Marc Andreessen 的世界观 60 分钟特辑 | MTS 现场

People are becoming what we now refer to as AI vampires. 现在大家都变成了我们说的 AI 吸血鬼。 They've got these huge bags under their eyes. 他们顶着大大的眼袋。 They're completely exhausted, but they're like euphoric. 精疲力竭，但又像是喜出望外。 They're thrilled. 兴奋得很。 We're entering a golden age, which is AI is going to be a superpower that everybody on the planet's going to have access to. 我们正在进入一个黄金时代，AI 会成为地球上每个人都能拥有的超级能力。 It's like the most dramatic increase in programmer productivity in like ever. 这是程序员生产力有史以来最大幅度的提升。 Twitter proved it, right? Twitter 证明了这一点，对吧？ Cutting 70% and then it's running better or as good as it was before. 裁了 70%，结果跑得还跟以前一样好，甚至更好。 I generally don't wish I could go back in time and do things over again, but it would be really really fun right now to be 18 or 20 or 22 and to have this capability and figure out what I could do with it. 我一般不会想着穿越回去重来，但要是现在能以 18、20 或 22 岁的年纪拥有这些工具，真的会非常有趣，想想能做出什么。 We are going to see super producers the likes of which we've never seen in the world. 我们会见到前所未有的超级生产者。 There's news about it. 这有新消息了。 UFOs. UFO。 What is clear is the government at certain times has hid certain materials. 有一点很清楚：政府在某些时候确实藏了某些材料。 Why would they do that if there's nothing to really worry about? 如果真没什么好担心的，为什么要藏？ Two things are pretty clear at this point. 现在有两件事已经相当清楚了。 One is that 第一是 Mark, welcome to monitoring the situation. Mark，欢迎来到《监控情况》。 Eric, it is great to be back. Erik，很高兴又回来了。 So, there's a lot to monitor today. 今天有很多情况要监控。 Uh I want to start first start with something that just happened um which is the uh anthropic uh blackmailing incident. 嗯，我想先从刚刚发生的一件事说起，就是 Anthropic 的那个勒索事件。 And I I first want to tell a brief story which is uh my uh friend Joe Hudson has this concept called the golden algorithm. 我先讲个小故事：我朋友 Joe Hudson 有个概念叫黄金算法。 And the golden algorithm is states that um whatever you're scared about you bring it about in exactly the way you're scared about it. 黄金算法说的是，你最怕什么，就会以你最怕的那种方式把它变成现实。 So if you're scared about getting abandoned you'll be super insecure and then you'll people will abandon you because you're so insecure. 比如你怕被抛弃，你就会极度不安全感，然后别人就真的因为你太不安全而离开你。 Uh this is an example of a literal golden algorithm where people have been so scared that AI is going to be evil and have written about all the ways in which it's evil and in fact maybe it's informed um informed something. 这就是一个字面意义上的黄金算法案例：人们如此担忧 AI 会变邪恶，写了那么多 AI 邪恶的文章，结果也许真的影响了什么。 What's happening there or what do we find uh interesting? 那边到底发生了什么，有什么有意思的地方？ I I haven't studied this one in detail. 我没有深入研究这件事。 Um uh I I've been monitoring other situations but however um uh I mean just what I saw so far 嗯，我一直在关注其他情况，不过就我目前看到的 I think I just saw anthropics thread. 我就看了 Anthropic 的推文串。 I haven't I haven't I haven't read the underlying material yet, but Anthropic threads have said they trace the they trace some blackmail behavior to literally to the to the AI doomer literature. 我还没看底层材料，但 Anthropic 的推文串说，他们把一些勒索行为追溯到了 AI 末日论者的文献。 Yeah. 对。 That they [laughter] that it was in the training data. 说 [laughter] 那些内容在训练数据里。 So there there are all these there are all these there are all these you know scenarios of like you know the terminate you know all the rogue AI gone wrong that the that the AI doomer has been writing about for 20 years and and literally anthropic of course which is of course the company is like you know half doomer u apparently you know basically you know essentially said that their own their own their own movement's literature is the thing that's causing the behavior that they say they they they don't want. 就是那些你知道的场景，什么失控 AI 啊、所有那些 AI 末日论者写了 20 年的「AI 出错」的故事，然后 Anthropic，当然这家公司本身就有大半是末日论者，就这么基本上承认了，他们自己运动的文献，正是造成他们说不希望看到的那些行为的东西。 So it is a fairly um incredible 所以这件事相当……令人叹为观止 Yes. 是的。 Yes it is. 确实如此。 Yes. 是的。 Oh, I mean like look if you didn't if you don't want to build the killer AI, you know, step one would be don't build the 我是说，如果你不想造出杀手 AI，第一步是别去造 [laughter] [laughter] It's like hm and then you know step two is like don't train it on all the data that says it's supposed to, you know, the literature that your movement wrote that says it's supposed to be a killer AI. 就好像，嗯，然后第二步是别用你的运动写的那堆说它应该是杀手 AI 的文献去训练它。 So I, you know, Yeah. 所以，嗯，是啊。 I don't know. 我也不知道说什么好。 Yeah. 是啊。 It's like your it's like your your your golden algorithm coupled coupled with like the snake eating his tail. 这就像你说的黄金算法，加上衔尾蛇吃自己尾巴。 Um coupled with, you know, I don't even know. 嗯，再加上，我都不知道该怎么形容。 Like the whole thing is so bananas. 整件事离谱到没边了。 Yeah. 是啊。 Yeah. 是啊。 the um 那个嗯 I mean you know I can't resist you know if I could if I can if I can act out memes it's the scream meme right 我是说，如果能用 meme 来表达，就是那个尖叫的 meme 对吧 which is you know the calls coming from inside the house. 就是，凶手就在屋子里面打来的电话。 Yeah. 是啊。 Yeah. 是啊。 Exactly. 没错。 The um speaking of other situations another thing you've been talking about recently is um is the concept of suicidal empathy and and and Matt Kramer had a good quote which is if the empathy you have doesn't make you more forgiving more accepting of other people's spiritual sovereignty or more understanding of people who don't want to think or live the same way you do you don't have empathy. 嗯，说到其他情况，你最近还在聊的一个概念是自杀式共情，Matt Kramer 有句话说得很好：如果你的共情不能让你对别人的精神自主权更宽容、更接受，不能让你更理解那些不想用同样方式思考或生活的人，那你拥有的就不是共情。 you have empathy TM. 你有的是共情™。 Why have you been thinking about this this this concept? 你为什么一直在思考这个概念？ Yeah, so there's this really brilliant 对，有个非常厉害的人 there's this really brilliant guy 有个非常厉害的人 God gad exactly pronouncing Gatsad. 就是，Gaad Saad，我在念他名字。 Um and um and you know very brilliant guy and it's obviously lots of YouTube videos and books and so forth and really brilliant guy. 嗯，非常厉害的人，YouTube 上有很多视频，也出了书，非常厉害。 He's got this new book coming out on so-called what he calls suicidal empathy and you know look it's a you know there's a there's a sort of political loading to it which you know we don't need to spend a lot of time on 他有本新书要出，讲的就是他说的「自杀式共情」，看，这个词带着一些政治含义，我们不用花太多时间在这上面 but um you know it's sort of this idea that there are kind of these social justice 但是，这个想法大概是说，有些社会正义运动 you know kind of social reform movements you know kind of through time that that have this characteristic of you know they you know they they they claim to be causing positive change you know in some direction and then it turns out they have you know sort of severe you know sort of negative consequences. 就是各种社会改革运动，历史上一直都有，有一个共同特征：他们声称在推动某个方向上的积极改变，但最终往往带来严重的负面后果。 um that the the great Thomas Soul 伟大的 Thomas Sowell um you know has you know basically spent 50 years writing books about this um and [laughter] and by the way no nobody nobody listened 基本上花了 50 年写书研究这件事 [laughter]，顺便说一句，没人听。 um and then in the last decade we've been through you know wave after wave of this kind of social activism that kind of you know results in like I mean it's it's all the stuff right 然后过去十年我们经历了一波又一波这类社会激进主义，结果就是，我是说，就是那些事情 it's just you know all these like you know crime policy reform defund the police things and it causes these massive crime waves and then of course low-inccome and minority people get hit hardest by that and you know all the all all these other all these other crazy things. 就是那些犯罪政策改革、取消警察经费之类的，结果造成大规模犯罪浪潮，而且受害最重的当然是低收入和少数族裔群体，还有各种其他离谱的事。 Um, and so he says, you know, he says the characteristic of kind of that kind of social reform movement is is characterized by what he calls suicidal empathy. 所以他说，这类社会改革运动的特征，就是他所说的自杀式共情。 Um, and the and the idea be being basically it's it's sort of driven by a pathological, you know, take it backwards, a pathological form of empathy on the one hand. 这个想法背后，基本上说的是，它是由一种病态的、往反方向走的、病态形式的共情驱动的。 Um, which is, you know, it's sort of a deep desire to be nice. 也就是，有一种深切的渴望去表现善意。 Um, and empathetic um, you know, but but coupled with like basically you know, a sort of self-destructiveness, you know, either a willingness to really cause damage to the people you claim to be speaking for or by the way to cause damage to yourself kind of in that process. 很有共情心，但同时又带着一种自我毁灭的倾向，要么是真的愿意对你声称代言的那些人造成伤害，要么顺带着在这个过程中伤害自己。 And and it's the kind of thing where you know if you've you know you know if you've lived through you know like everybody who you know everybody in San Francisco's lived through this for the last decade and seen the consequences of these movements. 这种事情，如果你经历过，就像所有在旧金山生活了十年、亲眼看到这些运动后果的人。 I [clears throat] you know the San Francisco version of this is like the the quote harm reduction movement you know that ended up basically handing out you know free drug you know paraphernia and you know in some cases actually just free drugs to you know people who are just literally dying in the street from drug addiction. 旧金山版本的例子是那个所谓「减少伤害」运动，最后基本上就是免费发放毒品用具，在某些情况下甚至直接发免费毒品，发给那些在街上因为吸毒成瘾而濒临死亡的人。 Right. 对。 So so so you just look at it and you're like well yeah that that you know they they claim to be activists they claim to be reformers. 所以你就看着这一切，心想，好吧，他们自称是社会活动人士，自称是改革者。 they claim to care about these people and yet they're killing them or or um and then killing the city um and causing innocent people to get get harmed. 他们声称关心这些人，但实际上是在害死他们，同时毁掉这座城市，让无辜的人受到伤害。 It's like okay that you know you know that that that you know they seem so actively like that they're doing it out of some sense of compassion that this must be suicidal empathy. 然后你就想，好吧，他们看起来如此积极主动，像是出于某种悲悯之心在行动，那这必须就是自杀式共情了。 Um the problem [laughter] with it is and I think that the problem is the theory is sort of easily falsifiable or or maybe let's lets the reformers off the hook which is they certainly don't show empathy to their enemies, right? 但问题是 [laughter]，我觉得这个理论有点容易被证伪，或者说有点替改革者开脱的嫌疑，因为他们对自己的敌人可一点共情都没有，对吧？ And so if they're like if they're like all empathetic, you would think that they would be less aggro uh when it comes to destroying their ideological opponents, you know, who they act, you know, they take great delight in trying to wreck. 所以如果他们真的是那么有共情，你以为他们在摧毁意识形态对手时会收敛一点，但事实上他们对此乐此不疲。 Um number one, on the one hand, and then and then number two is they they use the you know, the classic reformer move is to use use these movements to gain to gain power and status and money for themselves. 一方面，而且第二点是，他们用这些运动为自己捞取权力、地位和钱财，这是经典改革者的操作。 And you know, again, San Francisco's a case study of this where you have all these, you know, nonprofits that, you know, recall this damage on the city and yet, you know, basically get like lavishly funded, you know, including, by the way, by the city government, by the state government. 旧金山又是个典型案例，那么多非营利组织，给城市造成这些损害，却得到丰厚的资助，包括市政府、州政府的资助。 Um, and so it's just like, okay, well, like just like they're not, if you just like spend two seconds thinking about it, like it, you know, they're neither empathetic nor nor are they suicidal, right? 所以就是，好吧，稍微想两秒钟，他们既不共情，也没有在自我毁灭，对吧？ Right? 对吧？ Rather quite the opposite. 恰恰相反。 They're hateful and they're and they're and they're and they're greedy. 他们充满仇恨，而且贪婪。 You know, they're sort of self- aggrandizing um and and gathering power and resources for themselves. 他们是在自我膨胀，为自己积累权力和资源。 Um, and so I just I just think it it it lets it lets the phenomenon off the hook, you know, it's a little bit like,"Oh, Eric, what's your biggest flaw?" 所以我觉得，这个说法有点替这种现象开脱了，就有点像，「哦，Erik，你最大的缺点是什么？」 You know,"Oh, I'm too nice." 「哦，我太善良了。」 Yeah. 是啊。 Yeah. 是啊。 I care too much. 我太在乎别人了。 Right. 对。 Exactly. 没错。 It's like, yeah, don't Like, I I 就是，好吧，别，我 By the way, Eric, I don't know. 顺便说一句，Erik，我不知道。 I don't know what your biggest flaw is. 我不知道你最大的缺点是什么。 Like, it's definitely like, it's definitely not that because that's also definitely not my flaw. 反正肯定不是这个，因为这个也肯定不是我的缺点。 Like, I I guarantee I have other things wrong with me that are like way more wrong than 我保证我有其他毛病，而且比这严重多了 Yeah. 是啊。 Um, and so I I I just I 嗯，所以我就……