别再盲目焦虑了！《经济学人》专访AI 大佬吴恩达：AI 不会
字幕摘录
| 时间 | 英文 | 中文 |
|---|---|---|
| 0:00 | there will be no AI job apocalypse. | 没有AI的工作启示。 |
| 0:03 | This idea of 50 percent being out of work, | 关于50%失业的想法, |
| 0:06 | 50 percent people being out of work, | 50%的人失业, |
| 0:07 | maybe rioting industries, I don't know. | 也许是骚乱行业 我不知道 |
| 0:10 | I don't see any realistic path to that. | 我没有看到任何现实的路 |
| 0:13 | In fact, early data, | 事实上,早期的数据, |
| 0:14 | even out of some of the organizations that were | 甚至从一些组织中 |
| 0:16 | hyping up the 50 percent unemployment job apocalypse scenarios, | 虚构了50%的失业 工作启示, |
| 0:21 | are showing that AI is not having | 表明大赦国际没有 |
| 0:24 | the negative impact on jobs that people were fearing. | 给人们担心的工作带来的负面影响。 |
展开字幕全文（130 条）
| 序号 | 英文 | 中文 |
|---|---|---|
| 1 | there will be no AI job apocalypse. | 没有AI的工作启示。 |
| 2 | This idea of 50 percent being out of work, | 关于50%失业的想法, |
| 3 | 50 percent people being out of work, | 50%的人失业, |
| 4 | maybe rioting industries, I don't know. | 也许是骚乱行业 我不知道 |
| 5 | I don't see any realistic path to that. | 我没有看到任何现实的路 |
| 6 | In fact, early data, | 事实上,早期的数据, |
| 7 | even out of some of the organizations that were | 甚至从一些组织中 |
| 8 | hyping up the 50 percent unemployment job apocalypse scenarios, | 虚构了50%的失业 工作启示, |
| 9 | are showing that AI is not having | 表明大赦国际没有 |
| 10 | the negative impact on jobs that people were fearing. | 给人们担心的工作带来的负面影响。 |
| 11 | Because AI ultimately is only part of someone's job, | 因为AI最终只是某人的工作的一部分, |
| 12 | it turns out we really need people to do the rest of their job still. | 结果我们真的需要人来完成剩下的工作 |
| 13 | What happens in software engineering is a predictor of | 在软件工程中发生的事情是预测 |
| 14 | what will happen in other job sectors as well. | 其他工作部门也将发生什么。 |
| 15 | I think it'll be fantastic and if anything, | 我认为这将是美妙的,如果有什么, |
| 16 | we can't find enough software engineers right now, | 我们现在找不到足够的软件工程师, |
| 17 | and I'm already seeing we can't find enough AI native marketers, | 我已经看到我们找不到足够的 AI本地营销员, |
| 18 | or enough AI native salespeople, | 或足够的AI本地推销员, |
| 19 | enough AI native operations people for the challenges of skilling. | 足够的AI本土操作人员应对技能挑战. |
| 20 | Because for all of these job roles, | 因为所有这些工作角色 |
| 21 | we have to help people learn new skills, | 我们必须帮助人们学习新的技能, |
| 22 | and I don't just mean learn AI skills, | 我不是说学人工智能 |
| 23 | but it turns out that maybe I'll take one example of software engineering. | 但事实证明,也许我会拿一个软件工程的例子。 |
| 24 | A lot of people used to be maybe still are front-end engineers, | 过去很多人也许还是前端工程师 |
| 25 | meaning they build a website or an app to make it look good. | 意思是他们建立一个网站或一个应用程序 来让它看起来不错。 |
| 26 | But AI is pretty good at front-end engineering. | 但AI相当擅长前端工程. |
| 27 | A lot of front-end developers are now full-set developers, | 许多前端开发者现在都是完全设置的开发者, |
| 28 | meaning they don't just build the part that users look at, | 意思是他们不只是构建 用户看的部分, |
| 29 | but they build what runs in the background as well, | 但他们建造了 运行在背景, |
| 30 | and they build the mobile app, they build security things. | 他们制造移动应用程序, 他们制造安全的东西。 |
| 31 | A lot of developers are now taking on broader scope in their work. | 现在许多开发者正在他们的作品中占据更广泛的范围. |
| 32 | What that means is, instead of just focusing on what it looks like, | 这意味着, 而不是仅仅专注于它是什么样子, |
| 33 | they have to learn to build all these other pieces as well, | 他们必须学会 建造其他的棋子, |
| 34 | and they have to learn AI skills. | 他们必须学会人工智能 |
| 35 | When they do that, they're amazingly valuable in the job market, | 当他们这样做, 他们在就业市场 惊人的价值, |
| 36 | and we just can't find enough of these people. | 我们只是找不到足够的这些人。 |
| 37 | I'm seeing early signs of a similar transformation happening as well | 我也看到类似转变的早期迹象 |
| 38 | in other job categories. | 其他职类。 |
| 39 | I'm curious, though, you say there's not going to be a jobs apocalypse. | 不过我很好奇 你说这不会是启示录 |
| 40 | I'm curious what that implies about your sense of the eventual progress of AI, | 我很好奇你对AI最终进展的感知是什么 |
| 41 | because the fear about a jobs apocalypse isn't the AI we have in 2026, | 因为对一个启示录的恐惧 并不是2026年的人工智能 |
| 42 | it's the AI we have in 2030, or if your timelines are short, | 这是2030年的人工智能, 或者如果你的时间表很短, |
| 43 | it's the AI we have in six months, right? | 这是半年后的人工智能,对吧? |
| 44 | Few people are arguing that the AI today can do every task | 很少有人认为今天的人工智能可以完成每一项任务 |
| 45 | in a job like software engineering, | 在像软件工程这样的工作, |
| 46 | but the fear is that if we get this compounding rate of progress | 但恐惧是 如果我们得到这种复杂的进步速度 |
| 47 | for another year or two, it's hard to see where it stops. | 再过一两年 很难看它停在哪里 |
| 48 | Are you implicitly arguing that you think it will stop, | 你是在暗示你认为会停止吗? |
| 49 | that we will hit a hard limit at some sort of your scaling laws, | 我们将会在你的缩放定律上 达到极限, |
| 50 | or some architectural change will be needed, | 或者需要一些建筑改变, |
| 51 | and there will be some tasks that the AI systems of tomorrow | 还有明天的人工智能系统 |
| 52 | simply will not be able to do and probably never will be able to do? | 根本无法做到,而且可能永远也无法做到? |
| 53 | Well, I don't think we'll hit a hard limit, | 好吧,我不认为我们会达到一个硬极限, |
| 54 | but progress will be incremental for a long time to come. | 但是,今后很长一段时间,进展将是渐进的。 |
| 55 | So I would love if AI could do any intellectual task a human can, | 所以我希望AI能完成人类所能完成的任何智力任务, |
| 56 | maybe go on holiday or something. | 也许去度假什么的。 |
| 57 | But realistically, that would take multiple novel breakthroughs, | 但实际上,这需要 多个新的突破, |
| 58 | and I'd love to see it within our lifetime, | 我很乐意在一生中看到它 |
| 59 | but whether it'll take decades or even longer than decades to get there, | 但无论要花几十年甚至更长的时间才能到达那里, |
| 60 | I don't know. | 师曰. |
| 61 | And until AI could do 100% of what people can do, | 直到AI能做100%的人能做的, |
| 62 | which seems incredibly far away, I don't think we'll get there anytime, | 似乎离得很远,我觉得我们随时都到不了那里 |
| 63 | certainly not in six months. | 当然不是在6个月。 |
| 64 | Almost very unlikely to get there by 2030. | 到2030年几乎不可能实现。 |
| 65 | There's still plenty of work for people to do. | 仍令众生多业. |
| 66 | And use the term compounding progress. | 并用"复合进步"一词. |
| 67 | One of the terms of hype in Silicon Valley | 硅谷的杂音词之一 |
| 68 | has been compounding progress or recursive improvements and so on. | 使进展或递归性改进等复杂化。 |
| 69 | And at least right now, that's much more hype than reality. | 至少现在,这比现实要多得多。 |
| 70 | For decades, computers have made computers better. | 几十年来,计算机使计算机变得更好. |
| 71 | I mean, we used the last generation of semiconductor microchips | 我们用了上一代半导体微芯片 |
| 72 | to build the current generation of semiconductor microchips. | 用于构建当前一代半导体微芯片. |
| 73 | And yes, at the Frontier Labs, | 是的 在边境实验室 |
| 74 | AI writes all the code or almost all the code. | AI写出所有代码或几乎所有代码. |
| 75 | So of course, AI is helping us make better AI. | 当然,AI帮助我们改进AI。 |
| 76 | But it turns out that if we look at the bottlenecks of AI progress, | 但事实证明 如果我们看看AI进步的瓶颈 |
| 77 | a lot of bottlenecks are we have to build new GPU servers, | 有很多瓶颈是我们必须建造新的GPU服务器, |
| 78 | we have to get power plants built, | 我们必须建造发电厂, |
| 79 | we have to run a lot of experiments to see what works. | 我们必须做很多实验,看看什么是可行的。 |
| 80 | So unfortunately, even super intelligent AI, if we can get there, | 所以不幸的是,即使超级智能AI, 如果我们能到达那里, |
| 81 | needs to obey the laws of physics. | 需要遵守物理规律 |
| 82 | And the bottlenecks are the time to build data centers | 瓶颈是建立数据中心的时候 |
| 83 | and run experiments to figure out what to do, | 并进行实验 找出该怎么做, |
| 84 | rather than just thinking on the box. | 而不是仅仅思考在盒子。 |
| 85 | Then those are fundamental bottlenecks, | 那么这些是根本的瓶颈 |
| 86 | because the bottlenecks of more intelligence | 因为更多的智慧的瓶颈 |
| 87 | is not just to be smarter about things, | 不只是要更聪明的事情, |
| 88 | that these are the bottlenecks as well that seem very difficult to overcome. | 这些也是似乎很难克服的瓶颈。 |
| 89 | If you're not concerned then about a jobs apocalypse, | 如果你不关心一个末日的工作, |
| 90 | are you still concerned about some narrow impacts? | 你还在担心一些狭窄的撞击吗? |
| 91 | I'm thinking firstly of the impact on early stage careers. | 我首先想到对早期职业的影响. |
| 92 | You brought up your Stanford colleague, Bryn Jolson, | 你提起你的斯坦福同事布林乔尔森 |
| 93 | who did some fantastic research last year, | 去年他做了一些了不起的研究, |
| 94 | showing that it seems like the more exposed a sector is to AI, | 显示一个部门受到AI的影响越大 |
| 95 | the worse job prospects are for people entering that sector. | 进入这个部门的人的工作前景最差。 |
| 96 | That research was pretty convincing to me, | 这研究对我来说很有说服力, |
| 97 | and it seems to be borne out anecdotally, | 这似乎是传闻的, |
| 98 | that getting a job as an entry-level software engineer | 作为初级软件工程师的工作 |
| 99 | with no skills, no previous work history, | 没有技能,没有以前的工作经历, |
| 100 | is getting quite hard, | 越来越难了 |
| 101 | because you don't have much to offer in a way that you may have, | 因为你没有太多 以你可能拥有的方式, |
| 102 | there were roles of just being a code monkey | 有个角色只是个密码猴子 |
| 103 | while you learn for a couple of years. | 你学了几年 |
| 104 | Eric's Canary paper was very thoughtful. | 埃里克的加那利论文非常周到. |
| 105 | I think he actually has new data since then | 我觉得他从那时起 有了新的数据 |
| 106 | that paints a more optimistic picture. | 这描绘了更乐观的画面。 |
| 107 | To take software engineering again as an example, | 再次以软件工程为例 |
| 108 | I think the challenge is not the lack of jobs, | 我认为挑战不是缺乏工作 |
| 109 | the challenge is the slowness of our educational system | 挑战在于我们教育系统的缓慢 |
| 110 | to adapt, to train graduates to do the jobs that are now in demand. | 适应,培训毕业生完成目前需要的工作。 |
| 111 | So I think it's really unfortunate that today in many universities | 所以,我认为这是非常不幸的, 今天在许多大学 |
| 112 | it's still possible to get a bachelor's degree in computer science | 计算机科学的学士学位还是有可能的 |
| 113 | without having made a single API call or network call | 在没有发出一个API呼叫或网络呼叫的情况下 |
| 114 | to a frontier AI model. | 一个前沿AI模型。 |
| 115 | And today I'm not going to hire a software engineer | 今天我不会请软件工程师 |
| 116 | that's never done that. | 从来没有做到这一点。 |
| 117 | It's still possible to get a bachelor's degree | 学士学位还是有可能的 |
| 118 | in computer science in some universities | 一些大学的计算机科学 |
| 119 | without using an AI coding agent, | 没有使用AI编码代理, |
| 120 | and frankly, none of my teams will hire a software engineer | 坦白说,我的团队中没有一个会雇用软件工程师 |
| 121 | that doesn't know how to use an AI coding agent. | 无法使用 AI 编码代理。 |
| 122 | So to me, the challenge is not this job apocalypse, | 所以对我来说,挑战 并不是这个工作启示录, |
| 123 | we don't need junior people. | 我们不需要年轻人。 |
| 124 | Oh, over this last summer, my team had a lot of interns, | 去年夏天 我的团队有很多实习生 |
| 125 | including high school students, current college students, fresh grads, | 包括高中生、现任大学生、研究生、 |
| 126 | and they were great. | 他们是伟大的。 |
| 127 | They did a ton of work. | 他们做了很多工作。 |
| 128 | I'm really glad I had all these interns, | 我很高兴有这么多实习生 |
| 129 | but the trick is I hired only AI-native interns. | 关键是我只雇了人工智能实习生 |
| 130 | So the jobs are there, but we've got to fix the educational system. | 因此,工作是在那里, 但我们必须修复教育系统。 |
该视频共有字幕 130 条。解锁更多字幕为会员功能，请移动到 价格