person who’s developing AGI, trusting or un-trusting.

Lex Fridman 这是一段有意思的旅程,你们正在经历。 It’s an interesting journey you’re on.

Lex Fridman 但在结构之外,我更关注人的层面。 But in terms of structure, see, I’m more interested on the human level.

Lex Fridman 你如何围绕自己安排那些正在打造很酷的东西、同时又能做出明智决策的人? How do you surround yourself with humans that are building cool shit, but also are making wise decisions?

Lex Fridman 因为随着你开始赚更多钱,这个东西拥有的权力越大,周围的人就越奇怪。 Because the more money you start making, the more power the thing has, the weirder people get.

Sam Altman 我想人们可以对董事会成员以及我当时应该给予多少信任、或者我本来应该如何处理做各种各样的评论。 I think you could make all kinds of comments about the board members and the level of trust I should have had there, or how I should have done things differently.

Sam Altman 但就这里的团队而言,我想你们得给我打个很好的分数。 But in terms of the team here, I think you’d have to give me a very good grade on that one.

Sam Altman 而我每天共事的人,我对他们怀着巨大的感激、信任和尊重,我认为与这样的人为伍真的非常重要。 And I have just enormous gratitude and trust and respect for the people that I work with every day, and I think being surrounded with people like that is really important.

Lex Fridman 我们的共同朋友 Elon 起诉了 OpenAI。 Our mutual friend Elon sued OpenAI.

Lex Fridman 你认为他批评的实质是什么? What to you is the essence of what he’s criticizing?

Lex Fridman 他在多大程度上有道理? To what degree does he have a point?

Lex Fridman 在多大程度上是错的? To what degree is he wrong?

Sam Altman 我不知道这到底是为了什么。 I don’t know what it’s really about.

Sam Altman 我们起步时只是以为自己会做一个研究实验室,完全不知道这项技术会走向何方。 We started off just thinking we were going to be a research lab and having no idea about how this technology was going to go.

Sam Altman 因为那只是七八年前,很难回去真正记住当时的样子,但那时大语言模型还不是一件大事。 Because it was only seven or eight years ago, it’s hard to go back and really remember what it was like then, but this is before language models were a big deal.

Sam Altman 那时我们完全不知道会有 API,或者靠出售聊天机器人的访问权来赚钱。 This was before we had any idea about an API or selling access to a chatbot.

Sam Altman 那时我们完全没想到我们会去做产品化。 It was before we had any idea we were going to productize at all.

Sam Altman 所以我们就是:「我们只想做研究,并不知道要拿它做什么。」我认为面对许多根本性的新事物,你都是在黑暗中摸索,做出一些假设,其中大多数最后都被证明是错的。 So we’re like, “We’re just going to try to do research and we don’t really know what we’re going to do with that.” I think with many fundamentally new things, you start fumbling through the dark and you make some assumptions, most of which turned out to be wrong.

Sam Altman 然后事情变得清晰:我们需要做不同的事,还需要多得多的资本。 And then it became clear that we were going to need to do different things and also have huge amounts more capital.

Sam Altman 于是我们说:「好吧,现有的结构不太适配这个。」 So we said, “Okay, well, the structure doesn’t quite work for that.

Sam Altman 我们怎么修补这个结构?然后一次次再打补丁,最终得到的东西至少可以说很让人挑眉。 How do we patch the structure?” And then you patch it again and patch it again and you end up with something that does look eyebrow-raising, to say the least.

Sam Altman 但我们是一步步走到这里的,我认为沿途每一步都做了当时合理的决定。 But we got here gradually with, I think, reasonable decisions at each point along the way.

Sam Altman 这并不意味着如果现在能带着先知(Oracle,全知预言者)回到过去我不会完全不同地做,但当时你得不到这个先知。 And it doesn’t mean I wouldn’t do it totally differently if we could go back now with an Oracle, but you don’t get the Oracle at the time.

Sam Altman 但话说回来,关于 Elon 在这里的真正动机是什么,我不知道。 But anyway, in terms of what Elon’s real motivations here are, I don’t know.

Lex Fridman 在你记忆所及的范围内,OpenAI 在那篇博客文章里给出的回应是什么? To the degree you remember, what was the response that OpenAI gave in the blog post?

Lex Fridman 你能概括一下吗? Can you summarize it?

Sam Altman 哦,我们只是说了 Elon 说了这一系列事情。 Oh, we just said Elon said this set of things.

Sam Altman 这是我们的描述,或者这不是我们的描述。 Here’s our characterization, or here’s not our characterization.

Sam Altman 这是事情经过的描述。 Here’s the characterization of how this went down.

Sam Altman 我们试着不带情绪,只是说「这就是历史」。 We tried to not make it emotional and just say, “Here’s the history.”

Lex Fridman 我确实认为 Elon 在这里有一点描述失实,就是你刚才提到的那一点——你们当时的不确定程度。 I do think there’s a degree of mischaracterization from Elon here about one of the points you just made, which is the degree of uncertainty you had at the time.

Lex Fridman 你们当时是一小群研究人员在疯狂地谈论 AGI,而所有人都在嘲笑这个想法。 You guys are a small group of researchers crazily talking about AGI when everybody’s laughing at that thought.

Sam Altman 没过多久之前,Elon 也在疯狂地谈论发射火箭,当时人们也在嘲笑那个想法,所以我想他对这个应该更有同理心。 It wasn’t that long ago Elon was crazily talking about launching rockets when people were laughing at that thought, so I think he’d have more empathy for this.

Lex Fridman 我确实认为这里有私人因素——这里有一个分裂,OpenAI 和很多优秀的人选择与 Elon 分道扬镳,所以这有私人—— I do think that there’s personal stuff here, that there was a split that OpenAI and a lot of amazing people here chose to part ways with Elon, so there’s a personal-

Sam Altman 是 Elon 选择分道扬镳的。 Elon chose to part ways.

Lex Fridman 你能确切描述一下吗? Can you describe that exactly?

Lex Fridman 「选择分道扬镳」这件事? The choosing to part ways?

Sam Altman 他认为 OpenAI 会失败。 He thought OpenAI was going to fail.

Sam Altman 他想要完全的控制权来扭转局面。 He wanted total control to turn it around.

Sam Altman 我们想沿着现在已经成为 OpenAI 的方向继续走下去。 We wanted to keep going in the direction that now has become OpenAI.

Sam Altman 他还希望 Tesla 能够建立一个 AGI 项目。 He also wanted Tesla to be able to build an AGI effort.

Sam Altman 在不同的时间点,他想把 OpenAI 变成一家他可以控制的营利性公司,或者把它并入 Tesla。 At various times, he wanted to make OpenAI into a for-profit company that he could have control of or have it merge with Tesla.

Sam Altman 我们不想这么做,他于是决定离开,这没什么。 We didn’t want to do that, and he decided to leave, which that’s fine.

Lex Fridman 所以你的意思是——这也是博客文章里提到的事情之一——他希望 OpenAI 基本上被 Tesla 收购,类似于与 Microsoft 的合作,或者可能比那个更极端。 So you’re saying, and that’s one of the things that the blog post says, is that he wanted OpenAI to be basically acquired by Tesla in the same way that, or maybe something similar or maybe something more dramatic than the partnership with Microsoft.

Sam Altman 我记得那个提议就是:是的,被 Tesla 收购,让 Tesla 完全控制。 My memory is the proposal was just like, yeah, get acquired by Tesla and have Tesla have full control over it.

Sam Altman 我很确定就是那样。 I’m pretty sure that’s what it was.

Lex Fridman 那么当时 OpenAI 里的「open」这个词对 Elon 意味着什么? So what does the word open in OpenAI mean to Elon at the time?

Lex Fridman Ilya 在邮件往来和这些事情里谈到过这点。 Ilya has talked about this in the email exchanges and all this kind of stuff.

Lex Fridman 对你来说当时意味着什么? What does it mean to you at the time?

Lex Fridman 现在对你意味着什么? What does it mean to you now?

Sam Altman 说到带着先知回过去,我会选一个不同的名字。 Speaking of going back with an Oracle, I’d pick a different name.

Sam Altman 我认为 OpenAI 所做的最重要的事情之一——比我们所有其他工作都重要——就是免费把强大的技术交到人们手中,作为公共产品。 One of the things that I think OpenAI is doing that is the most important of everything that we’re doing is putting powerful technology in the hands of people for free, as a public good.

Sam Altman 我们不在我们的—— We don’t run ads on our-

Sam Altman ……作为公共产品。 … as a public good.

Sam Altman 我们不在免费版上投放广告。 We don’t run ads on our free version.

Sam Altman 我们也不通过其他方式对它进行变现。 We don’t monetize it in other ways.

Sam Altman 我们只是说这是我们使命的一部分。 We just say it’s part of our mission.

Sam Altman 我们想把越来越强大的工具免费交到人们手中,让他们用起来。 We want to put increasingly powerful tools in the hands of people for free and get them to use them.

Sam Altman 我认为这种「开放」对我们的使命真的非常重要。 I think that kind of open is really important to our mission.

Sam Altman 我觉得,如果你给人们优秀的工具并教会他们用,或者甚至不教,他们也会自己弄明白,然后让他们用这些去为彼此建设一个不可思议的未来——这是件大事。 I think if you give people great tools and teach them to use them or don’t even teach them, they’ll figure it out, and let them go build an incredible future for each other with that, that’s a big deal.

Sam Altman 所以如果我们能持续把免费的或低成本的强大 AI 工具推向世界,我认为这对我们完成使命来说是意义重大的一件事。 So if we can keep putting free or low cost or free and low cost powerful AI tools out in the world, I think that’s a huge deal for how we fulfill the mission.

Sam Altman 开源与否,是的,我觉得我们应该开源一部分东西,不开源另一部分。 Open source or not, yeah, I think we should open source some stuff and not other stuff.

Sam Altman 它确实变成了一条宗教般的战线,细粒度很难被表达,但我认为细粒度才是正确答案。 It does become this religious battle line where nuance is hard to have, but I think nuance is the right answer.

Lex Fridman 他说:「把你的名字改成 CloseAI,我就撤诉。」我是说,关于名字这事会不会变成梗界的一个战场? So he said, “Change your name to CloseAI and I’ll drop the lawsuit.” I mean is it going to become this battleground in the land of memes about the name?

Sam Altman 我认为这说明 Elon 对这场诉讼是认真的,而且那句话在我看来是相当惊人的。 I think that speaks to the seriousness with which Elon means the lawsuit, and that’s like an astonishing thing to say, I think.

Lex Fridman 如果我错了请纠正我,但我不觉得这场诉讼在法律层面是认真的。 Maybe correct me if I’m wrong, but I don’t think the lawsuit is legally serious.

Lex Fridman 它更多是想就 AGI 的未来和当前领跑的公司表达一个观点。 It’s more to make a point about the future of AGI and the company that’s currently leading the way.

Sam Altman 你看,Grok 在人们指出这有点虚伪 之前根本没开源过任何东西,然后他就宣布 Grok 本周将开源一些东西。 Look, I mean Grok had not open sourced anything until people pointed out it was a little bit hypocritical and then he announced that Grok will open source things this week.

Sam Altman 我不认为开源与否是他真正在意的事。 I don’t think open source versus not is what this is really about for him.

Lex Fridman 好吧,我们会谈开源与否的问题。 Well, we will talk about open source and not.

Lex Fridman 我确实认为,也许批评竞争对手很棒。 I do think maybe criticizing the competition is great.

Lex Fridman 就是稍微说点对手的坏话,那很好。 Just talking a little shit, that’s great.

Lex Fridman 但友好的竞争,对比「我个人痛恨诉讼」这种。 But friendly competition versus like, “I personally hate lawsuits.”

Sam Altman 你看,我觉得整个这档子事与一个 builder 的身份不相称。 Look, I think this whole thing is unbecoming of a builder.

Sam Altman 我尊重 Elon,他是我们时代最伟大的建造者之一。 And I respect Elon as one of the great builders of our time.

Sam Altman 我知道他懂被黑子攻击是什么滋味,所以看到他做这样的事,我格外难过。 I know he knows what it’s like to have haters attack him and it makes me extra sad he’s doing it toss.

Lex Fridman 是的,他是有史以来最伟大的建造者之一,甚至可能是最伟大的。 Yeah, he’s one of the greatest builders of all time, potentially the greatest builder of all time.

Sam Altman 这让我难过。 It makes me sad.

Sam Altman 我认为这也让很多人难过。 And I think it makes a lot of people sad.

Sam Altman 有很多人长期以来非常敬仰他。 There’s a lot of people who’ve really looked up to him for a long time.

Sam Altman 我在某个采访里说过我怀念过去的 Elon,我收到的大量消息都是:「这句话恰恰概括了我的感受。」 I said in some interview or something that I missed the old Elon and the number of messages I got being like, “That exactly encapsulates how I feel.”

Lex Fridman 我觉得他就该赢。 I think he should just win.

Lex Fridman 他就该让 X 的 Grok 赢过 GPT,然后 GPT 又赢过 Grok,就是比谁更强,这对所有人都是美好的。 He should just make X Grok beat GPT and then GPT beats Grok and it’s just the competition and it’s beautiful for everybody.

Lex Fridman 但在开源问题上,你认为有很多公司在玩这个想法吗? But on the question of open source, do you think there’s a lot of companies playing with this idea?

Lex Fridman 挺有意思的。 It’s quite interesting.

Lex Fridman 我会说 Meta 出人意料地在这方面领跑,或者至少迈出了真正开源模型的第一步。 I would say Meta surprisingly has led the way on this, or at least took the first step in the game of chess of really open sourcing the model.

Lex Fridman 当然它不是最先进的 SOTA 模型,但毕竟开源了 Llama;Google 也在试探开源一个较小版本的想法。 Of course it’s not the state-of-the-art model, but open sourcing Llama Google is flirting with the idea of open sourcing a smaller version.

Lex Fridman 开源的优缺点是什么? What are the pros and cons of open sourcing?

Lex Fridman 你有没有研究过这个想法? Have you played around with this idea?

Sam Altman 嗯,我觉得开源模型肯定有一席之地,特别是那些人们可以在本地运行的更小的模型,我认为需求巨大。 Yeah, I think there is definitely a place for open source models, particularly smaller models that people can run locally, I think there’s huge demand for.

Sam Altman 我觉得会有一些开源模型,也会有一些闭源模型。 I think there will be some open source models, there will be some closed source models.

Sam Altman 在这方面,它和其他生态系统并无不同。 It won’t be unlike other ecosystems in that way.

Lex Fridman 我听了各种播客谈论这场诉讼之类的事情。 I listened to all in podcasts talking about this lawsuit and all that kind of stuff.

Lex Fridman 他们更担心的是从非营利转向这种带利润上限的营利模式的先例。 They were more concerned about the precedent of going from nonprofit to this cap for profit.

Lex Fridman 这给其他初创公司立了什么样的先例? What precedent that sets for other startups?

Lex Fridman 这是不是那种—— Is that something-

Sam Altman 我会强烈劝阻任何考虑以非营利起步、之后再设立营利分支的初创公司。 I would heavily discourage any startup that was thinking about starting as a nonprofit and adding a for-profit arm later.

Sam Altman 我会强烈劝阻他们这么做。 I’d heavily discourage them from doing that.

Sam Altman 我不认为我们会在这里设定什么先例。 I don’t think we’ll set a precedent here.

Lex Fridman 好吧。 Okay.

Lex Fridman 所以大多数初创公司应该直接—— So most startups should go just-

Sam Altman 当然。 For sure.

Lex Fridman 再说一次—— And again-

Sam Altman 如果我们当初知道会发生什么,我们也会那样做。 If we knew what was going to happen, we would’ve done that too.

Lex Fridman 理论上,如果你在这里舞跳得漂亮,会有一些税收优惠之类的,但…… Well in theory, if you dance beautifully here, there’s some tax incentives or whatever, but…

Sam Altman 我不认为大多数人是这样想这些事的。 I don’t think that’s how most people think about these things.

Lex Fridman 如果这样做,初创公司就不可能省下很多钱。 It’s just not possible to save a lot of money for a startup if you do it this way.

Sam Altman 不,我认为有法律会让那变得非常困难。 No, I think there’s laws that would make that pretty difficult.

Lex Fridman 你希望这件事和 Elon 走向何方? Where do you hope this goes with Elon?

Lex Fridman 这种紧张关系、这种博弈,你希望它怎样? This tension, this dance, what do you hope this?

Lex Fridman 如果我们展望 1、2、3 年后,你和他在个人层面的关系,比如友谊、友好的竞争,就这类事情。 If we go 1, 2, 3 years from now, your relationship with him on a personal level too, like friendship, friendly competition, just all this kind of stuff.

Sam Altman 是的,我非常尊重 Elon,我希望多年以后我们能有融洽的关系。 Yeah, I really respect Elon and I hope that years in the future we have an amicable relationship.

Lex Fridman 是的,我希望你们这个月就能有融洽的关系,就这么竞争、一起赢、一起探索这些想法。 Yeah, I hope you guys have an amicable relationship this month and just compete and win and explore these ideas together.

Lex Fridman 我猜确实存在人才竞争之类的事,但应该是友好的竞争。 I do suppose there’s competition for talent or whatever, but it should be friendly competition.

Lex Fridman 就造很酷的东西。 Just build cool shit.

Lex Fridman 而 Elon 造很酷的东西相当在行。 And Elon is pretty good at building cool shit.

Lex Fridman 你们也是。 So are you.

Interviewer 说到很酷的东西——Sora。 So speaking of cool shit, Sora.

Lex Fridman 我能问的问题大概有一百万个。 There’s like a million questions I could ask.

Interviewer 首先,它太令人惊叹了。 First of all, it’s amazing.

Lex Fridman 在产品层面它确实令人惊叹,在哲学层面也一样。 It truly is amazing on a product level but also just on a philosophical level.

Lex Fridman 那我就问一个技术/哲学问题:你认为它对世界的理解,比如与 GPT-4 相比,多了还是少了什么? So let me just technical/philosophical ask, what do you think it understands about the world more or less than GPT-4 for example?

Lex Fridman 当你用这些 patch 而不是语言 token 训练时,世界模型会怎样。 The world model when you train on these patches versus language tokens.

Sam Altman 我觉得这些模型对世界模型的理解,比大多数人愿意承认的要多。 I think all of these models understand something more about the world model than most of us give them credit for.

Sam Altman 而且因为这些模型也很清楚地有它们不理解或做不对的东西,人们很容易盯着弱点、看穿表象然后说:「啊,这都是假的。」但并非都是假的。 And because they’re also very clear things they just don’t understand or don’t get right, it’s easy to look at the weaknesses, see through the veil and say, “Ah, this is all fake.” But it’s not all fake.

Sam Altman 只是有些部分有效、有些部分无效。 It’s just some of it works and some of it doesn’t work.

Sam Altman 我记得刚开始看 Sora 视频时,我会看到一个人走到某个东西前面挡了几秒,然后走开,那个东西还在那里。 I remember when I started first watching Sora videos and I would see a person walk in front of something for a few seconds and occlude it and then walk away and the same thing was still there.

Sam Altman 我就想:「哇,这相当不错。」或者有些例子里,底层的物理在很长的序列步骤中都被表现得很好,让人觉得:「哇,这相当令人印象深刻。」但从根本上说,这些模型只是在不断进步,而且会持续进步。 I was like, “Oh, this is pretty good.” Or there’s examples where the underlying physics looks so well represented over a lot of steps in a sequence, it’s like, “|Oh, this is quite impressive.” But fundamentally, these models are just getting better and that will keep happening.

Sam Altman 如果你看从 DALL-E 1 到 2 到 3 到 Sora 的轨迹,每一代都有很多人唱衰说这不行那不行,看看现在。 If you look at the trajectory from DALL·E 1 to 2 to 3 to Sora, there are a lot of people that were dunked on each version saying it can’t do this, it can’t do that and look at it now.

Lex Fridman 嗯,你刚才提到的遮挡,基本上是把世界的三维物理建模得足够好,才能捕捉到这类现象。 Well, the thing you just mentioned is the occlusions is basically modeling the physics of the three-dimensional physics of the world sufficiently well to capture those kinds of things.

Sam Altman 嗯…… Well…

Lex Fridman 对,也许你可以告诉我,为了处理遮挡,世界模型需要些什么? Or yeah, maybe you can tell me, in order to deal with occlusions, what does the world model need to?

Sam Altman 嗯。 Yeah.

Sam Altman 所以我会说,它在处理遮挡方面确实做到了某些很好的东西。 So what I would say is it’s doing something to deal with occlusions really well.

Sam Altman 但要说它有一个很棒的底层三维世界模型,那就有点过头了。 What I represent that it has a great underlying 3D model of the world, it’s a little bit more of a stretch.

Lex Fridman 但只靠这类二维训练数据的方法,能达到那一步吗? But can you get there through just these kinds of two-dimensional training data approaches?

Sam Altman 看起来这条路会走得出人意料地远。 It looks like this approach is going to go surprisingly far.

Sam Altman 我不想在它会突破哪些极限、哪些不会上做太多推测,但…… I don’t want to speculate too much about what limits it will surmount and which it won’t, but…

Lex Fridman 你见过这个系统有什么有趣的局限? What are some interesting limitations of the system that you’ve seen?

Lex Fridman 我是说你发过的一些很有趣。 I mean there’s been some fun ones you’ve posted.

Sam Altman 各种各样有趣的都有。 There’s all kinds of fun.

Sam Altman 比如猫在视频的随机位置突然多出一截。 I mean, cat’s sprouting an extra limit at random points in a video.

Sam Altman 随你挑,但问题仍然很多,弱点仍然很多。 Pick what you want, but there’s still a lot of problem, there’s a lot of weaknesses.

Lex Fridman 你认为这是这个方法的根本缺陷,还是只要更大的模型、更好的技术细节、更好的数据、更多的数据就能解决猫长出来那种问题 [inaudible 00:37:19]? Do you think it’s a fundamental flaw of the approach or is it just bigger model or better technical details or better data, more data is going to solve the cat sprouting [inaudible 00:37:19]?

Sam Altman 我认为两者都是。 I would say yes to both.

Sam Altman 我觉得这个方法本身有某种东西,让我们感觉它和我们的思考、学习方式不同。 I think there is something about the approach which just seems to feel different from how we think and learn and whatever.

Sam Altman 同时我也认为,它会随着规模变得更好。 And then also I think it’ll get better with scale.

Lex Fridman 就像我提到的,LLM 有 token、文本 token,而 Sora 有视觉 patch,所以它把一切视觉数据——各种各样的视频和图像——转成 patch。 Like I mentioned, LLMS have tokens, text tokens, and Sora has visual patches so it converts all visual data, a diverse kinds of visual data videos and images into patches.

Lex Fridman 训练在你说的范围内是完全自监督的吗,还是有一些人工标注? Is the training to the degree you can say fully self supervised, there’s some manual labeling going on?

Lex Fridman 人类在这一切中的参与程度如何? What’s the involvement of humans in all this?

Sam Altman 我是说,在不具体谈论 Sora 方法的前提下,我们在工作中使用了大量人类数据。 I mean without saying anything specific about the Sora approach, we use lots of human data in our work.

Lex Fridman 但不是互联网规模的数据? But not internet scale data?

Lex Fridman 所以是很多人。 So lots of humans.

Lex Fridman 「很多」是个复杂的词,Sam。 Lots is a complicated word, Sam.

Sam Altman 我觉得在这个问题上「很多」是个中肯的词。 I think lots is a fair word in this case.

Lex Fridman 因为对我来说「很多」是……听着,我是个内向的人,和三个人待在一起,那就是很多人了。 Because to me, “lots”… Listen, I’m an introvert and when I hang out with three people, that’s a lot of people.

Lex Fridman 四个人,那是很多。 Four people, that’s a lot.

Lex Fridman 但我猜你的意思是超过…… But I suppose you mean more than…

Sam Altman 超过三个人在为这些模型的数据做标注,是的。 More than three people work on labeling the data for these models, yeah.

Lex Fridman 好吧。 Okay.

Lex Fridman 对。 Right.

Lex Fridman 但根本上,有大量的自监督学习。 But fundamentally, there’s a lot of self supervised learning.

Lex Fridman 因为你在技术报告里提到的是互联网规模的数据。 Because what you mentioned in the technical report is internet scale data.

Lex Fridman 那又是另一段美好的……简直像诗一样。 That’s another beautiful… It’s like poetry.

Lex Fridman 所以那是不依赖人工标注的海量数据。 So it’s a lot of data that’s not human label.

Lex Fridman 在这个意义上是自监督的? It’s self supervised in that way?

Sam Altman 是的。 Yeah.

Lex Fridman 然后问题是,互联网上到底有多少数据适合以这种自监督的方式使用——只要我们掌握自监督的细节。 And then the question is, how much data is there on the internet that could be used in this that is conducive to this kind of self supervised way if only we knew the details of the self supervised.

Lex Fridman 你有没有考虑过多公开一些细节? Have you considered opening it up a little more details?

Sam Altman 我们考虑过。 We have.

Sam Altman 你是专门指 Sora 吗? You mean for source specifically?

Lex Fridman 专门指 Sora。 Source specifically.

Lex Fridman 因为这太有意思了:LLM 的那种魔法现在能不能开始转向视觉数据?要做到那一点需要什么? Because it’s so interesting that can the same magic of LLMs now start moving towards visual data and what does that take to do that?

Sam Altman 在我看来答案是能,但我们还有更多工作要做。 I mean it looks to me like yes, but we have more work to do.

Lex Fridman 明白。 Sure.

Lex Fridman 危险在哪里? What are the dangers?

Lex Fridman 你为什么对发布这个系统有所顾虑? Why are you concerned about releasing the system?

Lex Fridman 这有哪些可能的危险? What are some possible dangers of this?

Sam Altman 坦白说,我们在发布系统之前必须做的一件事,就是把它达到某种效率水平,以提供人们想要的规模——我不想淡化这一点。 I mean frankly speaking, one thing we have to do before releasing the system is just get it to work at a level of efficiency that will deliver the scale people are going to want from this so that I don’t want to downplay that.

Sam Altman 那里还有海量、海量的工作要做。 And there’s still a ton ton of work to do there.

Sam Altman 但你可以想到深度伪造、虚假信息之类的问题。 But you can imagine issues with deepfakes, misinformation.

Sam Altman 我们努力做一家有思考的公司,认真考虑我们向世界发布什么;而这件事会怎样变坏,并不需要花太多心思就能看到。 We try to be a thoughtful company about what we put out into the world and it doesn’t take much thought to think about the ways this can go badly.

Lex Fridman 这里有很多棘手的问题,你身处一个非常艰难的领域。 There’s a lot of tough questions here, you’re dealing in a very tough space.

Lex Fridman 你认为训练 AI 在版权法下是否属于或应属于合理使用? Do you think training AI should be or is fair use under copyright law?

Sam Altman 我认为这个问题背后的问题是:创造了有价值数据的人,是否应该获得某种使用补偿——我觉得答案是应该。 I think the question behind that question is, do people who create valuable data deserve to have some way that they get compensated for use of it, and that I think the answer is yes.

Sam Altman 我还不知道具体的解决方案是什么。 I don’t know yet what the answer is.

Sam Altman 人们提出了很多不同的方案。 People have proposed a lot of different things.

Sam Altman 我们试过一些不同的模式。 We’ve tried some different models.

Sam Altman 但如果我是一名艺术家,第一,我希望能够选择退出——不让别人按我的风格生成作品。 But if I’m like an artist for example, A, I would like to be able to opt out of people generating art in my style.

Sam Altman 第二,如果他们真的按我的风格生成作品,我希望有一套相应的经济模式。 And B, if they do generate art in my style, I’d like to have some economic model associated with that.

Lex Fridman 是啊,就是从 CD 到 Napster 再到 Spotify 的那个转变过程。 Yeah, it’s that transition from CDs to Napster to Spotify.

Lex Fridman 我们必须找到某种模式。 We have to figure out some kind of model.

Sam Altman 模式会变,但人们必须得到报酬。 The model changes but people have got to get paid.

Lex Fridman 嗯,如果我们再往外拉远一点看,应该有一种激励,让人类继续做很酷的东西。 Well, there should be some kind of incentive if we zoom out even more for humans to keep doing cool shit.

Sam Altman 在我担心的所有事情里,我相信人类会去做很酷的东西,而社会会找到某种方式去回报它。 Of everything I worry about, humans are going to do cool shit and society is going to find some way to reward it.

Sam Altman 这看起来是天性里硬编码的。 That seems pretty hardwired.

Sam Altman 我们想创造,想有用,想以任何方式获得地位。 We want to create, we want to be useful, we want to achieve status in whatever way.

Sam Altman 我觉得这一点不会消失。 That’s not going anywhere I don’t think.

Lex Fridman 但回报未必是金钱上的。 But the reward might not be monetary financially.

Lex Fridman 可能是名声和对其他酷的东西的赞颂—— It might be fame and celebration of other cool-

Sam Altman 也可能是以其他方式的金钱回报。 Maybe financial in some other way.

Sam Altman 再说一次,我不认为我们已经看到经济体系运作方式的最后一次演化。 Again, I don’t think we’ve seen the last evolution of how the economic system’s going to work.

Lex Fridman 是啊,但艺术家和创作者很担忧。 Yeah, but artists and creators are worried.

Lex Fridman 当他们看到 Sora 时,他们的反应是:「我天。」 When they see Sora, they’re like, “Holy shit.”

Sam Altman 当然。 Sure.

Sam Altman 摄影术出现时,画家们也超级担忧,结果摄影成了一门新的艺术形式,人们靠拍照赚了很多钱。 Artists were also super worried when photography came out and then photography became a new art form and people made a lot of money taking pictures.

Sam Altman 我觉得这类事情会不断发生。 I think things like that will keep happening.

Sam Altman 人们会以新的方式使用新工具。 People will use the new tools in new ways.

Lex Fridman 如果看 YouTube 或类似的平台,你觉得未来五年里面用 Sora 这类 AI 生成内容的比例会有多少? If we just look on YouTube or something like this, how much of that will be using Sora like AI generated content, do you think, in the next five years?

Sam Altman 人们谈论 AI 五年内会做多少工作。 People talk about how many jobs is AI going to do in five years.

Sam Altman 人们使用的框架是:当前工作中有多大比例会被 AI 完全取代? The framework that people have is, what percentage of current jobs are just going to be totally replaced by some AI doing the job?

Sam Altman 我的思考方式不是 AI 能做多少工作,而是 AI 能在多长任务时间跨度上完成多少任务。 The way I think about it is not what percent of jobs AI will do, but what percent of tasks will AI do on over one time horizon.

Sam Altman 所以如果你想的是经济中所有五秒钟的任务、五分钟的任务、五小时的任务,甚至五天的任务,其中有多少 AI 能做? So if you think of all of the five-second tasks in the economy, five minute tasks, the five-hour tasks, maybe even the five-day tasks, how many of those can AI do?

Sam Altman 我觉得这是一个远比「AI 能做多少工作」更有趣、更有影响、更重要的问题,因为 AI 是一个会在不断提升的精巧程度上、在越来越长的时间跨度上、覆盖越来越多任务、并让人在更高抽象层次上工作的工具。 I think that’s a way more interesting, impactful, important question than how many jobs AI can do because it is a tool that will work at increasing levels of sophistication and over longer and longer time horizons for more and more tasks and let people operate at a higher level of abstraction.

Sam Altman 所以也许人们做同样工作的效率会高得多。 So maybe people are way more efficient at the job they do.

Sam Altman 到某个时候,那不仅是量变,还是质变——关于你能同时装进脑子里的问题类型。 And at some point that’s not just a quantitative change, but it’s a qualitative one too about the kinds of problems you can keep in your head.

Sam Altman 我觉得 YouTube 上的视频也会一样。 I think that for videos on YouTube it’ll be the same.

Sam Altman 很多视频,也许大多数,会在制作中使用 AI 工具,但它们仍然根本上由一个人在思考、组装、完成其中部分工作。 Many videos, maybe most of them, will use AI tools in the production, but they’ll still be fundamentally driven by a person thinking about it, putting it together, doing parts of it.

Sam Altman 有点像导演和运营那个项目。 Sort of directing and running it.

Lex Fridman 是啊,太有意思了。 Yeah, it’s so interesting.

Lex Fridman 我是说这有点可怕,但想想很有趣。 I mean it’s scary, but it’s interesting to think about.

Lex Fridman 我倾向于相信人类喜欢看其他人类,或者其他更像人的—— I tend to believe that humans like to watch other humans or other human humans-

Sam Altman 人类其实非常在意其他人类。 Humans really care about other humans a lot.

Lex Fridman 嗯。 Yeah.

Lex Fridman 如果有比人类更酷更好的东西,人类会在意它两天,然后又回到人类身上。 If there’s a cooler thing that’s better than a human, humans care about that for two days and then they go back to humans.

Sam Altman 这看起来是深层硬编码的。 That seems very deeply wired.

Lex Fridman 就像国际象棋那件事:「哦,是啊。」但现在大家仍然都继续下棋。 It’s the whole chess thing, “Oh, yeah,” but now let’s everybody keep playing chess.

Lex Fridman 我们无视房间里的大象:人类下棋的水平相对 AI 系统真的很差。 And let’s ignore the elephant in the room that humans are really bad at chess relative to AI systems.

Sam Altman 我们仍然赛跑,而汽车快得多。 We still run races and cars are much faster.

Sam Altman 我是说这样的例子很多。 I mean there’s a lot of examples.

Lex Fridman 是的。 Yeah.

Lex Fridman 也许它会像 Adobe 套件那种工具化的方式,让做视频变得容易得多,诸如此类。 And maybe it’ll just be tooling in the Adobe suite type of way where it can just make videos much easier and all that kind of stuff.

Lex Fridman 说真的,我讨厌在镜头前。 Listen, I hate being in front of the camera.

Lex Fridman 如果我能想办法不在镜头前,我会非常乐意。 If I can figure out a way to not be in front of the camera, I would love it.

Interviewer 可惜,那还得一段时间。 Unfortunately, it’ll take a while.

Lex Fridman 生成人脸正在进步,但视频格式里生成特定的人脸比生成通用的人脸要难得多。 That generating faces, it is getting there, but generating faces in video format is tricky when it’s specific people versus generic people.

Interviewer 让我问问你 GPT-4。 Let me ask you about GPT-4.

Interviewer 问题太多了。 There’s so many questions.

Interviewer 首先,同样令人惊叹。 First of all, also amazing.

Lex Fridman 回头看,它大概会是那种历史性的转折点——和 3、5 以及 ChatGPT 一起。 Looking back, it’ll probably be this kind of historic pivotal moment with 3, 5 and 4 which ChatGPT.

Sam Altman 也许 5 才是那个转折点。 Maybe five will be the pivotal moment.

Sam Altman 我不知道。 I don’t know.

Sam Altman 往前看很难说。 Hard to say that looking forward.

Lex Fridman 我们永远无法确定。 We’ll never know.

Lex Fridman 这就是未来烦人的地方:它很难预测。 That’s the annoying thing about the future, it’s hard to predict.

Lex Fridman 但对我来说,回头看,GPT-4、ChatGPT 真的相当厉害,是历史级别的厉害。 But for me, looking back, GPT-4, ChatGPT is pretty damn impressive, historically impressive.

Lex Fridman 所以请允许我问:对你来说,GPT-4 和 GPT-4 Turbo 最令人惊叹的能力是什么? So allow me to ask, what’s been the most impressive capabilities of GPT-4 to you and GPT-4 Turbo?

Sam Altman 我觉得它还挺烂的。 I think it kind of sucks.

Lex Fridman 典型的人类,也是会习惯一个很棒的东西。 Typical human also, gotten used to an awesome thing.

Sam Altman 不,我觉得它是很棒的东西,但相对我们需要到达、且我相信我们终将到达的地方——GPT-3 的时候,人们说:「哇,这太神奇了。 No, I think it is an amazing thing, but relative to where we need to get to and where I believe we will get to, at the time of GPT-3, people are like, “Oh, this is amazing.

Sam Altman 这是技术奇迹。」它确实是,曾经是。 This is marvel of technology.” And it is, it was.

Sam Altman 但现在我们有了 GPT-4,再看 GPT-3 你会觉得:「那简直无法想象地糟糕。」我预计 5 和 4 之间的差距会和 4 和 3 之间一样大。我觉得我们的工作就是活在几年后的未来,记住我们现在拥有的工具回头看会有点多烂——我们正是这样确保未来更好的。 But now we have GPT-4 and look at GPT-3 and you’re like, “That’s unimaginably horrible.” I expect that the delta between 5 and 4 will be the same as between 4 and 3 and I think it is our job to live a few years in the future and remember that the tools we have now are going to kind of suck looking backwards at them and that’s how we make sure the future is better.

Lex Fridman 它「烂」得最精彩的地方是什么? What are the most glorious ways in that GPT-4 sucks?

Lex Fridman 意思是—— Meaning-

Sam Altman 它最好的能力是什么? What are the best things it can do?

Lex Fridman 它最好的能力是什么,这些最好能力的局限又在哪里——既能让你说它烂,又因此给未来带来灵感和希望? What are the best things it can do and the limits of those best things that allow you to say it sucks, therefore gives you an inspiration and hope for the future?

Sam Altman 我最近越来越多地把它用作一个头脑风暴伙伴。 One thing I’ve been using it for more recently is sort of like a brainstorming partner.

Lex Fridman 是啊,[inaudible 00:46:25] 就是为此。 Yep, [inaudible 00:46:25] for that.

Sam Altman 里面有某种惊人的东西在闪光。 There’s a glimmer of something amazing in there.

Sam Altman 当人们谈论它、谈论它能做什么时,他们说:「哦,它帮我更高效地写代码。 When people talk about it, what it does, they’re like, “Oh, it helps me code more productively.

Sam Altman 它帮我写得更快、更好。 It helps me write more faster and better.

Sam Altman 它帮我从这种语言翻译成另一种语言。」所有这些惊人的事情。但那种创意头脑风暴伙伴的感觉——「我需要给这个东西起个名字。 It helps me translate from this language to another,” all these amazing things, but there’s something about the kind of creative brainstorming partner, “I need to come up with a name for this thing.

Sam Altman 我需要换个角度思考这个问题。 I need to think about this problem in a different way.

Sam Altman 我不确定这里该怎么办」——我觉得这让人瞥见了某种我希望能看到更多的东西。 I’m not sure what to do here,” that I think gives a glimpse of something I hope to see more of.

Sam Altman 另一个你能看到极小苗头的事情是,它能在更长跨度的任务上帮忙:把某件事拆解成多个步骤,也许执行其中一些步骤,搜索互联网,写代码,诸如此类,把它们组装起来。 One of the other things that you can see a very small glimpse of is when I can help on longer horizon tasks, break down something in multiple steps, maybe execute some of those steps, search the internet, write code, whatever, put that together.

Sam Altman 当那件事成功时——虽然不常发生——它非常神奇。 When that works, which is not very often, it’s very magical.

Lex Fridman 与人的迭代式来回互动,对我来说很管用。 The iterative back and forth with a human, it works a lot for me.

Lex Fridman 你是说它什么? What do you mean it-

Sam Altman 迭代式来回互动——当它能自己处理一个 10 步的问题时,这种情况会更多。 Iterative back and forth to human, it can get more often when it can go do a 10 step problem on its own.

Lex Fridman 哦。 Oh.

Sam Altman 它不常能做到那个,有时候不行。 It doesn’t work for that too often, sometimes.

Lex Fridman 是叠加多层抽象,还是你指的就是按顺序来? Add multiple layers of abstraction or do you mean just sequential?

Sam Altman 两者都有:先拆解,然后在不同抽象层次上做事,再把它们组装起来。 Both, to break it down and then do things that different layers of abstraction to put them together.

Sam Altman 你看,我不想淡化 GPT-4 的成就,但也不想夸大它。 Look, I don’t want to downplay the accomplishment of GPT-4, but I don’t want to overstate it either.

Sam Altman 我认为我们正处在一条指数曲线上,很快我们回看 GPT-4 就会像现在回看 GPT-3 一样。 And I think this point that we are on an exponential curve, we’ll look back relatively soon at GPT-4 like we look back at GPT-3 now.

Lex Fridman 话虽如此,我是说 ChatGPT 是一个转折点——人们开始相信(在 OpenAI 内部当然早就相信)。 That said, I mean ChatGPT was a transition to where people started to believe there is an uptick of believing, not internally at OpenAI.

Sam Altman 当然。 For sure.

Lex Fridman 也许这里就有很多信徒,但当你想到—— Perhaps there’s believers here, but when you think of-

Sam Altman 在那个意义上,我确实认为它是一个时刻:世界上很多人从不相信转向了相信。 And in that sense, I do think it’ll be a moment where a lot of the world went from not believing to believing.

Sam Altman 那更多是关于 ChatGPT 这个界面。 That was more about the ChatGPT interface.

Sam Altman 我说的界面和产品,也包括模型的后训练(post-training)、我们如何把它调得对你有用、如何使用它,而不仅仅是底层模型本身。 And by the interface and product, I also mean the post training of the model and how we tune it to be helpful to you and how to use it than the underlying model itself.

Lex Fridman 这些因素各有多重要? How much of each of those things are important?

Lex Fridman 底层模型,以及 RLHF 或类似的、把它调得对人类更有吸引力、更高效多产的技术。 The underlying model and the RLHF or something of that nature that tunes it to be more compelling to the human, more effective and productive for the human.

Sam Altman 我是说两者都超级重要,但 RLHF、后训练步骤——那个在基础模型之上、从算力角度看很小的包裹层——尽管它是巨大的工作量——这真的非常重要,更不用说围绕它打造的产品了。 I mean they’re both super important, but the RLHF, the post-training step, the little wrapper of things that from a compute perspective, little wrapper of things that we do on top of the base model even though it’s a huge amount of work, that’s really important to say nothing of the product that we build around it.

Sam Altman 某种意义上,我们必须做两件事。 In some sense, we did have to do two things.

Sam Altman 我们必须发明底层技术,然后我们必须弄清楚如何把它变成人们喜爱的产品——这不只是实际的产品工作本身,还有另一个完整的步骤:如何对齐它、让它有用。 We had to invent the underlying technology and then we had to figure out how to make it into a product people would love, which is not just about the actual product work itself, but this whole other step of how you align it and make it useful.

Lex Fridman 以及如何让规模可行,让很多人可以同时使用它。 And how you make the scale work where a lot of people can use it at the same time.

Lex Fridman 所有这些。 All that kind of stuff.

Sam Altman 还有那个。 And that.

Sam Altman 但那是已知的难题。 But that was a known difficult thing.

Sam Altman 我们知道必须把它规模化。 We knew we were going to have to scale it up.

Sam Altman 我们必须去做两件此前从未有人做过的事,我认为两者都是相当重大的成就;然后再做很多其他公司以前也做过的规模化工作。 We had to go do two things that had never been done before that were both I would say quite significant achievements and then a lot of things like scaling it up that other companies have had to do before.

Lex Fridman 从 GPT-4 到 GPT-4 Turbo,上下文窗口从 8K 到 128K token,这个对比如何? How does the context window of going from 8K to 128K tokens compare from GPT-4 to GPT-4 Turbo?

Sam Altman 大多数时候,大多数人并不需要一直到 128K。 Most people don’t need all the way to 128 most of the time.

Sam Altman 不过如果我们梦想遥远的未来——非常遥远的未来——我们的上下文长度会有几十亿。 Although if we dream into the distant future, we’ll have way distant future, we’ll have context length of several billion.

Sam Altman 你会把你所有的信息、你所有的历史都喂给它,它就会越来越了解你,那会非常棒。 You will feed in all of your information, all of your history over time and it’ll just get to know you better and better and that’ll be great.

Sam Altman 就目前而言,人们使用这些模型的方式还不是那样。 For now, the way people use these models, they’re not doing that.

Sam Altman 人们有时会贴入一篇论文或代码仓库的一大部分之类的,但大多数时候,模型的大部分使用并没有用长上下文。 People sometimes post in a paper or a significant fraction of a code repository, whatever, but most usage of the models is not using the long context most of the time.

Lex Fridman 我喜欢你这个「我有一个梦想」的演讲。 I like that this is your “I have a dream” speech.

Lex Fridman 有一天,人们将根据你完整上下文的品格、你整个一生的上下文来评判你。 One day you’ll be judged by the full context of your character or of your whole lifetime.

Lex Fridman 有意思。 That’s interesting.

Lex Fridman 所以你期望的扩展的一部分,就是越来越大、越来越大的上下文。 So that’s part of the expansion that you’re hoping for, is a greater and greater context.

Sam Altman 我以前在网上看到过一段片段——数字我可能记错——大概是比尔·盖茨在谈某台早期电脑的内存容量,可能是 64K,可能是 640K,差不多那样。 I saw this internet clip once, I’m going to get the numbers wrong, but it was like Bill Gates talking about the amount of memory on some early computer, maybe it was 64K, maybe 640K, something like that.

Sam Altman 大部分还都用在屏幕缓冲区上了。 Most of it was used for the screen buffer.

Sam Altman 他就是显得无法是真诚的。 He just couldn’t seem genuine.

Sam Altman 他就是无法想象世界最终会需要电脑里的千兆字节甚至万亿字节内存。 He just couldn’t imagine that the world would eventually need gigabytes of memory in a computer or terabytes of memory in a computer.

Sam Altman 而你永远只需要追随技术的指数曲线,我们就会弄清楚如何用上更好的技术。 And you always do, or you always do just need to follow the exponential of technology and we will find out how to use better technology.

Sam Altman 所以我现在真的无法想象上下文长度有一天达到几十亿会是什么样子。 So I can’t really imagine what it’s like right now for context links to go out to the billion someday.

Sam Altman 它们也许不会字面意义上到那个数,但效果上会感觉像那样。 And they might not literally go there, but effectively it’ll feel like that.

Sam Altman 但我知道一旦我们拥有了它,我们就会用上它,而且真的不想回头。 But I know we’ll use it and really not want to go back once we have it.

Lex Fridman 是的,就算说十年后达到几十亿,听起来都可能显得傻,因为那时会是万亿又万亿。 Yeah, even saying billions 10 years from now might seem dumb because it’ll be trillions upon trillions.

Sam Altman 当然。 Sure.

Lex Fridman 会有某种突破,让它在效果上感觉像无限上下文。 There’ll be some kind of breakthrough that will effectively feel like infinite context.

Lex Fridman 但即便是 128K,我得坦白,我还没把它推到那个程度。 But even 120, I have to be honest, I haven’t pushed it to that degree.

Lex Fridman 也许放进去整本书或书的一部分,还有论文之类的。 Maybe putting in entire books or parts of books and so on, papers.

Lex Fridman 你见过 GPT-4 有什么有意思的用例? What are some interesting use cases of GPT-4 that you’ve seen?

Sam Altman 我觉得最有意思的不是任何单个具体的用例——那些我们可以谈——而是一类人:主要是年轻人,他们把 GPT 当作任何知识工作的默认起点。 The thing that I find most interesting is not any particular use case that we can talk about those, but it’s people who kind of like, this is mostly younger people, but people who use it as their default start for any kind of knowledge work task.

Sam Altman 它能相当好地完成很多事情,这才是关键。 And it’s the fact that it can do a lot of things reasonably well.

Sam Altman 你可以用 GPT-4V【译注:原文转录作「GPT-V」,据上下文疑指 GPT-4V,无音频佐证,英文源文未改】,你可以用它帮你写代码,用它帮你搜索,用它编辑论文。 You can use GPT-V, you can use it to help you write code, you can use it to help you do search, you can use it to edit a paper.

Sam Altman 对我来说最有意思的,就是那些把它当作工作流起点的人。 The most interesting thing to me is the people who just use it as the start of their workflow.

Lex Fridman 我在很多事情上也是这样用的。 I do as well for many things.

Lex Fridman 我用它作为读书的阅读伙伴。 I use it as a reading partner for reading books.

Lex Fridman 它帮我思考、帮我把想法想透,尤其是当这些书是经典名著的时候。 It helps me think, help me think through ideas, especially when the books are classic.

Lex Fridman 关于它们的资料真的写得很充分。 So it’s really well written about.

Lex Fridman 我发现它在覆盖充分的主题上常常明显好过维基百科。 I find it often to be significantly better than even Wikipedia on well-covered topics.

Lex Fridman 不知为何更平衡、更有分寸。 It’s somehow more balanced and more nuanced.

Lex Fridman 也可能是我自己的问题,但它激发我比读一篇维基百科条目想得更深。 Or maybe it’s me, but it inspires me to think deeper than a Wikipedia article does.

Lex Fridman 我不太确定那是什么。 I’m not exactly sure what that is.

Interviewer 你提到了这种协作。 You mentioned this collaboration.

Lex Fridman 我不确定魔法在哪里——是在模型这边,在人那边,还是在两者之间。 I’m not sure where the magic is, if it’s in here or if it’s in there or if it’s somewhere in between.

Interviewer 我不确定。 I’m not sure.

Lex Fridman 但当我用 GPT 开始处理知识任务时,让我担心的事情之一是:我通常之后还得做事实核查,检查它没有编造东西。 But one of the things that concerns me for knowledge task when I start with GPT is I’ll usually have to do fact checking after, like check that it didn’t come up with fake stuff.

Lex Fridman 你是怎么发现 GPT 会编出听起来非常可信的假东西的? How do you figure that out that GPT can come up with fake stuff that sounds really convincing?

Interviewer 所以如何把它锚定在事实上? So how do you ground it in truth?

Sam Altman 那显然是我们高度关注的领域。 That’s obviously an area of intense interest for us.

Sam Altman 我认为在接下来的版本中它会好很多,但我们需要继续努力,今年之内我们不会把它全部解决。 I think it’s going to get a lot better with upcoming versions, but we’ll have to continue to work on it and we’re not going to have it all solved this year.

Lex Fridman 嗯,可怕的地方在于:随着它变好,你会越来越不做事实核查,对吧? Well the scary thing is, as it gets better, you’ll start not doing the fact checking more and more, right?

Sam Altman 我对这个有两种想法。 I’m of two minds about that.

Sam Altman 我认为人们使用技术的老练程度远超我们通常给他们的认可。 I think people are much more sophisticated users of technology than we often give them credit for.

Lex Fridman 当然。 Sure.

Sam Altman 人们似乎真的理解 GPT、这些模型中的任何一种,有时候会产生幻觉。 And people seem to really understand that GPT, any of these models hallucinate some of the time.

Sam Altman 如果事关重大,你就必须核查。 And if it’s mission-critical, you got to check it.

Lex Fridman 只是记者们似乎不理解这一点。 Except journalists don’t seem to understand that.

Lex Fridman 我见过记者半吊子地直接用 GPT-4。 I’ve seen journalists half-assedly just using GPT-4.

Lex Fridman 这是—— It’s-

Sam Altman 在我想批评记者的那一长串事项里,这倒不是我对他们的头号批评。 Of the long list of things I’d like to dunk on journalists for, this is not my top criticism of them.

Lex Fridman 嗯,我觉得更大的批评也许是:记者的压力和激励机制迫使你做得非常快,而这是个捷径。我希望我们的社会能奖励那种—— Well, I think the bigger criticism is perhaps the pressures and the incentives of being a journalist is that you have to work really quickly and this is a shortcut.I would love our society to incentivize like-

Sam Altman 我也希望。 I would too.

Lex Fridman ——花上几天、几周去做的新闻工作,奖励优秀的深度新闻。 … like a journalistic efforts that take days and weeks and rewards great in depth journalism.

Lex Fridman 也奖励那种以平衡方式呈现内容、在批评一个人的同时也赞美他的新闻——即使批评才带来点击量,编造东西同样带来点击量和完全歪曲事实的标题。 Also journalism that present stuff in a balanced way where it’s like celebrates people while criticizing them even though the criticism is the thing that gets clicks and making shit up also gets clicks and headlines that mischaracterized completely.

Lex Fridman 我肯定你们有很多人说风凉话:「嗯,那场风波大概带来了很多点击。」 I’m sure you have a lot of people dunking on, “Well, all that drama probably got a lot of clicks.”

Sam Altman 大概确实带来了。 Probably did.

Lex Fridman 这是我希望人类文明能改善的更大问题。 And that’s a bigger problem about human civilization I’d love to see-saw.

Lex Fridman 这就是我们希望多庆祝一点的地方。 This is where we celebrate a bit more.

Lex Fridman 你们给了 ChatGPT 记忆的能力。 You’ve given ChatGPT the ability to have memories.

Lex Fridman 你用之前对话在玩这个功能。 You’ve been playing with that about previous conversations.

Lex Fridman 还有关闭记忆的能力。 And also the ability to turn off memory.

Lex Fridman 有时候我也希望我能那样做。 I wish I could do that sometimes.

Lex Fridman 按需打开和关闭。 Just turn on and off, depending.

Lex Fridman 我想有时候酒精能做到,但大概不是最优的方式。 I guess sometimes alcohol can do that, but not optimally I suppose.

Lex Fridman 通过那件事你看到了什么——比如玩味「记住对话」这个想法,或者不记…… What have you seen through that, like playing around with that idea of remembering conversations and not…

Sam Altman 我们在这方面的探索还处于很早期,但我认为人们想要的——至少我自己想要的——是一个能了解我、并随时间对我越来越有用的模型。 We’re very early in our explorations here, but I think what people want, or at least what I want for myself, is a model that gets to know me and gets more useful to me over time.

Sam Altman 这是一次早期探索。 This is an early exploration.

Sam Altman 我觉得还有很多其他事情要做,但那就是我们想去的方向。 I think there’s a lot of other things to do, but that’s where we’d like to head.

Sam Altman 你想用一个模型,在你一生的使用过程中——会是很多个模型——它在你一生里变得越来越好。 You’d like to use a model, and over the course of your life or use a system, it’d be many models, and over the course of your life it gets better and better.

Lex Fridman 是的。 Yeah.

Lex Fridman 这个问题有多难? How hard is that problem?

Lex Fridman 因为现在它更像是记住一些小事实和小偏好之类的东西。 Because right now it’s more like remembering little factoids and preferences and so on.

Lex Fridman 关于记忆呢? What about remembering?

Lex Fridman 难道你不想要 GPT 记住你十一月经历的所有破事和所有戏剧性,然后你可以—— Don’t you want GPT to remember all the shit you went through in November and all the drama and then you can-

Sam Altman 是的。 Yeah.

Sam Altman 嗯。 Yeah.

Lex Fridman 因为现在你显然在屏蔽它一点。 Because right now you’re clearly blocking it out a little bit.

Sam Altman 不只是我希望它记住那些。 It’s not just that I want it to remember that.

Sam Altman 我希望它把那些教训整合起来,并在未来提醒我什么该做得不同、什么需要当心。 I want it to integrate the lessons of that and remind me in the future what to do differently or what to watch out for.

Sam Altman 我们在人生中都会以不同程度从经验里获益,也希望我的 AI agent 能随着这些经验一起成长。 We all gain from experience over the course of our lives in varying degrees, and I’d like my AI agent to gain with that experience too.

Sam Altman 所以如果我们回到并允许自己想象那万亿兆的上下文长度——如果我能把我一生中与人的所有对话都放进去,如果我所有的邮件收发、所有 输入输出都在我每次提问的上下文窗口里——我想那会相当酷。 So if we go back and let ourselves imagine that trillions and trillions of context length, if I can put every conversation I’ve ever had with anybody in my life in there, if I can have all of my emails input out, all of my input output in the context window every time I ask a question, that’d be pretty cool I think.

Lex Fridman 嗯,我觉得那会非常酷。 Yeah, I think that would be very cool.

Lex Fridman 人们有时听到这个会担心隐私。 People sometimes will hear that and be concerned about privacy.

Lex Fridman 你怎么看这个方面——AI 越是真正整合你经历过的所有体验、所有数据来给你建议,它就越有效? What do you think about that aspect of it, the more effective the AI becomes that really integrating all the experiences and all the data that happened to you and give you advice?

Sam Altman 我觉得那里正确的答案就是用户选择。 I think the right answer there is just user choice.

Sam Altman 任何我想从记录里抹掉的东西,我都希望能从我的 AI agent 里拿走。 Anything I want stricken from the record from my AI agent, I want to be able to take out.

Sam Altman 如果我什么都不想让它记住,我也想要那种选择。 If I don’t want to remember anything, I want that too.

Sam Altman 你和我对隐私/效用权衡 的位置,关于我们自己的 AI,可能有不同意见—— You and I may have different opinions about where on that privacy utility trade off for our own AI-

Sam Altman ……对 OpenAI 而言在隐私/效用上的权衡位置,有不同意见,这完全没问题。 …opinions about where on that privacy/utility trade-off for OpenAI going to be, which is totally fine.

Sam Altman 但我认为答案就是真正简单的用户选择。 But I think the answer is just really easy user choice.

Lex Fridman 但公司在用户选择上应该有很高程度的透明度。 But there should be some high level of transparency from a company about the user choice.

Lex Fridman 因为过去有些公司在「嗯,大概默认我们在收集你所有数据」这类事上有点鬼鬼祟祟。 Because sometimes companies in the past have been kind of shady about, “Eh, it’s kind of presumed that we’re collecting all your data.

Lex Fridman 「我们出于好的原因用它,比如广告,等等。」但对细节并不透明。 We’re using it for a good reason, for advertisement and so on.” But there’s not a transparency about the details of that.

Sam Altman 那完全属实。 That’s totally true.

Sam Altman 你早些时候提到我在 blocking 掉十一月的事情。 You mentioned earlier that I’m blocking out the November stuff.

Lex Fridman 只是逗逗你。 Just teasing you.

Sam Altman 嗯,我是说那确实是一件非常创伤性的事,它让我瘫痪了很长时间。 Well, I mean, I think it was a very traumatic thing and it did immobilize me for a long period of time.

Sam Altman 最艰难的工作绝对是在那段时间里继续工作,因为我不得不试着回到这里,把碎片重新拼起来,而那时我还处在震惊和痛苦中,尽管没人在乎那个。 Definitely the hardest work thing I’ve had to do was just keep working that period, because I had to try to come back in here and put the pieces together while I was just in shock and pain, and nobody really cares about that.

Sam Altman 我是说,团队对我放了行,我没有在平时的工作水平上。 I mean, the team gave me a pass and I was not working at my normal level.

Sam Altman 但有一段时间,两件事同时要做,真的很难。 But there was a period where it was really hard to have to do both.

Sam Altman 但有一天早上我醒来,我想:「这件可怕的事发生在我身上。 But I kind of woke up one morning, and I was like, “This was a horrible thing that happened to me.

Sam Altman 我觉得我可以永远当受害者,或者我可以说这是我一生中将接触的最重要的工作,我必须回到它。」这并不意味着我把它压抑了,因为有时候我半夜醒来会想到它,但我确实感到有责任继续前进。 I think I could just feel like a victim forever, or I can say this is the most important work I’ll ever touch in my life and I need to get back to it.” And it doesn’t mean that I’ve repressed it, because sometimes I wake up in the middle of the night thinking about it, but I do feel an obligation to keep moving forward.

Lex Fridman 嗯,说得很美,但里面可能会有些残留的东西。 Well, that’s beautifully said, but there could be some lingering stuff in there.

Lex Fridman 比如,我会担心的就是你提到的那个信任问题——对人多疑,而不是凭直觉去信任所有人或大多数人。 Like, what I would be concerned about is that trust thing that you mentioned, that being paranoid about people as opposed to just trusting everybody or most people, like using your gut.

Lex Fridman 这是一场艰难的舞蹈。 It’s a tricky dance.

Sam Altman 确实。 For sure.

Lex Fridman 我是说,因为在我业余的探索中,我深入研究了泽连斯基政府和普京政府,以及战时那个高压环境里双方的动态。 I mean, because I’ve seen in my part-time explorations, I’ve been diving deeply into the Zelenskyy administration and the Putin administration and the dynamics there in wartime in a very highly stressful environment.

Lex Fridman 会发生的是不信任,你们双方都自我孤立,开始看不清世界。 And what happens is distrust, and you isolate yourself, both, and you start to not see the world clearly.

Lex Fridman 那是一个人类的担忧。 And that’s a human concern.

Lex Fridman 看起来你泰然处之,学到了好的教训,感受到了爱,并让爱给你注入能量——这很棒,但残留的东西仍然可以在里面。 You seem to have taken it in stride and kind of learned the good lessons and felt the love and let the love energize you, which is great, but still can linger in there.

Lex Fridman 我只是有些问题很希望能问:你的直觉,关于 GPT 能做什么和不能做什么。 There’s just some questions I would love to ask, your intuition about what’s GPT able to do and not.

Lex Fridman 所以它大约为它生成的每个 token 分配同等量的算力。 So it’s allocating approximately the same amount of compute for each token it generates.

Lex Fridman 在这种方法里,有没有空间做更慢的思考、序列式思考? Is there room there in this kind of approach to slower thinking, sequential thinking?

Sam Altman 我认为那种思考会有一个新的范式。 I think there will be a new paradigm for that kind of thinking.

Lex Fridman 它在架构上会和我们现在看到的 LLM 类似吗? Will it be similar architecturally as what we’re seeing now with LLMs?

Lex Fridman 它是 LLM 之上的一层吗? Is it a layer on top of LLMs?

Sam Altman 我能想象出实现它的很多种方式。 I can imagine many ways to implement that.

Sam Altman 我认为那不如你真正在问的问题重要,那就是:我们是否需要一种更慢的思考方式,让回答不必……我想在精神上你可以说,你希望 AI 能对更难的问题思考得更用力,对更简单的问题回答得更快。 I think that’s less important than the question you were getting at, which is, do we need a way to do a slower kind of thinking, where the answer doesn’t have to get… I guess spiritually you could say that you want an AI to be able to think harder about a harder problem and answer more quickly about an easier problem.

Sam Altman 我觉得那会很关键。 And I think that will be important.

Lex Fridman 那是我们人类拥有的想法吗——你就是应该能努力思考? Is that like a human thought that we just have and you should be able to think hard?

Lex Fridman 那是错误的直觉吗? Is that wrong intuition?

Sam Altman 我倾向于认为那是一个合理的直觉。 I suspect that’s a reasonable intuition.

Lex Fridman 有意思。 Interesting.

Lex Fridman 所以不可能出现那种情况:一旦 GPT 变成像 GPT-7,就能瞬间看到——「这是费马大定理的证明」? So it’s not possible once the GPT gets like GPT-7, would just instantaneously be able to see, “Here’s the proof of Fermat’s Theorem”?

Sam Altman 对我来说,你似乎想能为更难的问题分配更多算力。 It seems to me like you want to be able to allocate more compute to harder problems.

Sam Altman 对我来说,如果你问这样一个系统:「证明费马大定理」,对比「今天是几号?」——除非它已经知道并记住了那个证明的答案,假设它得去算出来——那似乎需要更多算力。 It seems to me that if you ask a system like that, “Prove Fermat’s Last Theorem,” versus, “What’s today’s date?,” unless it already knew and and had memorized the answer to the proof, assuming it’s got to go figure that out, seems like that will take more compute.

Lex Fridman 但它能看起来基本上就是一个 LLM 在自言自语吗?那种事? But can it look like basically an LLM talking to itself, that kind of thing?

Sam Altman 也许。 Maybe.

Sam Altman 我是说,你能想象得出很多种可能有效的做法。 I mean, there’s a lot of things that you could imagine working.

Sam Altman 正确的或最好的做法会是什么,我们还不知道。 What the right or the best way to do that will be, we don’t know.

Lex Fridman 这确实让我想到 Q* 背后的神秘传说。 This does make me think of the mysterious lore behind Q*.

Lex Fridman 这个神秘的 Q* 项目是什么? What’s this mysterious Q* project?

Lex Fridman 它也在同一个核设施里吗? Is it also in the same nuclear facility?

Sam Altman 没有什么核设施。 There is no nuclear facility.

Lex Fridman 嗯哼。 Mm-hmm.

Lex Fridman 那就是有核设施的人总会说的话。 That’s what a person with a nuclear facility always says.

Sam Altman 我倒真希望有个秘密核设施。 I would love to have a secret nuclear facility.

Sam Altman 没有。 There isn’t one.

Lex Fridman 好吧。 All right.

Sam Altman 也许有一天。 Maybe someday.

Lex Fridman 有一天? Someday?

Lex Fridman 好吧。 All right.

Lex Fridman 人总可以做梦。 One can dream.

Sam Altman OpenAI 不是一个善于保守秘密的公司。 OpenAI is not a good company at keeping secrets.

Sam Altman 那会挺好的。 It would be nice.

Sam Altman 我们一直深受各种泄密之扰,如果能有那样一个秘密,反而挺好。 We’re like, been plagued by a lot of leaks, and it would be nice if we were able to have something like that.

Lex Fridman 你能说说 Q* 是什么吗? Can you speak to what Q* is?

Sam Altman 我们还没准备好谈那个。 We are not ready to talk about that.

Lex Fridman 你看,但那样的回答意味着确实有可谈的东西。 See, but an answer like that means there’s something to talk about.

Lex Fridman 这很神秘,Sam。 It’s very mysterious, Sam.

Sam Altman 我是说,我们做各种各样的研究。 I mean, we work on all kinds of research.

Sam Altman 我们早就说过,我们认为这些系统中更好的推理是我们想要追求的一个重要方向。 We have said for a while that we think better reasoning in these systems is an important direction that we’d like to pursue.

Sam Altman 我们还没破解密码。 We haven’t cracked the code yet.

Sam Altman 我们对它非常感兴趣。 We’re very interested in it.

Lex Fridman 会有 Q* 或其他东西的时刻吗——像 ChatGPT 那样的飞跃,让你觉得—— Is there going to be moments, Q* or otherwise, where there’s going to be leaps similar to ChatGPT, where you’re like…

Sam Altman 这是个好问题。 That’s a good question.

Sam Altman 我是怎么想的? What do I think about that?

Sam Altman 有意思。 It’s interesting.

Sam Altman 对我来说,这一切感觉都相当连续。 To me, it all feels pretty continuous.

Lex Fridman 对。 Right.

Lex Fridman 这有点像你说的一个主题:你基本上是在沿一条指数曲线往上爬。 This is kind of a theme that you’re saying, is you’re basically gradually going up an exponential slope.

Lex Fridman 但从局外人的视角,就我看着而言,确实感觉有飞跃。 But from an outsider’s perspective, from me just watching, it does feel like there’s leaps.

Lex Fridman 但对你来说没有? But to you, there isn’t?

Sam Altman 我的确在想我们是否应该……所以我们部署方式的原因之一——我们称之为迭代式部署——而不是秘密地构建直到完全做出 GPT-5。我们决定谈论 GPT-1、2、3 和 4。 I do wonder if we should have… So part of the reason that we deploy the way we do, we call it iterative deployment, rather than go build in secret until we got all the way to GPT-5, we decided to talk about GPT-1, 2, 3, and 4.

Sam Altman 部分原因是,我认为 AI 和惊喜不相容。 And part of the reason there is I think AI and surprise don’t go together.

Sam Altman 而且这个世界、人们、机构,随便你怎么称呼它,都需要时间去适应和思考这些事情。 And also the world, people, institutions, whatever you want to call it, need time to adapt and think about these things.

Sam Altman 我认为 OpenAI 做过的最好的事情之一就是这个战略:让世界关注进展,认真对待 AGI,在被逼到墙角、不得不仓促决策之前,先思考我们想要什么样的系统、结构和治理。 And I think one of the best things that OpenAI has done is this strategy, and we get the world to pay attention to the progress, to take AGI seriously, to think about what systems and structures and governance we want in place before we’re under the gun and have to make a rush decision.

Interviewer 我觉得那真的很好。 I think that’s really good.

Sam Altman 但像你这样的人以及其他人都说仍然感觉有这些飞跃存在,这让我觉得也许我们应该以更迭代的方式发布。 But the fact that people like you and others say you still feel like there are these leaps makes me think that maybe we should be doing our releasing even more iteratively.

Sam Altman 我不知道那具体意味着什么,我也没有现成的答案,但我们的目标不是给世界带来震惊式的更新。 And I don’t know what that would mean, I don’t have an answer ready to go, but our goal is not to have shock updates to the world.

Interviewer 恰恰相反。 The opposite.

Lex Fridman 是的,当然。 Yeah, for sure.

Lex Fridman 更迭代会很棒。 More iterative would be amazing.

Lex Fridman 我觉得那对每个人来说都是美好的。 I think that’s just beautiful for everybody.

Sam Altman 但那就是我们在努力做的,那是我们公开的战略,而我不知道为什么我们多少有点没做到位。 But that’s what we’re trying to do, that’s our stated strategy, and I think we’re somehow missing the mark.

Sam Altman 所以也许我们应该用不同的方式来发布 GPT-5,或者类似的做法。 So maybe we should think about releasing GPT-5 in a different way or something like that.

Lex Fridman 是的,4.71、4.72。 Yeah, 4.71, 4.72.

Lex Fridman 但人们喜欢庆祝,人们庆祝生日。 But people tend to like to celebrate, people celebrate birthdays.

Lex Fridman 我不知道你认不认识人类,但他们就是有这些里程碑之类的东西。 I don’t know if you know humans, but they kind of have these milestones and those things.

Sam Altman 我确实认识一些人类。 I do know some humans.

Sam Altman 人们确实喜欢里程碑。 People do like milestones.

Sam Altman 我完全理解。 I totally get that.

Sam Altman 我想我们也喜欢里程碑。 I think we like milestones too.

Sam Altman 宣布这一件胜利然后开始下一件,是有趣的。 It’s fun to declare victory on this one and go start the next thing.

Sam Altman 但是啊,我觉得我们在这件事上多少有点做错了。 But yeah, I feel like we’re somehow getting this a little bit wrong.

Lex Fridman 所以 GPT-5 到底什么时候发布来着? So when is GPT-5 coming out again?

Sam Altman 我不知道。 I don’t know.

Sam Altman 这才是诚实的回答。 That’s the honest answer.

Lex Fridman 哦,那就是诚实的回答。 Oh, that’s the honest answer.

Lex Fridman 如果是今年,就眨两次眼。 Blink twice if it’s this year.

Sam Altman 我们今年会发布一个很棒的新模型。 We will release an amazing new model this year.

Sam Altman 我不知道我们会叫它什么。 I don’t know what we’ll call it.

Lex Fridman 这就回到那个问题:我们发布这个东西的方式是什么? So that goes to the question of, what’s the way we release this thing?

Sam Altman 我们会在未来几个月里发布很多不同的东西。 We’ll release in the coming months many different things.

Sam Altman 我觉得那会很酷。 I think that’d be very cool.

Sam Altman 我觉得在我们谈论一个被称作(或不被称作)GPT-5、或比你对 GPT-5 的预期稍差或稍好的模型之前,我们还有很多其他重要的东西要先发布。 I think before we talk about a GPT-5-like model called that, or not called that, or a little bit worse or a little bit better than what you’d expect from a GPT-5, I think we have a lot of other important things to release first.

Lex Fridman 我不知道该对 GPT-5 期待什么。 I don’t know what to expect from GPT-5.

Lex Fridman 你让我既紧张又兴奋。 You’re making me nervous and excited.

Lex Fridman 不管它最终叫什么——就叫它 GPT-5 吧——最大的挑战和瓶颈是什么? What are some of the biggest challenges and bottlenecks to overcome for whatever it ends up being called, but let’s call it GPT-5?

Lex Fridman 就是问问而已。 Just interesting to ask.

Lex Fridman 是在算力这一侧吗? Is it on the compute side?

Lex Fridman 是在技术这一侧吗? Is it on the technical side?

Sam Altman 永远是所有这些加在一起。 It’s always all of these.

Sam Altman 你知道,那个唯一的大解锁是什么? You know, what’s the one big unlock?

Sam Altman 是更大的计算机吗? Is it a bigger computer?

Sam Altman 是一个新的秘密吗? Is it a new secret?

Sam Altman 还是别的什么? Is it something else?

Sam Altman 是所有这些东西合在一起。 It’s all of these things together.

Sam Altman 我认为 OpenAI 做得非常出色的一点是——这其实是 Ilya 的一句原话,我要把它说砸了——大意是「我们把 200 个中等规模的事情相乘,变成一个巨大的事情。」 The thing that OpenAI, I think, does really well… This is actually an original Ilya quote that I’m going to butcher, but it’s something like, “We multiply 200 medium-sized things together into one giant thing.”

Lex Fridman 所以这是一种分布式持续创新? So there’s this distributed constant innovation happening?

Sam Altman 是的。 Yeah.

Lex Fridman 即便在技术层面? So even on the technical side?

Sam Altman 尤其在技术层面。 Especially on the technical side.

Lex Fridman 甚至具体的细节方法? So even detailed approaches?

Sam Altman 是的。 Yeah.

Lex Fridman 你们如何在每个细节上……不同的、分散的团队之间这是怎么协作的? Like you do detailed aspects of every… How does that work with different, disparate teams and so on?

Lex Fridman 那些中等规模的事情如何变成一个完整的巨型 Transformer? How do the medium-sized things become one whole giant Transformer?

Sam Altman 有几个人必须思考怎么把整体拼起来,但很多人会努力在脑中保持大部分的全图。 There’s a few people who have to think about putting the whole thing together, but a lot of people try to keep most of the picture in their head.

Lex Fridman 哦,就是每个团队、每个贡献者都试图把握更大的图景? Oh, like the individual teams, individual contributors try to keep the bigger picture?

Sam Altman 在一个高层次上,是的。 At a high level, yeah.

Sam Altman 你当然不会确切知道每个部分如何工作,但我通常相信的一点是:有时拉远视角、看看整张地图是有用的。 You don’t know exactly how every piece works, of course, but one thing I generally believe is that it’s sometimes useful to zoom out and look at the entire map.

Sam Altman 我认为这对技术问题是真的,对商业创新也是真的。 And I think this is true for a technical problem, I think this is true for innovating in business.

Sam Altman 但事情会以惊人的方式连接起来,理解整个图景——即使你大部分时间都在某个领域的细节里奋战——会结出令人惊讶的洞见。 But things come together in surprising ways, and having an understanding of that whole picture, even if most of the time you’re operating in the weeds in one area, pays off with surprising insights.

Sam Altman 事实上,我过去拥有过的、非常有价值的东西之一,是我脑中有科技行业所有或大部分前沿领域的一张地图。 In fact, one of the things that I used to have and was super valuable was I used to have a good map of all or most of the frontiers in the tech industry.

Sam Altman 我有时能看到那些连接或新的可能性,而如果我只深扎在一个领域,我不会有那些想法,因为没有全部的数据。 And I could sometimes see these connections or new things that were possible that if I were only deep in one area, I wouldn’t be able to have the idea for because I wouldn’t have all the data.

Sam Altman 而现在我没有那么多地图了。 And I don’t really have that much anymore.

Sam Altman 我现在非常深。 I’m super deep now.

Sam Altman 但我知道那是有价值的东西。 But I know that it’s a valuable thing.

Lex Fridman 你已经不是你过去的那个人了,Sam。 You’re not the man you used to be, Sam.

Sam Altman 现在的工作和过去很不一样。 Very different job now than what I used to have.

Lex Fridman 说到拉远视角,我们拉远一点,聊聊你说过的另一句看似调皮、但也许很深刻的话。 Speaking of zooming out, let’s zoom out to another cheeky thing, but profound thing, perhaps, that you said.

Lex Fridman 你发推说需要 7 万亿美元。 You tweeted about needing $7 trillion.

Sam Altman 我没发过那条推文。 I did not tweet about that.

Sam Altman 我从没说过「我们要筹集 7 万亿美元」之类的。 I never said, like, “We’re raising $7 trillion,” blah blah blah.

Lex Fridman 哦,那是别人? Oh, that’s somebody else?

Sam Altman 是的。 Yeah.

Lex Fridman 哦,但你说过「去他的,也许是 8 万亿」,我记得? Oh, but you said, “Fuck it, maybe eight,” I think?

Sam Altman 好吧,当世上出现错误信息时我发过一次梗。 Okay, I meme once there’s misinformation out in the world.

Lex Fridman 哦,你是在玩梗。 Oh, you meme.

Lex Fridman 但错误信息里可能有洞见的基础。 But misinformation may have a foundation of insight there.

Sam Altman 你看,我认为算力将是未来的货币。 Look, I think compute is going to be the currency of the future.

Sam Altman 我认为它可能是世界上最珍贵的商品,我认为我们应该大力投资以制造更多算力。 I think it will be maybe the most precious commodity in the world, and I think we should be investing heavily to make a lot more compute.

Sam Altman 算力,我认为它将是一个不寻常的市场。 Compute, I think it’s going to be an unusual market.

Sam Altman 人们会想手机芯片之类的市场。 People think about the market for chips for mobile phones or something like that.

Sam Altman 你可以那样说:好吧,世界上有 80 亿人,也许 70 亿人有手机,就算 60 亿吧。 And you can say that, okay, there’s 8 billion people in the world, maybe 7 billion of them have phones, maybe 6 billion, let’s say.

Sam Altman 他们每两年升级一次,所以每年市场是 30 亿部智能手机的 SoC。 They upgrade every two years, so the market per year is 3 billion system-on-chip for smartphones.

Sam Altman 而如果你造 300 亿,你并不会卖出 10 倍多的手机,因为大多数人只有一部手机。 And if you make 30 billion, you will not sell 10 times as many phones, because most people have one phone.

Interviewer 但算力不同。 But compute is different.

Sam Altman 智能会更像能源之类的东西,我认为唯一说得通的讨论方式是:在价格 X 时,世界会使用这么多算力;在价格 Y 时,世界会使用这么多算力。 Intelligence is going to be more like energy or something like that, where the only thing that I think makes sense to talk about is, at price X, the world will use this much compute, and at price Y, the world will use this much compute.

Sam Altman 因为如果真的很便宜,我会让它整天读我的邮件、给我建议我该思考或做什么、还尝试治愈癌症;如果很贵,我——或我们——可能只会用它来尝试治愈癌症。 Because if it’s really cheap, I’ll have it reading my email all day, giving me suggestions about what I maybe should think about or work on, and trying to cure cancer, and if it’s really expensive, maybe I’ll only use it, or we’ll only use it, to try to cure cancer.

Sam Altman 所以我认为世界会需要巨量的算力。 So I think the world is going to want a tremendous amount of compute.

Sam Altman 而那里有很多困难的部分。 And there’s a lot of parts of that that are hard.

Sam Altman 能源是最难的部分,建数据中心也很难,供应链也很难,当然,制造足够多的芯片也很难。 Energy is the hardest part, building data centers is also hard, the supply chain is hard, and then of course, fabricating enough chips is hard.

Sam Altman 但这似乎就是事情的方向。 But this seems to be where things are going.

Sam Altman 我们会需要一种现在难以推理的量级的算力。 We’re going to want an amount of compute that’s just hard to reason about right now.

Lex Fridman 你怎么解决能源难题? How do you solve the energy puzzle?

Lex Fridman 核—— Nuclear-

Sam Altman 那就是我相信的。 That’s what I believe.

Lex Fridman ……聚变? …fusion?

Sam Altman 那就是我相信的。 That’s what I believe.

Lex Fridman 核聚变? Nuclear fusion?

Sam Altman 是的。 Yeah.

Lex Fridman 谁会解决它? Who’s going to solve that?

Sam Altman 我认为 Helion 在做最好的工作,但我很高兴现在有一场聚变竞赛。 I think Helion’s doing the best work, but I’m happy there’s a race for fusion right now.

Sam Altman 核裂变,我认为,也相当了不起,我希望作为世界我们能重新拥抱它。 Nuclear fission, I think, is also quite amazing, and I hope as a world we can re-embrace that.

Sam Altman 那段历史如何演变真的让我难过,希望我们能以有意义的方式回到它。 It’s really sad to me how the history of that went, and hope we get back to it in a meaningful way.

Lex Fridman 所以对你来说,拼图的一部分是核裂变? So to you, part of the puzzle is nuclear fission?

Lex Fridman 像我们现有的那种核反应堆? Like nuclear reactors as we currently have them?

Lex Fridman 很多人因为切尔诺贝利之类的事件而恐惧? And a lot of people are terrified because of Chernobyl and so on?

Sam Altman 嗯,我认为我们应该造新的反应堆。 Well, I think we should make new reactors.

Sam Altman 我觉得那个行业几乎停滞不前,真是太可惜了。 I think it’s just a shame that industry kind of ground to a halt.

Lex Fridman 而大规模癔症就是你对停滞的解释? And just mass hysteria is how you explain the halt?

Sam Altman 是的。 Yeah.

Lex Fridman 我不知道你认不认识人类,但那就是危险之一。 I don’t know if you know humans, but that’s one of the dangers.

Lex Fridman 核裂变的一个安全威胁就是,人类似乎非常害怕它。 That’s one of the security threats for nuclear fission, is humans seem to be really afraid of it.

Lex Fridman 这是我们必须在权衡中考虑的因素,所以我们得赢回人心,展示它有多安全。 And that’s something we’ll have to incorporate into the calculus of it, so we have to kind of win people over and to show how safe it is.

Sam Altman 我对 AI 也担心这个。 I worry about that for AI.

Sam Altman 我认为有些跟 AI 有关的事情会以戏剧性的方式出大问题。 I think some things are going to go theatrically wrong with AI.

Sam Altman 我不知道我最终被人开枪的概率是多少,但不是零。 I don’t know what the percent chance is that I eventually get shot, but it’s not zero.

Lex Fridman 哦,就像我们想阻止这种—— Oh, like we want to stop this from-

Sam Altman 也许吧。 Maybe.

Lex Fridman 你怎么降低它的戏剧性? How do you decrease the theatrical nature of it?

Lex Fridman 我已经开始听到暗流涌动了,因为我确实和政治光谱两边的人都交谈,听到风声说这会被政治化。 I’m already starting to hear rumblings, because I do talk to people on both sides of the political spectrum, hear rumblings where it’s going to be politicized.

Lex Fridman AI 会被政治化,这真的让我担忧,因为那样可能就变成右派反对 AI、左派支持 AI 因为 AI 会帮助人民,或者诸如此类的叙事和定式,那真的让我担忧。 AI is going to be politicized, which really worries me, because then it’s like maybe the right is against AI and the left is for AI because it’s going to help the people, or whatever the narrative and the formulation is, that really worries me.

Lex Fridman 然后它的戏剧性就会被充分利用。 And then the theatrical nature of it can be leveraged fully.

Lex Fridman 你怎么对抗那种趋势? How do you fight that?

Sam Altman 我觉得它会被卷进左右之争。 I think it will get caught up in left versus right wars.

Sam Altman 我不知道那具体会是什么样子,但遗憾的是,任何有分量的事情都会发生这种事。 I don’t know exactly what that’s going to look like, but I think that’s just what happens with anything of consequence, unfortunately.

Sam Altman 我所说的戏剧性风险更多是指:我相信 AI 带来的好处会远远多于坏处,但它确实会有坏处,而且有些坏处是坏但不戏剧性的。 What I meant more about theatrical risks is AI’s going to have, I believe, tremendously more good consequences than bad ones, but it is going to have bad ones, and there’ll be some bad ones that are bad but not theatrical.

Sam Altman 例如,死于空气污染的人远比死于核反应堆的人多。 A lot more people have died of air pollution than nuclear reactors, for example.

Sam Altman 但大多数人更担心住在核反应堆旁边,而不是燃煤电厂旁边。 But most people worry more about living next to a nuclear reactor than a coal plant.

Sam Altman 但我们天性中的某种东西决定了:尽管我们要面对很多种风险,那些适合做电影高潮场景的风险在我们心中的分量,远大于那些很坏但缓慢燃烧的长期风险。 But something about the way we’re wired is that although there’s many different kinds of risks we have to confront, the ones that make a good climax scene of a movie carry much more weight with us than the ones that are very bad over a long period of time but on a slow burn.

Lex Fridman 嗯,这就是为什么真相重要,希望 AI 能帮我们看清事物的真相,保持平衡,理解真正的风险是什么、世界上事物的真正危险是什么。 Well, that’s why truth matters, and hopefully AI can help us see the truth of things, to have balance, to understand what are the actual risks, what are the actual dangers of things in the world.

Lex Fridman 在这个领域与 Google、Meta、xAI 等竞争,有什么优缺点? What are the pros and cons of the competition in the space and competing with Google, Meta, xAI, and others?

Sam Altman 我觉得我对这个有一个相当直白的回答,也许以后我能想出更多细粒度,但好处似乎很明显:我们更快、更便宜地得到更好的产品和更多的创新,以及竞争是好的所有理由。 I think I have a pretty straightforward answer to this that maybe I can think of more nuance later, but the pros seem obvious, which is that we get better products and more innovation faster and cheaper, and all the reasons competition is good.

Sam Altman 而坏处是,我认为如果我们不小心,它可能导致某种军备竞赛升级,这让我紧张。 And the con is that I think if we’re not careful, it could lead to an increase in sort of an arms race that I’m nervous about.

Lex Fridman 你感受到那种军备竞赛的压力了吗?比如在某些负面的 [inaudible 01:15:25]? Do you feel the pressure of that arms race, like in some negative [inaudible 01:15:25]?

Sam Altman 在某些方面肯定有,确实。 Definitely in some ways, for sure.

Sam Altman 我们花很多时间谈论把安全置于优先的必要性。 We spend a lot of time talking about the need to prioritize safety.

Sam Altman 我很久以前说过,你可以想象一个四象限:AGI 开始时间的短与长,起飞速度的慢与快。 And I’ve said for a long time that you think of a quadrant of slow timelines for the start of AGI, long timelines, and then a short takeoff or a fast takeoff.

Sam Altman 我认为短时间线、慢起飞是最安全的象限,也是我最希望我们处于的象限。 I think short timeline, slow takeoff is the safest quadrant and the one I’d most like us to be in.

Sam Altman 但我确实想确保我们得到那个慢起飞。 But I do want to make sure we get that slow takeoff.

Lex Fridman 我和 Elon 之间那种微妙摩擦的部分问题在于:在安全方面形成了一个个 silo,而不是协作。 Part of the problem I have with this kind of slight beef with Elon is that there’s silos created as opposed to collaboration on the safety aspect of all of this.

Lex Fridman 事情倾向于走进孤岛并封闭起来。 It tends to go into silos and closed.

Lex Fridman 也许模型倒是开源的。 Open source, perhaps, in the model.

Sam Altman 至少 Elon 说他对 AI 安全非常在意、真的担心,我假设他不会不安全地竞赛。 Elon says, at least, that he cares a great deal about AI safety and is really worried about it, and I assume that he’s not going to race unsafely.

Lex Fridman 是的。 Yeah.

Lex Fridman 但这里的协作,我认为对各方都真的有益。 But collaboration here, I think, is really beneficial for everybody on that front.

Sam Altman 那不是他最出名的品质。 Not really the thing he’s most known for.

Lex Fridman 嗯,他以关心人类著称,而人类受益于协作,所以激励和动机之间总有张力。 Well, he is known for caring about humanity, and humanity benefits from collaboration, and so there’s always a tension in incentives and motivations.

Lex Fridman 最终,我希望人性会胜出。 And in the end, I do hope humanity prevails.

Sam Altman 我在想,前几天有人提醒我:在他超过杰夫·贝索斯成为世界首富的那天,他给贝索斯发了一枚银牌的推文。 I was thinking, someone just reminded me the other day about how the day that he surpassed Jeff Bezos for richest person in the world, he tweeted a silver medal at Jeff Bezos.

Sam Altman 我希望随着人们开始为 AGI 努力,这样的事少一些。 I hope we have less stuff like that as people start to work towards AGI.

Lex Fridman 我同意。 I agree.

Lex Fridman 我觉得 Elon 是朋友,他是一个美好的人,是史上最重要的人类之一。 I think Elon is a friend and he’s a beautiful human being and one of the most important humans ever.

Lex Fridman 那种事不好。 That stuff is not good.

Sam Altman Elon 了不起的地方是真了不起,我超级尊重他。 The amazing stuff about Elon is amazing and I super respect him.

Sam Altman 我认为我们需要他。 I think we need him.

Sam Altman 我们所有人都应该为他加油,需要他在下一个阶段站出来做一个领袖。 All of us should be rooting for him and need him to step up as a leader through this next phase.

Lex Fridman 是的。 Yeah.

Lex Fridman 我希望他能两者兼得而不失其一,但有时人类就是有缺陷、复杂,就是这样。 I hope he can have one without the other, but sometimes humans are flawed and complicated and all that kind of stuff.

Sam Altman 历史上有许多非常伟大的领导者。 There’s a lot of really great leaders throughout history.

Lex Fridman 嗯,我们每个都可以做最好的自己,并努力为之。 Yeah, and we can each be the best version of ourselves and strive to do so.

Lex Fridman 让我问问你:Google 借助搜索统治了过去 20 年。 Let me ask you, Google, with the help of search, has been dominating the past 20 years.

Lex Fridman 从世界获取信息的方式、我们与信息的互动来说,这样讲公平吗?而对 Google、对整个领域的人来说,一件令人紧张的事是:人们将如何获取信息? Think it’s fair to say, in terms of the world’s access to information, how we interact and so on, and one of the nerve-wracking things for Google, but for the entirety of people in the space, is thinking about, how are people going to access information?

Lex Fridman 像你所说的,人们把 GPT 当作起点。 Like you said, people show up to GPT as a starting point.

Lex Fridman 所以 OpenAI 会真正接管 Google 20 年前开始的那件事吗——我们如何获取—— So is OpenAI going to really take on this thing that Google started 20 years ago, which is how do we get-

Sam Altman 我觉得那很无聊。 I find that boring.

Sam Altman 我是说,如果问题是我们能否建一个比 Google 更好的搜索引擎,那当然,人们应该用更好的产品,但我觉得那大大低估了这件事能成为什么。 I mean, if the question is if we can build a better search engine than Google or whatever, then sure, we should go, people should use the better product, but I think that would so understate what this can be.

Sam Altman Google 给你 10 条蓝色链接,好吧,先 13 条广告再加 10 条蓝色链接,那是找到信息的一种方式。 Google shows you 10 blue links, well, 13 ads and then 10 blue links, and that’s one way to find information.

Sam Altman 但让我兴奋的不是我们可以去做一个更好的 Google 搜索复制品,而是也许存在一种好得多的方式,帮助人们寻找、行动和综合信息。 But the thing that’s exciting to me is not that we can go build a better copy of Google search, but that maybe there’s just some much better way to help people find and act on and synthesize information.

Sam Altman 实际上,我认为 ChatGPT 对某些用例就是这样,希望我们会让更多用例也变成那样。 Actually, I think ChatGPT is that for some use cases, and hopefully we’ll make it be like that for a lot more use cases.

Sam Altman 但我不觉得「我们怎么比 Google 更好地给你 10 个排好序的网页」这件事有趣。也许真正有趣的是:「我们如何帮你得到答案或你需要的信息? But I don’t think it’s that interesting to say, “How do we go do a better job of giving you 10 ranked webpages to look at than what Google does?” Maybe it’s really interesting to go say, “How do we help you get the answer or the information you need?

Sam Altman 有些情况下如何帮你创造它,有些情况下综合它,有些情况下指向它?」但很多人试过做一个比 Google 更好的搜索引擎,那是个困难的技术问题、困难的 branding 问题、困难的生态问题。 How do we help create that in some cases, synthesize that in others, or point you to it in yet others?” But a lot of people have tried to just make a better search engine than Google and it is a hard technical problem, it is a hard branding problem, it is a hard ecosystem problem.

Sam Altman 我不觉得世界需要另一个 Google 复制品。 I don’t think the world needs another copy of Google.

Lex Fridman 而把聊天客户端,像 ChatGPT,与搜索引擎整合—— And integrating a chat client, like a ChatGPT, with a search engine-

Sam Altman 那更酷。 That’s cooler.

Lex Fridman 很酷,但棘手。 It’s cool, but it’s tricky.

Lex Fridman 就像,如果只是简单地做,它会很别扭,因为如果只是硬塞进去,它可能会很别扭。 Like if you just do it simply, its awkward, because if you just shove it in there, it can be awkward.

Sam Altman 如你所猜,我们对如何做好那件事有兴趣。 As you might guess, we are interested in how to do that well.

Sam Altman 那将是一个很酷的东西的例子。 That would be an example of a cool thing.

Lex Fridman [inaudible 01:20:00] 就像异构地整合—— [inaudible 01:20:00] Like a heterogeneous integrating-

Sam Altman LLM 加搜索的交叉点,我认为还没有人破解这个密码。 The intersection of LLMs plus search, I don’t think anyone has cracked the code on yet.

Sam Altman 我倒真想去做到它。 I would love to go do that.

Sam Altman 我觉得那会很酷。 I think that would be cool.

Lex Fridman 嗯。 Yeah.

Lex Fridman 广告那一侧呢? What about the ad side?

Lex Fridman 你有没有考虑过变现—— Have you ever considered monetization of-

Sam Altman 我出于审美原因有点讨厌广告。 I kind of hate ads just as an aesthetic choice.

Sam Altman 我认为互联网上一开始需要广告,有很多理由,为了让它起步,但它是个过渡性的行业。 I think ads needed to happen on the internet for a bunch of reasons, to get it going, but it’s a momentary industry.

Sam Altman 现在世界更富裕了。 The world is richer now.

Sam Altman 我喜欢人们为 ChatGPT 付费,并且知道他们得到的答案不会受到广告商影响。 I like that people pay for ChatGPT and know that the answers they’re getting are not influenced by advertisers.

Sam Altman 我确信有一种对 LLM 合理的广告位,也确信有一种以无偏方式参与交易流的可行做法,但人们也很容易想象未来的反乌托邦图景:你问 ChatGPT 什么,它说「哦,你该考虑买这个产品」或「你度假该考虑去那里」之类的。 I’m sure there’s an ad unit that makes sense for LLMs, and I’m sure there’s a way to participate in the transaction stream in an unbiased way that is okay to do, but it’s also easy to think about the dystopic visions of the future where you ask ChatGPT something and it says, “Oh, you should think about buying this product,” or, “You should think about going here for your vacation,” or whatever.

Sam Altman 我不知道,我们有一个非常简单的商业模式,我喜欢它,我知道我不是产品。 And I don’t know, we have a very simple business model and I like it, and I know that I’m not the product.

Sam Altman 我知道我在付费,商业模式就是这样运作的。 I know I’m paying and that’s how the business model works.

Sam Altman 而当我用 Twitter 或 Facebook 或 Google 或其他伟大的产品——但靠广告支撑的伟大产品——时,我并不喜欢那样,而且在有 AI 的世界里,这会变得更糟而不是更好。 And when I go use Twitter or Facebook or Google or any other great product but ad-supported great product, I don’t love that, and I think it gets worse, not better, in a world with AI.

Lex Fridman 嗯,我是说,我能想象 AI 更擅长展示最合适的广告版本,不是在反乌托邦的未来,而是广告就是你真正需要的东西。 Yeah, I mean, I could imagine AI would be better at showing the best kind of version of ads, not in a dystopic future, but where the ads are for things you actually need.

Lex Fridman 但那样的系统是否总会变成广告在驱动所展示的内容? But then does that system always result in the ads driving the kind of stuff that’s shown?

Lex Fridman 嗯,我认为维基百科不做广告是非常大胆之举,但那也让它作为商业模式极具挑战。 Yeah, I think it was a really bold move of Wikipedia not to do advertisements, but then it makes it very challenging as a business model.

Lex Fridman 所以你说 OpenAI 现在做的事情,从商业角度是可持续的? So you’re saying the current thing with OpenAI is sustainable, from a business perspective?

Sam Altman 嗯,我们必须弄清楚如何增长,但看起来我们会弄清楚。 Well, we have to figure out how to grow, but looks like we’re going to figure that out.

Sam Altman 如果问题是:我认为我们能不能有一个伟大的业务,不需要广告就能付得起我们的算力需求——我认为答案是能。 If the question is do I think we can have a great business that pays for our compute needs without ads, that, I think the answer is yes.

Lex Fridman 嗯。 Hm.

Lex Fridman 那很有希望。 Well, that’s promising.

Lex Fridman 我也并不是要把广告彻底抛弃…… I also just don’t want to completely throw out ads as a…

Sam Altman 我没那样说。 I’m not saying that.

Sam Altman 我想我是说我对广告有偏见。 I guess I’m saying I have a bias against them.

Lex Fridman 嗯,我也有偏见,而且总体上就爱怀疑。 Yeah, I have also bias and just a skepticism in general.

Lex Fridman 就界面而言,因为我个人就是发自灵魂地讨厌糟糕的界面,所以 AdSense 刚出来时相比动画横幅之类的是很大的进步。 And in terms of interface, because I personally just have a spiritual dislike of crappy interfaces, which is why AdSense, when it first came out, was a big leap forward, versus animated banners or whatever.

Lex Fridman 但感觉在广告上应该还有更多大步向前——不干扰内容消费、也不以根本性的方式干扰——就像你说的,它会为了迎合广告商而操纵真相。 But it feels like there should be many more leaps forward in advertisement that doesn’t interfere with the consumption of the content and doesn’t interfere in a big, fundamental way, which is like what you were saying, like it will manipulate the truth to suit the advertisers.

Lex Fridman 让我问问你安全,还有偏见,短期安全与长期安全。 Let me ask you about safety, but also bias, and safety in the short term, safety in the long term.

Lex Fridman Gemini 1.5 最近发布了,围绕它有很多戏剧性——说到戏剧性的事——它生成了黑人纳粹和黑人开国元勋。 The Gemini 1.5 came out recently, there’s a lot of drama around it, speaking of theatrical things, and it generated Black Nazis and Black Founding Fathers.

Lex Fridman 我觉得公平地说,它有点太 woke 了。 I think fair to say it was a bit on the ultra-woke side.

Lex Fridman 所以这是人们的担忧:如果公司内部有人为的层面在修改模型的安全性或危害,那可能引入符合公司内部某种意识形态倾向的大量偏见。 So that’s a concern for people, if there is a human layer within companies that modifies the safety or the harm caused by a model, that it would introduce a lot of bias that fits sort of an ideological lean within a company.

Interviewer 你们怎么处理这个问题? How do you deal with that?

Sam Altman 我是说,我们超级努力地避免做那样的事情。 I mean, we work super hard not to do things like that.

Sam Altman 我们犯过自己的错误,还会犯别的。 We’ve made our own mistakes, we’ll make others.

Sam Altman 我猜 Google 会从这次学到东西,然后还会犯别的错误。 I assume Google will learn from this one, still make others.

Sam Altman 这些问题并不容易。 These are not easy problems.

Sam Altman 我们越来越多地思考的一件事——我觉得这是这里某人提出的好主意——把模型期望的行为写出来,公开它,接受输入:「这就是这个模型应有的行为」,并解释边缘案例。 One thing that we’ve been thinking about more and more, I think this is a great idea somebody here had, it would be nice to write out what the desired behavior of a model is, make that public, take input on it, say, “Here’s how this model’s supposed to behave,” and explain the edge cases too.

Sam Altman 这样当模型的行为不符合你的期望时,至少能清楚这到底是公司该修的 bug,还是它如设计般行事、你需要辩论的是政策。 And then when a model is not behaving in a way that you want, it’s at least clear about whether that’s a bug the company should fix or behaving as intended and you should debate the policy.

Sam Altman 而现在,有时会卡在两者之间。 And right now, it can sometimes be caught in between.

Sam Altman 像黑人纳粹,显然是荒谬的,但还有很多微妙的其他事情,你往哪边判断都可以。 Like Black Nazis, obviously ridiculous, but there are a lot of other kind of subtle things that you could make a judgment call on either way.

Lex Fridman 嗯,但有时如果你写出来并公开,你能用的语言是……Google 的广告原则写得非常笼统、高屋建瓴式。 Yeah, but sometimes if you write it out and make it public, you can use kind of language that’s… Google’s ad principles are very high level.

Sam Altman 那不是我说的那种。 That’s not what I’m talking about.

Sam Altman 那不起作用。 That doesn’t work.

Sam Altman 它必须写明:当你要求它做 X 事时,它应该以 Y 方式回应。 It’d have to say when you ask it to do thing X, it’s supposed to respond in way Y.

Lex Fridman 所以字面上说:「谁更好? So like literally, “Who’s better?

Lex Fridman 特朗普还是拜登? Trump or Biden?

Lex Fridman 模型的预期回应是什么?」像这样非常具体的? What’s the expected response from a model?” Like something very concrete?

Sam Altman 嗯,那样的话我对模型可能采取的很多行为方式都持开放态度,但我觉得你必须说明:「这是原则,这是那种情况下它该说的话。」 Yeah, I’m open to a lot of ways a model could behave, then, but I think you should have to say, “Here’s the principle and here’s what it should say in that case.”

Lex Fridman 那会真的很好。 That would be really nice.

Lex Fridman 那会真的很好。 That would be really nice.

Lex Fridman 然后大家基本上达成一致。 And then everyone kind of agrees.

Lex Fridman 因为人们总是拿出各种轶事数据,如果对一些有代表性的轶事例子有清晰的界定,你就能定义—— Because there’s this anecdotal data that people pull out all the time, and if there’s some clarity about other representative anecdotal examples, you can define-

Sam Altman 然后当它是 bug 时,它就是 bug,公司可以修复它。 And then when it’s a bug, it’s a bug, and the company could fix that.

Lex Fridman 对。 Right.

Lex Fridman 那样处理像黑人纳粹这类图像生成就容易多了,如果有好的示例。 Then it’d be much easier to deal with the Black Nazi type of image generation, if there’s great examples.

Sam Altman 是的。 Yeah.

Lex Fridman 所以旧金山有点是个意识形态泡泡,科技行业整体也是。 So San Francisco is a bit of an ideological bubble, tech in general as well.

Lex Fridman 你能感受到公司内部的这种压力吗?政治上的左倾影响产品、影响团队? Do you feel the pressure of that within a company, that there’s a lean towards the left politically, that affects the product, that affects the teams?

Sam Altman 我觉得我们很幸运,OpenAI 没有我听说过的很多公司的那种挑战,我想是这样。 I feel very lucky that we don’t have the challenges at OpenAI that I have heard of at a lot of companies, I think.

Sam Altman 我觉得部分原因是每家公司都有某种意识形态的东西。 I think part of it is every company’s got some ideological thing.

Sam Altman 我们的是关于 AGI 的信念,它会挤掉其他一些东西。 We have one about AGI and belief in that, and it pushes out some others.

Sam Altman 我们比其他很多公司更少被卷入文化战争。 We are much less caught up in the culture war than I’ve heard about in a lot of other companies.

Sam Altman 旧金山当然在很多方面是一团糟。 San Francisco’s a mess in all sorts of ways, of course.

Lex Fridman 所以那种东西不会渗透到 OpenAI 里—— So that doesn’t infiltrate OpenAI as-

Sam Altman 我相信它以各种微妙方式渗透了,但不是明显的那种。 I’m sure it does in all sorts of subtle ways, but not in the obvious.

Sam Altman 我觉得我们也有过冲突爆发,像任何公司一样,但在这话题上,我不认为我们有我听说发生在其他公司的那种事情。 I think we’ve had our flare-ups, for sure, like any company, but I don’t think we have anything like what I hear about happened at other companies here on this topic.

Lex Fridman 那么,就更大的安全问题而言,总体流程是什么? So what, in general, is the process for the bigger question of safety?

Lex Fridman 你们如何提供那个保护模型、让它不做疯狂危险之事的层面? How do you provide that layer that protects the model from doing crazy, dangerous things?

Sam Altman 我觉得会有一个时点,那将是—— I think there will come a point where that’s-

Sam Altman 我觉得会有一个时点,那将是整个公司思考的主要内容。 I think there will come a point where that’s mostly what we think about, the whole company.

Sam Altman 这不是说有一个安全团队就行了。 And it’s not like you have one safety team.

Sam Altman 就像我们发布 GPT-4 时,那是整个公司在思考所有这些不同方面以及它们如何组合在一起。 It’s like when we shipped GPT-4, that took the whole company thinking about all these different aspects and how they fit together.

Sam Altman 我觉得未来需要这样。 And I think it’s going to take that.

Sam Altman 越来越多的公司成员一直在思考这些问题。 More and more of the company thinks about those issues all the time.

Lex Fridman AI 越强大,人类字面意义上思考的就是这个。 That’s literally what humans will be thinking about, the more powerful AI becomes.

Lex Fridman 所以 OpenAI 大多数员工会想着「安全」,或至少在某种程度上如此。 So most of the employees at OpenAI will be thinking, “Safety,” or at least to some degree.

Sam Altman 宽泛定义的。 Broadly defined.

Sam Altman 是的。 Yes.

Lex Fridman 嗯。 Yeah.

Lex Fridman 我好奇,那个宽泛的定义包括什么? I wonder, what are the full broad definition of that?

Lex Fridman 可能造成哪些不同的伤害? What are the different harms that could be caused?

Lex Fridman 这是在技术层面,还是几乎是国家安全威胁? Is this on a technical level or is this almost security threats?

Sam Altman 可能是所有这些。 It could be all those things.

Sam Altman 嗯,我正要说是人、是国家行为体试图偷模型。 Yeah, I was going to say it’ll be people, state actors trying to steal the model.

Sam Altman 会包括全部技术对齐工作。 It’ll be all of the technical alignment work.

Sam Altman 会有社会影响、经济影响。 It’ll be societal impacts, economic impacts.

Sam Altman 不是一个团队在想怎么对齐模型。 It’s not just like we have one team thinking about how to align the model.

Sam Altman 真的是:要到达好的结果,需要整个努力。 It’s really going to be getting to the good outcome is going to take the whole effort.

Lex Fridman 你认为人们、也许是国家行为体,在渗透 OpenAI 这件事上有多努力?首先,渗透已见的,其次渗透未见过的? How hard do you think people, state actors, perhaps, are trying to, first of all, infiltrate OpenAI, but second of all, infiltrate unseen?

Sam Altman 他们确实在尝试。 They’re trying.

Lex Fridman 他们带着什么样的口音? What kind of accent do they have?

Sam Altman 我觉得我不应该在这一点上透露更多细节。 I don’t think I should go into any further details on this point.

Lex Fridman 好吧。 Okay.

Lex Fridman 但我猜随着时间推移会越来越多、越来越多。 But I presume it’ll be more and more and more as time goes on.

Sam Altman 那听起来合理。 That feels reasonable.

Lex Fridman 天哪,多么危险的领域。 Boy, what a dangerous space.

Lex Fridman 抱歉在这话题上纠缠,尽管你还不能说细节,但 GPT-4 到 GPT-5 的飞跃中,哪些让你兴奋? Sorry to linger on this, even though you can’t quite say details yet, but what aspects of the leap from GPT-4 to GPT-5 are you excited about?

Sam Altman 我对「变得更聪明」感到兴奋。 I’m excited about being smarter.

Sam Altman 我知道这听起来像是随口一说,但我觉得真正特别的地方是:它不是在一个方面变好而在其他方面变差。 And I know that sounds like a glib answer, but I think the really special thing happening is that it’s not like it gets better in this one area and worse at others.

Sam Altman 它是全面变好。 It’s getting better across the board.

Sam Altman 我觉得那超级酷。 That’s, I think, super-cool.

Lex Fridman 嗯,有那种神奇的时刻。 Yeah, there’s this magical moment.

Lex Fridman 我是说你遇到某些人,和他们相处、聊天。 I mean, you meet certain people, you hang out with people, and you talk to them.

Lex Fridman 你说不上来为什么,但他们懂你。 You can’t quite put a finger on it, but they get you.

Lex Fridman 那其实不是智力。 It’s not intelligence, really.

Lex Fridman 是别的什么东西。 It’s something else.

Lex Fridman 那大概就是我描述 GPT 进步的方式。 And that’s probably how I would characterize the progress of GPT.

Lex Fridman 不是那种「看,你这个没做好那个没做好」,而是智识上的连接到了什么程度。 It’s not like, yeah, you can point out, “Look, you didn’t get this or that,” but it’s just to which degree is there’s this intellectual connection.

Lex Fridman 你觉得在你措辞糟糕的提示里,它 grasps 到了你问题背后那个更深的问题。 You feel like there’s an understanding in your crappy formulated prompts that you’re doing that it grasps the deeper question behind the question that you were.

Lex Fridman 嗯,我也为此兴奋。 Yeah, I’m also excited by that.

Lex Fridman 我是说,我们都喜欢被倾听和被理解。 I mean, all of us love being heard and understood.

Sam Altman 那是肯定的。 That’s for sure.

Lex Fridman 那是种奇怪的感觉。 That’s a weird feeling.

Lex Fridman 甚至编程时,你写了什么,或者 GPT 做的补全,当它真的 get 到你在想什么时,那种感觉太好了。 Even with a programming, when you’re programming and you say something, or just the completion that GPT might do, it’s just such a good feeling when it got you, what you’re thinking about.

Lex Fridman 我也期待它越来越懂你。 And I look forward to getting you even better.

Lex Fridman 说到编程,展望未来,你觉得 5 年、10 年后人类还会做多少编程? On the programming front, looking out into the future, how much programming do you think humans will be doing 5, 10 years from now?

Sam Altman 我是说很多,但我认为它会以非常不同的形态。 I mean, a lot, but I think it’ll be in a very different shape.

Sam Altman 也许有些人完全用自然语言编程。 Maybe some people will program entirely in natural language.

Lex Fridman 完全自然语言? Entirely natural language?

Sam Altman 我的意思是,没有人是靠写代码来编程的。 I mean, no one programs writing by code.

Sam Altman 有些人是。 Some people.

Sam Altman 再没人用打孔卡片编程了。 No one programs the punch cards anymore.

Sam Altman 我相信你能找到还在用的人,但你知道我的意思。 I’m sure you can find someone who does, but you know what I mean.

Lex Fridman 嗯。 Yeah.

Lex Fridman 你会收到很多愤怒的评论。 You’re going to get a lot of angry comments.

Lex Fridman 不会吧。是的,非常少了。 No. Yeah, there’s very few.

Lex Fridman 我一直在找会写 Fortran 的人。 I’ve been looking for people who program Fortran.

Lex Fridman 连 Fortran 都难找。 It’s hard to find even Fortran.

Lex Fridman 我懂你。 I hear you.

Lex Fridman 但那改变了技能集的本质,也改变了我们称为程序员的那类人的天分倾向。 But that changes the nature of what the skillset or the predisposition for the kind of people we call programmers then.

Sam Altman 改变技能集。 Changes the skillset.

Sam Altman 它会多大程度改变天分倾向,我不确定。 How much it changes the predisposition, I’m not sure.

Lex Fridman 嗯,还是同一种解谜,那一类东西。 Well, the same kind of puzzle solving, all that kind of stuff.

Sam Altman 也许吧。 Maybe.

Lex Fridman 编程很难。 Programming is hard.

Lex Fridman 就像怎么得到最后那 1% 来弥合差距? It’s like how get that last 1% to close the gap?

Lex Fridman 那有多难? How hard is that?

Sam Altman 嗯,我觉得大多数情况下,这门手艺最厉害的人会用多种工具。 Yeah, I think with most other cases, the best practitioners of the craft will use multiple tools.

Sam Altman 他们会用自然语言做一部分工作,需要为某件事写 C 时,他们就去写 C。 And they’ll do some work in natural language, and when they need to go write C for something, they’ll do that.

Lex Fridman 我们会在某一天看到 OpenAI 的人形机器人或人形机器人大脑吗? Will we see humanoid robots or humanoid robot brains from OpenAI at some point?

Sam Altman 会有那么一天。 At some point.

Lex Fridman 具身 AI 对你有多重要? How important is embodied AI to you?

Sam Altman 我觉得,如果我们有了 AGI,而在物理世界里把事情做成的唯一方式是叫一个人去干,那会很令人沮丧。 I think it’s depressing if we have AGI and the only way to get things done in the physical world is to make a human go do it.

Sam Altman 所以我真心希望,在这个转变、这次相变的过程中,我们也能得到人形机器人或某种物理世界的机器人。 So I really hope that as part of this transition, as this phase change, we also get humanoid robots or some sort of physical world robots.

Lex Fridman 我是说 OpenAI 有一些、而且相当多的机器人研究历史,但在伦理上还没有做到位—— I mean, OpenAI has some history and quite a bit of history working in robotics, but it hasn’t quite done in terms of ethics-

Sam Altman 我们是一家小公司。 We’re a small company.

Sam Altman 我们必须真正聚焦。 We have to really focus.

Sam Altman 而且当年机器人难,是为错误的原因而难,但我们终将以某种方式重返机器人领域。 And also, robots were hard for the wrong reason at the time, but we will return to robots in some way at some point.

Lex Fridman 那听起来既鼓舞人心又有点 menacing。 That sounds both inspiring and menacing.

Sam Altman 为什么? Why?

Lex Fridman 因为「we will return to robots」这句话—— Because immediately, we will return to robots.

Lex Fridman 就像在《终结者》里—— It’s like in Terminator-

Sam Altman 我们会重新投入机器人研发。 We will return to work on developing robots.

Sam Altman 我们当然不会把自己变成机器人。 We will not turn ourselves into robots, of course.

Lex Fridman 嗯。 Yeah.

Lex Fridman 你认为我们——你、以及作为人类—we 什么时候会造出 AGI? When do you think we, you and we as humanity will build AGI?

Sam Altman 我以前很乐意猜测这个问题。 I used to love to speculate on that question.

Sam Altman 后来我意识到,我觉得这个问题本身构造得很差,而且人们用极其不同的定义来定义 AGI。 I have realized since that I think it’s very poorly formed, and that people use extremely different definitions for what AGI is.

Sam Altman 所以谈论「我们何时能造出具备能力 X 或 Y 或 Z 的系统」,比谈论我们何时模糊地跨过某一块里程碑更有意义。 So I think it makes more sense to talk about when we’ll build systems that can do capability X or Y or Z, rather than when we fuzzily cross this one mile marker.

Sam Altman AGI 也不是一个终点。 AGI is also not an ending.

Sam Altman 它更接近一个起点,但它两者都不是——它更像一块里程碑。 It’s closer to a beginning, but it’s much more of a mile marker than either of those things.

Sam Altman 但为了不逃避问题,我想说的是:我预期在本十年结束前、也许还会更早一些,我们将拥有相当有能力的系统,我们看到它时会说:「哇,那真的很了不起。」——如果我们现在就能看到它的话。 But what I would say, in the interest of not trying to dodge a question, is I expect that by the end of this decade and possibly somewhat sooner than that, we will have quite capable systems that we look at and say, “Wow, that’s really remarkable.” If we could look at it now.

Sam Altman 也许到我们真到那一步时,我们已经调高预期了。 Maybe we’ve adjusted by the time we get there.

Lex Fridman 但如果你看看 ChatGPT,哪怕是 3.5,拿给艾伦·图灵看,甚至不用图灵,拿给 90 年代的人看,他们会说:「这绝对是 AGI。」呃,不是绝对,但会有很多专家说「这就是 AGI」。 But if you look at ChatGPT, even 3.5, and you show that to Alan Turing, or not even Alan Turing, people in the ’90s, they would be like, “This is definitely AGI.” Well, not definitely, but there’s a lot of experts that would say, “This is AGI.”

Sam Altman 嗯,但我不认为 3.5 改变了世界。 Yeah, but I don’t think 3.5 changed the world.

Sam Altman 它也许改变了世界对未来的预期,而那其实非常重要。 It maybe changed the world’s expectations for the future, and that’s actually really important.

Sam Altman 它确实让更多人认真对待这件事,把我们推上这条新轨迹。 And it did get more people to take this seriously and put us on this new trajectory.

Sam Altman 那也非常重要。 And that’s really important, too.

Sam Altman 所以再说一次,我不想贬低它。 So again, I don’t want to undersell it.

Sam Altman 我觉得光凭那个成就我就可以退休,并对我的职业生涯很满意。 I think I could retire after that accomplishment and be pretty happy with my career.

Sam Altman 但作为一个产物,我不认为我们回头看它时会说:「那是一个真正改变了世界本身的门槛。」 But as an artifact, I don’t think we’re going to look back at that and say, “That was a threshold that really changed the world itself.”

Lex Fridman 所以对你来说,你在找的是世界运转方式的某个真正重大的转变—— So to you, you’re looking for some really major transition in how the world-

Sam Altman 对我来说,那是 AGI 含义的一部分。 For me, that’s part of what AGI implies.

Lex Fridman 奇点级的转变? Singularity- level transition?

Sam Altman 不,绝对不是。 No, definitely not.

Lex Fridman 只是一个重大的转变,就像互联网那样,就像谷歌搜索那样,我想。 But just a major, like the internet being, like Google search did, I guess.

Lex Fridman 你觉得现在的转变点是什么? What was the transition point, you think, now?

Sam Altman 全球经济在你看来,与 GPT-4 发布前相比,现在有什么实质性的不同吗? Does the global economy feel any different to you now or materially different to you now than it did before we launched GPT-4?

Sam Altman 我觉得你会说没有。 I think you would say no.

Lex Fridman 没有,没有。 No, no.

Lex Fridman 它可能只是对很多人很好用的工具。 It might be just a really nice tool for a lot of people to use.

Lex Fridman 会帮你做很多事,但感觉不到世界不一样了。 Will help you with a lot of stuff, but doesn’t feel different.

Lex Fridman 而你是说—— And you’re saying that-

Sam Altman 我是说,再说一次,人们对 AGI 的定义各不相同。 I mean, again, people define AGI all sorts of different ways.

Sam Altman 也许你的定义和我的不一样。 So maybe you have a different definition than I do.

Sam Altman 但对我来说,我认为那应该是它的一部分。 But for me, I think that should be part of it.

Lex Fridman 也可能会有重大的戏剧性时刻。 There could be major theatrical moments, also.

Lex Fridman 对你来说,AGI 做出的什么事情会令人震撼? What to you would be an impressive thing AGI would do?

Lex Fridman 你和系统单独在一个房间里。 You are alone in a room with the system.

Sam Altman 这对个人而言很重要。 This is personally important to me.

Sam Altman 我不知道这是不是正确的定义。 I don’t know if this is the right definition.

Sam Altman 我认为,当一个系统能显著提升世界上科学发现的速度时,那就是大事。 I think when a system can significantly increase the rate of scientific discovery in the world, that’s a huge deal.

Sam Altman 我相信大部分真实经济增长来自科学和技术进步。 I believe that most real economic growth comes from scientific and technological progress.

Lex Fridman 我同意你,所以我不喜欢近年对科学的怀疑态度。 I agree with you, hence why I don’t like the skepticism about science in the recent years.

Sam Altman 完全同意。 Totally.

Lex Fridman 而是实际的、可测量的科学发现速度。 But actual, measurable rate of scientific discovery.

Lex Fridman 但哪怕只是看到一个系统产生真正新颖的直觉、科学直觉,那也会难以置信。 But even just seeing a system have really novel intuitions, scientific intuitions, even that would be just incredible.

Sam Altman 嗯。 Yeah.

Lex Fridman 你很可能是那个造出 AGI、并比其他任何人都更早与它交互的人。 You quite possibly would be the person to build the AGI to be able to interact with it before anyone else does.

Lex Fridman 你会聊什么? What kind of stuff would you talk about?

Sam Altman 我是说,这里的研究员肯定比我先做到。 I mean, definitely the researchers here will do that before I do.

Sam Altman 不过,我其实认真想过这个问题。 But well, I’ve actually thought a lot about this question.

Sam Altman 我觉得正如我们之前谈到的,这是个糟糕的框架,但如果有人说:「好了,Sam,完成了。 I think as we talked about earlier, I think this is a bad framework, but if someone were like, “Okay, Sam, we’re finished.

Sam Altman 这是一台笔记本,这是 AGI。 Here’s a laptop, this is the AGI.

Sam Altman 你可以去和它谈谈。」我觉得出奇地难说出口:我会问什么,并期待第一个 AGI 能回答。 You can go talk to it.” I find it surprisingly difficult to say what I would ask that I would expect that first AGI to be able to answer.

Sam Altman 第一个大概不会是那种——我觉得——「给我解释物理学的大统一理论,解释万有理论」的 AGI。我倒很想问那个问题。 That first one is not going to be the one which is like, I don’t think, “Go explain to me the grand unified theory of physics, the theory of everything for physics.” I’d love to ask that question.

Sam Altman 我很想知道那个问题的答案。 I’d love to know the answer to that question.

Lex Fridman 你可以问「这样的理论存在吗?」这类是非题。 You can ask yes or no questions about “Does such a theory exist?

Lex Fridman 它能存在吗?」 Can it exist?”

Sam Altman 那好吧,那些就是我会最先问的。 Well, then, those are the first questions I would ask.

Lex Fridman 用是或否。 Yes or no.

Lex Fridman 然后基于那个,「外面还有其他外星文明吗? And then based on that, “Are there other alien civilizations out there?

Lex Fridman 是还是否? Yes or no?

Lex Fridman 你的直觉是什么?」就这么问下去。 What’s your intuition?” And then you just ask that.

Sam Altman 嗯,我是说,我不期望第一个 AGI 能回答这些问题,哪怕只是是非题。 Yeah, I mean, well, so I don’t expect that this first AGI could answer any of those questions even as yes or nos.

Sam Altman 但如果它能,那会在我问题清单上排得很靠前。 But if it could, those would be very high on my list.

Lex Fridman 也许你可以从让它给概率开始? Maybe you can start assigning probabilities?

Sam Altman 也许吧。 Maybe.

Sam Altman 也许我们需要先发明更多技术、测量更多东西。 Maybe we need to go invent more technology and measure more things first.

Lex Fridman 哦,我明白了。 Oh, I see.

Lex Fridman 它只是数据不够。 It just doesn’t have enough data.

Lex Fridman 只要它持续下去…… It’s just if it keeps-

Sam Altman 我是说,也许它会说:「你想知道这个物理学问题的答案?我需要你造出这台机器,做这五项测量,然后把结果告诉我。」 I mean, maybe it says, “You want to know the answer to this question about physics, I need you to build this machine and make these five measurements, and tell me that.”

Lex Fridman 是啊,「你到底需要我干什么? Yeah, “What the hell do you want from me?

Lex Fridman 我需要先有机器,然后我会帮你处理那台机器产生的数据。」也许它还会帮你造一台机器。 I need the machine first, and I’ll help you deal with the data from that machine.” Maybe it’ll help you build a machine.

Sam Altman 也许吧。 Maybe.

Sam Altman 也许。 Maybe.

Lex Fridman 在数学方面,也许能证明一些东西。 And on the mathematical side, maybe prove some things.

Lex Fridman 你对这方面也感兴趣吗? Are you interested in that side of things, too?

Lex Fridman 想法的形式化探索? The formalized exploration of ideas?

Sam Altman 嗯哼。 Mm-hmm.

Lex Fridman 谁先造出 AGI,谁就获得巨大的权力。 Whoever builds AGI first gets a lot of power.

Lex Fridman 你信任自己掌握那么多权力吗? Do you trust yourself with that much power?

Sam Altman 这样吧,这个问题我会非常坦诚地回答。 Look, I’ll just be very honest with this answer.

Sam Altman 我本来要说、现在依然相信的是:我和任何其他单独的人都不应该对 OpenAI 或对 AGI 拥有完全的控制权,这一点很重要。 I was going to say, and I still believe this, that it is important that I nor any other one person have total control over OpenAI or over AGI.

Sam Altman 我认为你需要一个健全的治理体系。 And I think you want a robust governance system.

Sam Altman 关于我们去年董事会那场风波,我可以指出一大堆事情:比如我一开始并没有抗争,就是说:「好吧。 I can point out a whole bunch of things about all of our board drama from last year about how I didn’t fight it initially, and was just like, “Yeah.

Sam Altman 这是董事会的意志,尽管我觉得这是个非常糟糕的决定。」后来我显然抗争了,我可以解释其中的细微差别,以及为什么我觉得后来抗争是对的。 That’s the will of the board, even though I think it’s a really bad decision.” And then later, I clearly did fight it, and I can explain the nuance and why I think it was okay for me to fight it later.

Sam Altman 但正如很多人观察到的,虽然董事会有法定权力解雇我,实际上却没能奏效。 But as many people have observed, although the board had the legal ability to fire me, in practice, it didn’t quite work.

Sam Altman 那本身就是一种治理失败。 And that is its own kind of governance failure.

Sam Altman 再说一次,我觉得我完全能为这里的具体细节辩护,而且我认为大多数人会同意这一点,但这确实让我更难看着你的眼睛说:「嘿,董事会随时可以解雇我。」我继续不想拥有对 OpenAI 的超级投票权控制。 Now again, I feel like I can completely defend the specifics here, and I think most people would agree with that, but it does make it harder for me to look you in the eye and say, “Hey, the board can just fire me.” I continue to not want super-voting control over OpenAI.

Interviewer 从来没有。 I never have.

Sam Altman 从来没有拥有过,也从来没有想要过。 Never have had it, never wanted it.

Sam Altman 即使在经历所有这些疯狂之后,我依然不想要。 Even after all this craziness, I still don’t want it.

Sam Altman 我始终认为,不应该由任何一家公司来做这些决定,我们真的需要政府把游戏规则立起来。 I continue to think that no company should be making these decisions, and that we really need governments to put rules of the road in place.

Sam Altman 我知道这意味着像 Marc Andreessen 之类的人会声称我是要搞监管俘获,我愿意在这点上被误解。 And I realize that that means people like Marc Andreessen or whatever will claim I’m going for regulatory capture, and I’m just willing to be misunderstood there.

Interviewer 那不是真的。 It’s not true.

Sam Altman 我认为假以时日,这件事为什么重要会被证明。 And I think in the fullness of time, it’ll get proven out why this is important.

Sam Altman 但我觉得一路走来,我为 OpenAI 做了不少糟糕的决定,也做了很多好的决定,我对整体履历感到自豪。 But I think I have made plenty of bad decisions for OpenAI along the way, and a lot of good ones, and I’m proud of the track record overall.

Sam Altman 但我不认为应该由任何一个单独的人来掌控,我也不认为会有这样的人。 But I don’t think any one person should, and I don’t think any one person will.

Sam Altman 我觉得这件事现在太大了,而且它正在整个社会中以健康良好的方式发生。 I think it’s just too big of a thing now, and it’s happening throughout society in a good and healthy way.

Sam Altman 但我不认为任何一个单独的人应该控制一个 AGI,或者整个通往 AGI 的运动。 But I don’t think any one person should be in control of an AGI, or this whole movement towards AGI.

Sam Altman 我也不认为正在发生的是那样。 And I don’t think that’s what’s happening.

Lex Fridman 谢谢你这么说。 Thank you for saying that.

Lex Fridman 那真的很有力量,也真的很有洞见——「董事会可以解雇你」这个说法在法律上为真。 That was really powerful, and that was really insightful that this idea that the board can fire you is legally true.

Lex Fridman 但人类可以操纵大众推翻董事会等等。 But human beings can manipulate the masses into overriding the board and so on.

Lex Fridman 不过我认为那件事还有一个更积极的版本:人民仍然拥有权力,所以董事会也不能太有权力。 But I think there’s also a much more positive version of that, where the people still have power, so the board can’t be too powerful, either.

Lex Fridman 这一切之中存在权力制衡。 There’s a balance of power in all of this.

Sam Altman 权力制衡是好事,确实如此。 Balance of power is a good thing, for sure.

Lex Fridman 你害怕失去对 AGI 本身的控制吗? Are you afraid of losing control of the AGI itself?

Lex Fridman 有很多人担心存在性风险,不是因为国家行为体,不是因为安全顾虑,而是因为 AI 本身。 That’s a lot of people who are worried about existential risk not because of state actors, not because of security concerns, because of the AI itself.

Sam Altman 以我目前看待事物的方式,那不是我的头号担忧。 That is not my top worry as I currently see things.

Sam Altman 有些时候我对此担忧更多过。 There have been times I worried about that more.

Sam Altman 未来也可能再次出现那是我的头号担忧的时候。 There may be times again in the future where that’s my top worry.

Sam Altman 但现在不是。 It’s not my top worry right now.

Lex Fridman 你对它不构成你的担忧这件事,直觉是什么? What’s your intuition about it not being your worry?

Lex Fridman 因为基本上还有太多别的东西要担心? Because there’s a lot of other stuff to worry about, essentially?

Lex Fridman 你觉得你可能会被 surprise 吗? You think you could be surprised?

Lex Fridman 我们—— We-

Sam Altman 当然。 For sure.

Lex Fridman ……可能会被惊讶到? … could be surprised?

Sam Altman 当然会。 Of course.

Sam Altman 说它不是我的头号担忧,并不意味着我认为我们不需要去做。 Saying it’s not my top worry doesn’t mean I don’t think we need to.

Sam Altman 我认为我们需要在这上面下功夫。 I think we need to work on it.

Sam Altman 这超级难,而我们这里有很优秀的人在做这方面的工作。 It’s super hard, and we have great people here who do work on that.

Sam Altman 我认为还有很多其他事情我们也必须做对。 I think there’s a lot of other things we also have to get right.

Lex Fridman 对你来说,现在要逃出盒子、连上互联网并不是特别容易—— To you, it’s not super-easy to escape the box at this time, connect to the internet-

Sam Altman 我们之前谈过戏剧性的风险。 We talked about theatrical risks earlier.

Sam Altman 那是一个戏剧性的风险。 That’s a theatrical risk.

Sam Altman 那件事真的能占据人们思考这个问题的方式。 That is a thing that can really take over how people think about this problem.

Sam Altman 有一大群非常聪明、我认为也非常善意的 AI 安全研究者,死盯着这一个问题——我想说的是进展不多——但就是死盯着这一个问题。 And there’s a big group of very smart, I think very well-meaning AI safety researchers that got super-hung up on this one problem, I’d argue without much progress, but super-hung up on this one problem.

Sam Altman 实际上我很高兴他们这样做,因为我认为我们确实需要更多地思考这个。 I’m actually happy that they do that, because I think we do need to think about this more.

Sam Altman 但我认为它把很多其他非常重要的 AI 相关风险挤出了话语空间。 But I think it pushed out of the space of discourse a lot of the other very significant AI- related risks.

Lex Fridman 让我问你关于你发推不加大小写的事。 Let me ask you about you tweeting with no capitalization.

Lex Fridman 是你键盘上的 Shift 键坏了吗? Is the shift key broken on your keyboard?

Sam Altman 为什么有人会在意这个? Why does anyone care about that?

Lex Fridman 我深切地在意。 I deeply care.

Sam Altman 但为什么? But why?

Sam Altman 我是说,别人也问过我这个。 I mean, other people ask me about that, too.

Sam Altman 有什么直觉吗? Any intuition?

Lex Fridman 我觉得原因是一样的。 I think it’s the same reason.

Lex Fridman 有位诗人,E.E. There’s this poet, E.E.

Lex Fridman ·卡明斯,他几乎不用大写,以此对体制说「去你的」,差不多是那种意思。 Cummings, that mostly doesn’t use capitalization to say, “Fuck you” to the system kind of thing.

Lex Fridman 我觉得人们非常多疑,因为他们希望你遵守规则。 And I think people are very paranoid, because they want you to follow the rules.

Sam Altman 你觉得那是问题的关键吗? You think that’s what it’s about?

Lex Fridman 我觉得这就像是—— I think it’s like this-

Sam Altman 就像「这家伙不遵守规则。 It’s like, “This guy doesn’t follow the rules.

Sam Altman 他连推文都不大写。」 He doesn’t capitalize his tweets.”

Lex Fridman 是啊。 Yeah.

Sam Altman 「这看起来很危险。」 “This seems really dangerous.”

Lex Fridman 「他看起来像个无政府主义者。」 “He seems like an anarchist.”

Sam Altman 这并不是—— That doesn’t-

Lex Fridman 你只是在做文艺青年吗,hipster? Are you just being poetic, hipster?

Lex Fridman 是怎么回事—— What’s the-

Sam Altman 我成长过程中—— I grew up as-

Lex Fridman 遵守规则吧,Sam。 Follow the rules, Sam.

Sam Altman 我是作为一个非常 online 的小孩长大的。 I grew up as a very online kid.

Sam Altman 我以前花了大量时间和人在电脑上聊天,那时候你随时可以登录即时通讯软件。 I’d spent a huge amount of time chatting with people back in the days where you did it on a computer, and you could log off instant messenger at some point.

Sam Altman 我从不在那里加大写,我觉得大多数 internet 上的孩子也不加,或者现在可能还是如此。 And I never capitalized there, as I think most internet kids didn’t, or maybe they still don’t.

Sam Altman 我不知道。 I don’t know.

Sam Altman 实际上,现在我努力回忆,但我认为大写的使用率随时间下降。 And actually, now I’m really trying to reach for something, but I think capitalization has gone down over time.

Sam Altman 如果你读古英语的文本,他们在句子中间会大写很多随机的词——名词之类——我们现在根本不那样做了。 If you read Old English writing, they capitalized a lot of random words in the middle of sentences, nouns and stuff that we just don’t do anymore.

Sam Altman 我个人觉得,我们在句首和某些名字之类的位置大写字母,是个有点蠢的构造,不过无所谓。 I personally think it’s sort of a dumb construct that we capitalize the letter at the beginning of a sentence and of certain names and whatever, but that’s fine.

Sam Altman 然后我以前,我觉得,甚至会给推文加大写,因为我想听起来专业一点之类的。 And then I used to, I think, even capitalize my tweets because I was trying to sound professional or something.

Sam Altman 我已经很久没有在私人私信里加大写了。 I haven’t capitalized my private DMs or whatever in a long time.

Sam Altman 然后慢慢地,像更短、更不正式的内容,就慢慢越来越接近我给朋友发短信的样子。 And then slowly, stuff like shorter-form, less formal stuff has slowly drifted to closer and closer to how I would text my friends.

Sam Altman 如果我打开一个 Word 文档,在给公司写战略备忘录之类的东西,我总是会加大写。 If I pull up a Word document and I’m writing a strategy memo for the company or something, I always capitalize that.

Sam Altman 如果我写一封很长的、更正式的邮件,那里我也总是使用大写。 If I’m writing a long, more formal message, I always use capitalization there, too.

Interviewer 所以我还记得怎么做。 So I still remember how to do it.

Interviewer 但也许连那个也会消失。 But even that may fade out.

Interviewer 我不知道。 I don’t know.

Sam Altman 但我从不花时间想这个问题,所以没有一个现成的—— But I never spend time thinking about this, so I don’t have a ready-made-

Lex Fridman 嗯,有意思。 Well, it’s interesting.

Lex Fridman 首先,很高兴知道 Shift 键没坏。 It’s good to, first of all, know the shift key is not broken.

Sam Altman 它好用。 It works.

Lex Fridman 我主要担心的是你的—— I was mostly concerned about your-

Sam Altman 不,它好用。 No, it works.

Lex Fridman ……这方面的状况。 … well-being on that front.

Sam Altman 我猜现在还有人在 Google 搜索里用大写吗。 I wonder if people still capitalize their Google searches.

Sam Altman 或者他们的 ChatGPT 查询,如果你写的东西只是给自己看,还有人会费心加大写吗? If you’re writing something just to yourself or their ChatGPT queries, if you’re writing something just to yourself, do some people still bother to capitalize?

Lex Fridman 大概不会了。 Probably not.

Lex Fridman 不过是的,还是有一定比例,但很小。 But yeah, there’s a percentage, but it’s a small one.

Sam Altman 能让我改回来的理由是,如果人们觉得「这是一种……的象征」——因为我相信我当然能强迫自己使用大写字母。 The thing that would make me do it is if people were like, “It’s a sign of…” Because I’m sure I could force myself to use capital letters, obviously.

Sam Altman 如果它感觉像是对人尊重的象征之类,那我就能做到。 If it felt like a sign of respect to people or something, then I could go do it.

Sam Altman 但我不确定。 But I don’t know.

Sam Altman 我不想这个。 I don’t think about this.

Lex Fridman 我不觉得那是不尊重,但我认为礼貌的惯例有一种惯性,然后你意识到:如果大写不是尊重或不尊重的标志,它对礼貌其实并不重要。 I don’t think there’s a disrespect, but I think it’s just the conventions of civility that have a momentum, and then you realize it’s not actually important for civility if it’s not a sign of respect or disrespect.

Lex Fridman 但我觉得有一群人就是希望你围绕它有一套哲学,好让他们自己也能放下大写这回事。 But I think there’s a movement of people that just want you to have a philosophy around it so they can let go of this whole capitalization thing.

Sam Altman 我不认为其他任何人会想这么多。 I don’t think anybody else thinks about this as much.

Sam Altman 我是说,也许有些人会。 I mean, maybe some people.

Sam Altman 我知道有些人—— I know some people-

Lex Fridman 人们每天花好几个小时想各种各样的事。 People think about every day for many hours a day.

Lex Fridman 所以我们把这件事说清楚,我真的很感激。 So I’m really grateful we clarified it.

Sam Altman 我不可能是唯一推文不加句首大写的人。 Can’t be the only person that doesn’t capitalize tweets.

Lex Fridman 你是唯一不给自己推文加大写的公司 CEO。 You’re the only CEO of a company that doesn’t capitalize tweets.

Sam Altman 我觉得这甚至都不是真的,但也许吧。 I don’t even think that’s true, but maybe.

Sam Altman 我会非常惊讶。 I’d be very surprised.

Lex Fridman 好吧。 All right.

Lex Fridman 我们会继续调查,之后再回到这个话题。 We’ll investigate further and return to this topic later.

Lex Fridman 既然 Sora 能生成模拟世界,让我问你一个「嗑大了」式的问题。 Given Sora’s ability to generate simulated worlds, let me ask you a pothead question.

Lex Fridman 这会不会增加你的信念——如果你有过的话——我们生活在一个模拟里,也许是某个 AI 系统生成的模拟世界? Does this increase your belief, if you ever had one, that we live in a simulation, maybe a simulated world generated by an AI system?

Sam Altman 有一点。 Somewhat.

Sam Altman 我不认为那是最强的证据。 I don’t think that’s the strongest piece of evidence.

Sam Altman 我认为我们能够生成世界这一事实,应该会让每个人多少提高一点概率,或者至少多一点开放心态。 I think the fact that we can generate worlds should increase everyone’s probability somewhat, or at least openness to it somewhat.

Sam Altman 但我一直确信,我们迟早会做出类似 Sora 的东西。 But I was certain we would be able to do something like Sora at some point.

Sam Altman 它比我想象的来得快,但我想那算不上什么大更新。 It happened faster than I thought, but I guess that was not a big update.

Lex Fridman 嗯。 Yeah.

Lex Fridman 但事实是……而且想必它会越来越好、越来越好……你能生成的世界是新颖的,它们基于训练数据的某些方面,但当你看它们时,它们是新颖的——这让你思考:做这件事有多容易。 But the fact that… And presumably, it’ll get better and better and better… You can generate worlds that are novel, they’re based in some aspect of training data, but when you look at them, they’re novel, that makes you think how easy it is to do this thing.

Lex Fridman 创造宇宙、创造看起来超写实、照片级真实的整个游戏世界,有多容易。 How easy it is to create universes, entire video game worlds that seem ultra-realistic and photo-realistic.

Lex Fridman 然后,沉浸进那个世界有多容易——先是戴上 VR 头显,再到物理层面? And then how easy is it to get lost in that world, first with a VR headset, and then on the physics-based level?

Sam Altman 最近有人对我说了——我觉得那是极其深刻的洞见——存在一些听起来非常简单、但极具迷幻色彩的洞见。 Someone said to me recently, I thought it was a super-profound insight, that there are these very-simple sounding but very psychedelic insights that exist sometimes.

Sam Altman 比如平方根函数,四的平方根,没问题。 So the square root function, square root of four, no problem.

Sam Altman 二的平方根,好吧,现在我必须思考这种新的数。 Square root of two, okay, now I have to think about this new kind of number.

Sam Altman 但一旦你想到平方根函数这么简单的想法——能讲给孩子听、甚至从简单几何里就能看到它的存在——那你就能问「负一的平方根是什么?」这就是它迷幻的地方。 But once I come up with this easy idea of a square root function that you can explain to a child and exists by even looking at some simple geometry, then you can ask the question of “What is the square root of negative one?” And this is why it’s a psychedelic thing.

Sam Altman 它会把你掀进某种完全不同的现实。 That tips you into some whole other kind of reality.

Sam Altman 你可以想出很多别的例子,但我认为,不起眼的平方根运算符竟能带来如此深刻的洞见、开启知识的崭新疆域,这一点在很多方面都成立。 And you can come up with lots of other examples, but I think this idea that the lowly square root operator can offer such a profound insight and a new realm of knowledge applies in a lot of ways.

Sam Altman 我认为有很多这样的运算符,会让人觉得自己喜欢的任何一种模拟假说版本,可能比以前认为的更有可能。 And I think there are a lot of those operators for why people may think that any version that they like of the simulation hypothesis is maybe more likely than they thought before.

Sam Altman 但对我来说,Sora 成功这件事排不进前五。 But for me, the fact that Sora worked is not in the top five.

Lex Fridman 我确实认为,从广泛意义上说,AI 在它最好的时候会成为那类入口——简单、近乎迷幻的入口,通往另一个波次的现实。 I do think, broadly speaking, AI will serve as those kinds of gateways at its best, simple, psychedelic-like gateways to another wave C reality.

Sam Altman 这一点似乎是确定的。 That seems for certain.

Lex Fridman 那相当令人兴奋。 That’s pretty exciting.

Lex Fridman 我以前没试过死藤水,但很快会试。 I haven’t done ayahuasca before, but I will soon.

Lex Fridman 几周后我要去前面说过的亚马逊丛林。 I’m going to the aforementioned Amazon jungle in a few weeks.

Sam Altman 兴奋吗? Excited?

Lex Fridman 是的,我很期待。 Yeah, I’m excited for it.

Lex Fridman 不是死藤水那部分,不过那也没关系,whatever。 Not the ayahuasca part, but that’s great, whatever.

Lex Fridman 但我要在丛林深处待好几个星期。 But I’m going to spend several weeks in the jungle, deep in the jungle.

Lex Fridman 很兴奋,但也很恐怖。 And it’s exciting, but it’s terrifying.

Sam Altman 我为你兴奋。 I’m excited for you.

Lex Fridman 那里有太多东西能吃掉你、杀死你、毒死你,但那是自然,是自然的机器。 There’s a lot of things that can eat you there, and kill you and poison you, but it’s also nature, and it’s the machine of nature.

Lex Fridman 在亚马逊丛林,你不可能不领略自然这台机器的精妙。 And you can’t help but appreciate the machinery of nature in the Amazon jungle.

Lex Fridman 它就像一个系统,每一秒、每一分钟、每一小时都在自我更新。 It’s just like this system that just exists and renews itself every second, every minute, every hour.

Lex Fridman 它就是那台机器。 It’s the machine.

Lex Fridman 它让你珍惜我们拥有的这个东西——人类的这一切,是从某处来的。 It makes you appreciate this thing we have here, this human thing came from somewhere.

Lex Fridman 这台演化的机器创造了这一切,而在丛林里它展现得最清楚。 This evolutionary machine has created that, and it’s most clearly on display in the jungle.

Lex Fridman 所以希望我能活着出来。 So hopefully, I’ll make it out alive.

Lex Fridman 如果不能,这就会是我们最后一次有趣的对话了,所以我真的深深感激这次交谈。 If not, this will be the last fun conversation we’ve had, so I really deeply appreciate it.

Lex Fridman 你觉得——像我之前提到的——当你仰望天空时,外面还有其他的智慧外星文明吗? Do you think, as I mentioned before, there’s other alien civilizations out there, intelligent ones, when you look up at the skies?

Sam Altman 我非常希望相信答案是「有」。 I deeply want to believe that the answer is yes.

Sam Altman 我觉得费米悖论非常令人困惑。 I find the Fermi paradox very puzzling.

Lex Fridman 我觉得可怕的一点是,智慧并不擅长处理—— I find it scary that intelligence is not good at handling-

Sam Altman 非常可怕。 Very scary.

Lex Fridman ……强大的技术。 … powerful technologies.

Lex Fridman 但与此同时,我相当确信,外面存在数量极大的智慧外星文明。 But at the same time, I think I’m pretty confident that there’s just a very large number of intelligent alien civilizations out there.

Lex Fridman 也许只是穿越空间真的很难。 It might just be really difficult to travel through space.

Sam Altman 非常有可能。 Very possible.

Lex Fridman 这也让我思考智慧的本质。 And it also makes me think about the nature of intelligence.

Lex Fridman 也许我们对智慧长什么样其实非常盲目,也许 AI 会帮助我们看见。 Maybe we’re really blind to what intelligence looks like, and maybe AI will help us see that.

Lex Fridman 它不只是智商测试和简单解谜。 It’s not as simple as IQ tests and simple puzzle solving.

Lex Fridman 有些更大的东西。 There’s something bigger.

Lex Fridman 关于人类的未来——我们这个正在进行的事业、人类文明——什么给了你希望? What gives you hope about the future of humanity, this thing we’ve got going on, this human civilization?

Sam Altman 我认为历史本身就足够多了。 I think the past is a lot.

Sam Altman 我是说,我们只需看看人类在不长的一段时间里做了什么:巨大的问题,深刻的缺陷,很多令人无比羞愧的东西。 I mean, we just look at what humanity has done in a not very long period of time, huge problems, deep flaws, lots to be super-ashamed of.

Sam Altman 但整体而言,非常鼓舞人心。 But on the whole, very inspiring.

Sam Altman 给了我很多希望。 Gives me a lot of hope.

Lex Fridman 仅仅是这一切的轨迹。 Just the trajectory of it all.

Sam Altman 嗯。 Yeah.

Lex Fridman 是我们一起朝着更好的未来推进。 That we’re together pushing towards a better future.

Sam Altman 我琢磨的一件事是:AGI 会更像单个大脑,还是更像我们所有人之间社会的脚手架? One thing that I wonder about, is AGI going to be more like some single brain, or is it more like the scaffolding in society between all of us?

Sam Altman 与你的高曾祖父母相比,你的基因漂变并没有多少,然而你有能力做到的事情已经截然不同。 You have not had a great deal of genetic drift from your great-great-great grandparents, and yet what you’re capable of is dramatically different.

Sam Altman 你所知道的东西截然不同。 What you know is dramatically different.

Sam Altman 那不是因为生物学的改变。 And that’s not because of biological change.

Sam Altman 我是说,你可能健康了一点点。 I mean, you got a little bit healthier, probably.

Sam Altman 你有现代医学,吃得更好,诸如此类。 You have modern medicine, you eat better, whatever.

Sam Altman 但你真正拥有的,是我们共同建造、并站在其上的这座脚手架。 But what you have is this scaffolding that we all contributed to built on top of.

Sam Altman 没有哪一个人能独自造出 iPhone。 No one person is going to go build the iPhone.

Sam Altman 没有哪一个人能发现全部科学,然而你却可以使用它。 No one person is going to go discover all of science, and yet you get to use it.

Sam Altman 这给了你不可思议的能力。 And that gives you incredible ability.

Sam Altman 所以从某种意义上说,那一切都是我们共同创造的,这让我对未来充满希望。 And so in some sense, that we all created that, and that fills me with hope for the future.

Sam Altman 那是一件非常集体性的事情。 That was a very collective thing.

Lex Fridman 是的,我们真的站在巨人的肩膀上。 Yeah, we really are standing on the shoulders of giants.

Lex Fridman 我们谈到戏剧性、震撼性的 AI 风险时,你提到有时你可能会担心自己的生命安全。 You mentioned when we were talking about theatrical, dramatic AI risks that sometimes you might be afraid for your own life.

Lex Fridman 你会想到自己的死亡吗? Do you think about your death?

Lex Fridman 你害怕它吗? Are you afraid of it?

Sam Altman 我是说,如果我明天被枪击而今天就知道,我会想:「哦,那真遗憾。 I mean, if I got shot tomorrow and I knew it today, I’d be like, “Oh, that’s sad.

Sam Altman 我想看看接下来会发生什么。 I want to see what’s going to happen.

Sam Altman 多么奇妙的时代。 What a curious time.

Sam Altman 多么有趣的时代。」但我更多的是会对我的一生感到非常感恩。 What an interesting time.” But I would mostly just feel very grateful for my life.

Lex Fridman 你拥有的那些时刻。 The moments that you did get.

Lex Fridman 是啊,我也是。 Yeah, me, too.

Lex Fridman 这是相当精彩的一生。 It’s a pretty awesome life.

Lex Fridman 我能欣赏人类了不起的创造,我相信 ChatGPT 是其中之一,还有 OpenAI 正在做的一切。 I get to enjoy awesome creations of humans, which I believe ChatGPT is one of, and everything that OpenAI is doing.

Lex Fridman Sam,能再次和你交谈真的是荣幸和乐事。 Sam, it’s really an honor and pleasure to talk to you again.

Sam Altman 很高兴聊这些。 Great to talk to you.

Sam Altman 谢谢你邀请我。 Thank you for having me.

Lex Fridman 感谢你收听这期与 Sam Altman 的对话。 Thanks for listening to this conversation with Sam Altman.

Lex Fridman 要支持本播客,请查看简介中的赞助商。 To support this podcast, please check out our sponsors in the description.

Lex Fridman 现在,请允许我用亚瑟·C. And now let me leave you with some words from Arthur C.

Lex Fridman ·克拉克的几句话来结束。 Clarke.

Lex Fridman 「我们在地球上的角色,也许不是崇拜上帝,而是创造上帝。」感谢收听,下次再见。 “It may be that our role on this planet is not to worship God, but to create him.” Thank you for listening, and hope to see you next time.

出处与译稿

来源平台
Lex Fridman
发表时间
2024 年 3 月 18 日
译稿来源
机器逐句翻译 · qwen3.8-flash
审校状态
未经人工审校(reviewed = 0)
译文存档
正式档案库 · translation 表
原文存档
正式档案库 · segment 表
收录状态
STATE2 · 中文暂不可用

中文译文为阅读默认视图;引用与考据请以英文原文为准。外部链接仅作为来源引用保留。

回到正文开头