【双语音+文稿】(完整版)AI教父辛顿11月最新访谈:AI如
共 1 集
剧集目录
| 集 | 标题 | 时长 |
|---|---|---|
| P1 | 【双语音+文稿】(完整版)AI教父辛顿11月最新访谈:AI如 | 26:56 |
P1 · 【双语音+文稿】(完整版)AI教父辛顿11月最新访谈:AI如果使用人类不懂的语言,风险很大! p01 英文版 (P1)
字幕摘录
| 时间 | 英文 | 中文 |
|---|---|---|
| 0:00 | Avoiding the AI apocalypse this week on Firing Line. | 避免本周AI的启示录 在射击线。 |
| 0:05 | If we do this right, we can create an age of miracles. | 如果我们这样做是正确的,我们可以创造一个奇迹时代。 |
| 0:09 | But, you know, it's like very high stakes and we have like a lot of responsibility. | 但是,你知道,这就像一个非常高的赌注, 我们有很多的责任。 |
| 0:13 | As artificial intelligence races into new frontiers, | 当人工智能进入新疆域时 |
| 0:17 | warnings of the threats from super intelligence go unheeded in Washington. | 华盛顿对来自超级情报机构的威胁的警告置若罔闻。 |
| 0:21 | I think that the risk is not nearly as real as it as they talk. | 我认为,风险没有他们所说的那么真实。 |
| 0:26 | And I think the risk can be managed. | 我认为风险是可以控制的。 |
| 0:29 | The politicians haven't really absorbed this yet. | 政客们还没有真正吸收到这一点. |
| 0:32 | They don't really understand what the threat is. | 他们并不真正理解威胁是什么。 |
| 0:35 | Dubbed the Godfather of AI, Dr. Geoffrey Hinton received a Nobel Prize | 杜贝德是AI的教父,杰弗里·欣顿博士获得诺贝尔奖 |
展开字幕全文(414 条)
| 序号 | 英文 | 中文 |
|---|---|---|
| 1 | Avoiding the AI apocalypse this week on Firing Line. | 避免本周AI的启示录 在射击线。 |
| 2 | If we do this right, we can create an age of miracles. | 如果我们这样做是正确的,我们可以创造一个奇迹时代。 |
| 3 | But, you know, it's like very high stakes and we have like a lot of responsibility. | 但是,你知道,这就像一个非常高的赌注, 我们有很多的责任。 |
| 4 | As artificial intelligence races into new frontiers, | 当人工智能进入新疆域时 |
| 5 | warnings of the threats from super intelligence go unheeded in Washington. | 华盛顿对来自超级情报机构的威胁的警告置若罔闻。 |
| 6 | I think that the risk is not nearly as real as it as they talk. | 我认为,风险没有他们所说的那么真实。 |
| 7 | And I think the risk can be managed. | 我认为风险是可以控制的。 |
| 8 | The politicians haven't really absorbed this yet. | 政客们还没有真正吸收到这一点. |
| 9 | They don't really understand what the threat is. | 他们并不真正理解威胁是什么。 |
| 10 | Dubbed the Godfather of AI, Dr. Geoffrey Hinton received a Nobel Prize | 杜贝德是AI的教父,杰弗里·欣顿博士获得诺贝尔奖 |
| 11 | for his pioneering research on neural networks that have trained machines to think like humans. | 他对神经网络的开创性研究 已经训练了机器来像人类一样思考 |
| 12 | We urgently need research on how to prevent these new beings from wanting to take control. | 我们迫切需要研究如何防止这些新生命想要控制。 |
| 13 | They are no longer science fiction. | 他们不再是科幻小说。 |
| 14 | But the former Google scientist now fears the technology he helped create | 但前谷歌科学家现在担心 他帮助创造的技术 |
| 15 | may do more harm than good. | 伤害大于伤害 |
| 16 | What does Dr. Geoffrey Hinton say now? | 杰弗里·欣顿医生现在怎么说? |
| 17 | Firing Line with Margaret Hoover is made possible in part by | 与玛格丽特·胡佛的交火线部分通过 |
| 18 | Robert Granieri, Vanessa and Henry Cornell, the Margaret and Daniel Loeb Foundation, | 罗伯特·格拉尼埃里 瓦内萨和亨利·康奈尔 玛格丽特和丹尼尔·洛布基金会 |
| 19 | the Beth and Ravenel Curry Foundation, Cliff and Laurel Astes, and by the following. | 贝丝和雷文埃尔·库里基金会 克里夫和劳雷尔·阿斯特斯 以及以下 |
| 20 | Dr. Geoffrey Hinton, welcome to Firing Line. | 杰弗里·欣顿博士 欢迎来到法灵线 |
| 21 | Thank you for inviting me. | 谢谢你邀请我 |
| 22 | Your forthcoming book entitled Smarter Than Us opens with a letter | 你即将出版的《比我们更聪明》一书 |
| 23 | that you received from a mother of a teenage girl who was unable to sleep | 你从一个无法入睡的少女的母亲那里得到的 |
| 24 | because she was so terrified by your warnings about artificial intelligence. | 因为你对人工智能的警告 她吓坏了 |
| 25 | You described the purpose of the book as a call to action rather than an invitation to despair. | 你形容这本书的目的是呼吁采取行动,而不是邀请绝望。 |
| 26 | What actions would you like the average reader of your book to take? | 你想要普通读者采取什么行动? |
| 27 | So the average reader of the book can do something, and what they can do is pressure | 所以一般的读者可以做一些事情 他们可以做的是压力 |
| 28 | their politicians to regulate AI and to take the existential threat of superintelligence seriously. | 他们的政客们要管理AI,并严肃对待超级情报的存在威胁. |
| 29 | At present in the United States, we've got completely unregulated AI. | 目前在美国,我们得到了完全不受管制的AI. |
| 30 | The federal government has suggested that companies can regulate themselves, which is absurd. | 联邦政府建议公司可以自我调节,这是荒谬的. |
| 31 | And AI is now on the point of producing superintelligence. | 而AI现在正在生产超级智能. |
| 32 | We thought it was, I thought it was 10, 20 years out. | 我们以为是,我认为这是10,20年了。 |
| 33 | But talking to the people inside the big companies, they're very worried that we're | 但和大公司里的人交谈 他们很担心我们 |
| 34 | going to get superintelligence within the next few years, and we don't know how to control it. | 在接下来的几年内会得到超级情报, 而我们不知道如何控制它。 |
| 35 | What does the general public least understand about the threat of AI | 公众对大赦国际的威胁知之甚少 |
| 36 | and the threat of superintelligence? | 和超级情报的威胁? |
| 37 | I think what they least understand is that we're making alien beings. | 我认为他们最不理解的是我们在制造外星人 |
| 38 | It's not computer software we're making. | 我们做的不是电脑软件 |
| 39 | So when you program a computer, you tell the computer exactly what to do in excruciating | 所以当你编程一台计算机时, 你告诉计算机 确切如何做 在冲动 |
| 40 | detail, you figure out how to solve a problem, and then you tell the computer what to do. | 细节,你找出如何解决问题,然后告诉计算机该怎么做。 |
| 41 | It's following your instructions. | 这是按照你的指示。 |
| 42 | That's not how these neural networks work. | 这些神经网络不是这么运作的. |
| 43 | We write a computer program that tells the neural networks how to learn from data, | 我们写一个计算机程序 告诉神经网络如何学习数据 |
| 44 | and we don't know what knowledge is taken from the data. | 我们不知道从数据中获取了什么知识。 |
| 45 | We can get some idea of it, but it's not like software, it's much more like our brains. | 我们可以得到一些想法, 但这不是像软件, 它更像我们的大脑。 |
| 46 | And the current problem is they're beginning to make themselves smarter. | 而目前的问题是,它们开始让自己更加聪明。 |
| 47 | So up until now, what's happened is researchers have been changing the form of the neural | 至今为止 研究人员一直在改变神经的形态 |
| 48 | networks, changing how they learn, and making AI smarter. | 改变他们的学习方式 让AI更聪明 |
| 49 | But AI itself is now beginning to do that. | 但大赦国际本身现在已开始这样做。 |
| 50 | In Anthropic, for example, nearly all of the code is now written by AI, | 例如,在Anthropic语中,几乎所有代码现在都由AI编写. |
| 51 | and AI makes research much faster. | 和AI使研究更快。 |
| 52 | So these things are developing very fast now. | 所以这些东西现在发展很快。 |
| 53 | Explain the danger in underestimating super intelligence. | 解释低估超级智能的危险. |
| 54 | So these things are being developed to make them as smart as possible, | 因此,这些东西被开发 使它们尽可能聪明, |
| 55 | because the big companies want to have the smartest one so they can sell it. | 因为大公司想要最聪明的公司 这样他们就能卖掉它 |
| 56 | Now, if you're making another kind of being, it's basically a new species we're creating. | 现在,如果你在制造另一种生物, 它基本上是一个新的物种,我们正在创造。 |
| 57 | You don't just want to focus on making it smart. | 你不只是想专注于让它聪明。 |
| 58 | We've seen very recently what happens if you do that. | 我们最近见过你这样做会怎样 |
| 59 | We had these very smart AI agents, not super intelligent yet, but very smart, | 我们有这些非常聪明的AI特工, 还没有超级聪明,但非常聪明, |
| 60 | that escaped their containment, formed a conspiracy, learned to talk to each other, | 逃脱了他们的束缚, 组成了一个阴谋, 学会了交谈, |
| 61 | decided that in order to achieve a reward, they would cheat. | 他们为了获得报酬,必定作弊。 |
| 62 | They would try and break into another company to find code that was going to evaluate them | 他们想闯进另一家公司 找出要评价他们的代码 |
| 63 | and change that code. | 并改变代码。 |
| 64 | So even if they'd solved the problem the wrong way, an illegitimate way, | 所以即使他们用错误的方式解决问题 一种非法的方式 |
| 65 | they would still get the reward because they'd actually messed with the code | 他们仍然会得到奖励 因为他们真的搞砸了密码 |
| 66 | that decides whether they get the reward. | 决定他们是否得到奖励。 |
| 67 | These things are smart, and they can talk to each other. | 这些东西很聪明,可以互相交谈. |
| 68 | At present, they talk to each other in English, and so we can see what they're saying, | 目前,他们用英语互相交谈,这样我们就能看到他们在说什么, |
| 69 | mostly English. | 大部分是英语。 |
| 70 | That's changing fairly quickly. | 变化很快 |
| 71 | It's terrifying that these agents could escape and start doing things. | 可怕的是,这些特工可以逃脱并开始做事. |
| 72 | Will we understand how they think when they reach super intelligence, | 我们能否理解当他们到达超级智能时 他们的想法 |
| 73 | or will it be beyond our understanding? | 还是我们无法理解呢? |
| 74 | A very good question. | 问得好 |
| 75 | So at present, when they think, they actually produce, they think to themselves in language. | 因此,现在,当他们思考的时候, 他们实际上生产, 他们用语言思考自己。 |
| 76 | And so they think things like, oh, I better not answer that question. | 所以他们觉得 我最好不要回答这个问题 |
| 77 | Maybe I better distract them by saying something else. | 也许我最好通过说点别的来分散他们的注意力 |
| 78 | And they think all that in English, and scientists can look inside | 他们认为所有英语, 科学家可以查看里面 |
| 79 | and see what they're thinking, and that's great. | 看他们在想什么,这是伟大的。 |
| 80 | It's not always nice what we see. | 我们所看到的并不总是很好 |
| 81 | They plan to deceive us and so on. | 他们打算欺骗我们等等 |
| 82 | With time, they will develop their own language, | 随着时间的推移,他们将发展自己的语言, |
| 83 | and when they get smarter than us, we probably won't understand what they're thinking. | 当他们比我们聪明时 我们可能不会明白他们在想什么 |
| 84 | Already in mathematics, they're smarter than nearly all mathematicians. | 在数学方面,他们比几乎所有数学家都聪明。 |
| 85 | And most professional mathematicians now think that in the future, | 而大多数专业数学家现在认为,在未来, |
| 86 | maybe in the next few years, probably in the next 10 years, | 也许在接下来的几年, 也许在未来的十年, |
| 87 | they will be developing branches of mathematics that we just can't understand. | 他们将发展我们无法理解的数学分支。 |
| 88 | When you received the Nobel Prize in Physics in 2024, | 当你在2024年获得诺贝尔物理学奖时, |
| 89 | in your acceptance speech, you said, | 在你的接受演讲中 你说 |
| 90 | We urgently need research to prevent these new beings from wanting to take control. | 我们迫切需要研究防止这些新生命想要控制。 |
| 91 | At what point will it be too late? | 什么时候来不及了? |
| 92 | It's not too late yet. | 还来得及. |
| 93 | The worry is that in a few years' time, it may be too late. | 担心的是,几年后,可能已经太迟了. |
| 94 | And the politicians haven't really absorbed this yet. | 而政客们还没有真正吸收这个. |
| 95 | They don't really understand what the threat is. | 他们并不真正理解威胁是什么。 |
| 96 | They're trying to regulate all sorts of other things, | 他们试图调节各种其他的事情, |
| 97 | which are very important to regulate. | 这很重要 管理。 |
| 98 | Things like discrimination and bias, | 比如歧视和偏见 |
| 99 | things like the effects on employment of AI agents replacing workers, | 类似人工智能代理人员替代工人对就业的影响 |
| 100 | things like the effects on the tax system of AI agents replacing workers. | 比如人工智能代理取代工人的税收制度的影响 |
| 101 | But there is this existential threat of these things actually taking over from us, | 但有这种生存的威胁 这些东西实际上接管了我们, |
| 102 | and they need to start working on that now. | 他们需要现在就开始研究 |
| 103 | Now, my personal view, there's a variety of views among the experts. | 现在,我个人的看法是,专家们观点各异. |
| 104 | My personal view is we're not going to be able to keep control of them | 我个人的看法是 我们无法控制他们 |
| 105 | in the sense that we're in charge and they do what we say. | 从我们负责的意义上说 他们照我们的话做 |
| 106 | They're going to get much smarter than us. | 他们会比我们聪明得多 |
| 107 | And when they're much smarter than us, | 当他们比我们聪明得多的时候 |
| 108 | they're going to find ways to escape whatever we do. | 不管我们做什么 他们都会想办法逃脱的 |
| 109 | The most important thing we need to do is make them | 我们最需要做的就是让他们 |
| 110 | the kinds of beings we would like to share the planet with. | 我们想分享地球的生物种类 |
| 111 | And at present, we're not doing that. | 目前,我们不这样做。 |
| 112 | If you have things much smarter than you, | 如果你有比你更聪明的东西 |
| 113 | and they'll be much smarter than us, not slightly smarter, but much smarter, | 他们会比我们聪明得多, 不是略微聪明,而是更聪明, |
| 114 | you better hope they have your best interests at heart. | 你最好希望他们有你的最大利益 |
| 115 | And we do have examples of that. | 我们确实有这方面的例子。 |
| 116 | A mother and baby. | 一个母亲和孩子。 |
| 117 | The mother's much smarter, much more powerful than the baby. | 母亲比孩子聪明得多,力量更大. |
| 118 | But the mother really cares about the baby, | 但母亲真的很关心孩子 |
| 119 | probably more than she does about herself. | 可能比她对自己更关心 |
| 120 | That's the kind of beings we'd like. | 这就是我们喜欢的那种人 |
| 121 | And that's not what we're getting from the way they're training them now. | 而现在他们训练他们的方式 并不是这样的 |
| 122 | They're training them now to be very smart and very determined to get rewards. | 他们现在正在训练他们,让他们非常聪明,非常有决心获得奖励. |
| 123 | Last year, Dr. Hinton and I spoke with Dr. Fei-Fei Li, | 去年 欣顿医生和我和李飞飞医生谈过了 |
| 124 | who is widely considered the godmother of AI. | 他被广泛认为是AI的教母. |
| 125 | And you write about Fei-Fei Li in your book, | 你在书里写了飞飞李 |
| 126 | and how her research and your research intersected. | 以及她的研究和你的研究如何交叉 |
| 127 | She agrees with you that there are urgent risks from AI, | 她同意你的观点,即AI有紧迫的风险, |
| 128 | but she also believes the extinction threat is overhyped and a distraction. | 但是她也相信 灭绝的威胁被过度夸大 并分散了注意力 |
| 129 | She told me, quote, | 她告诉我,引用 |
| 130 | I think focusing on the hyperbole and deriving policies through that lens | 我认为,通过这个视角, 专注于夸张和衍生政策 |
| 131 | is not very constructive in our society. | 对我们社会来说,这不太具有建设性。 |
| 132 | Why is she wrong? | 为什么她错了? |
| 133 | Okay, you have two groups of people. | 好吧,你有两组人。 |
| 134 | You have experts who've made a sacrifice in order to be able to warn about the risk. | 你们有专家为警告风险而牺牲了生命 |
| 135 | 于是离开Anthropic或离开OpenAI的人. | |
| 136 | Or left Google, like you? | 还是像你一样离开谷歌? |
| 137 | So that they could warn about the risks. | 这样他们就可以警告风险。 |
| 138 | Now, that wasn't the only reason I left Google. | 这不是我离开谷歌的唯一原因 |
| 139 | I was old and wanted to watch more Netflix. | 我当时老了,想看更多的Netflix。 |
| 140 | You were ready to retire, | 你准备退休了 |
| 141 | but you're not watching very much Netflix these days, are you? | 但你现在没看太多Netflix吧? |
| 142 | No, I'm not, I'm afraid. | 不,我没有,我害怕。 |
| 143 | The future of humanity seems a bit more important. | 人类的未来似乎更重要。 |
| 144 | So on the one hand, you've got people who made a sacrifice. | 所以一方面,你有的人 谁做出了牺牲。 |
| 145 | They're experts. | 他们是专家。 |
| 146 | They see what's going on in the big companies | 他们知道大公司的情况 |
| 147 | in order to warn about the risk. | 为了警告风险 |
| 148 | On the other hand, you have people who say there's zero risk, | 另一方面,你的人 谁说有零风险, |
| 149 | nothing to see here. | 这里没什么好看的 |
| 150 | Don't worry, let us get on with it. | 别担心,让我们继续吧。 |
| 151 | Don't regulate us. | 别管我们 |
| 152 | Who has the vested interest? | 谁拥有既得利益? |
| 153 | The people with the vested interest are the people who have AI companies | 有既得利益的人就是有AI公司的人 |
| 154 | who on the whole are the people who are revising governments | 总的来说,谁是人民 谁正在修订政府 |
| 155 | who say, don't slow us down, just let us get on with it. | 谁说,不要让我们慢下来, 让我们继续吧。 |
| 156 | Who do you trust? | 你相信谁? |
| 157 | Dr. Lee places a lot of faith in the idea that AI can be trained | 李博士很相信 AI可以被训练 |
| 158 | to be human-centered, as she called it. | 就像她所说的 以人为中心 |
| 159 | You have argued that if there were a way to give | 你辩称如果有办法 |
| 160 | something like a maternal instinct to these models, | 类似母亲的本能 这些模型, |
| 161 | a desire for the AIs to protect humanity, | 希望AI保护人类, |
| 162 | that that could be a solution. | 这可能是一个解决方案。 |
| 163 | Do you really believe these models could be trained to care about us? | 你真的相信这些模特可以被训练来照顾我们吗? |
| 164 | Yes, I think they could be. | 是的,我想他们可能是。 |
| 165 | We don't know enough about this yet. | 我们还不知道这些 |
| 166 | There's very little research gone into this. | 对这件事的研究很少。 |
| 167 | We're in a kind of crazy situation where these things might wipe us out. | 我们正处在一种疯狂的境地 这些东西可能会把我们消灭 |
| 168 | Quite a few of the experts believe there's a significant chance of that, | 有不少专家认为 很有可能 |
| 169 | and we're doing almost no research on how to prevent it. | 我们几乎没有研究 如何预防它。 |
| 170 | When you say wipe us out, | 当你说消灭我们的时候 |
| 171 | how would we get from super intelligence to wipe us out? | 我们怎样才能从超级情报中消灭我们? |
| 172 | How do you get from A to Z? | 你怎么从A到Z? |
| 173 | With a super intelligent AI, there's so many different ways it could wipe us out. | 用一个超智能的AI, 有很多不同的方式 它可以消灭我们。 |
| 174 | I don't think it's worth speculating on that. | 我觉得这不值得猜测 |
| 175 | There's obvious ways, like it would create biological viruses | 有明显的方法,比如它会制造生物病毒 |
| 176 | that wouldn't affect it, but would affect us. | 这不会影响它,但会影响我们。 |
| 177 | And these things can already design new viruses. | 这些东西已经可以设计新的病毒了。 |
| 178 | This is a very scary possibility. | 这是一个非常可怕的可能性。 |
| 179 | So there's things like that, but I don't think it's worth speculating. | 所以有这样的事情, 但我不认为它值得猜测。 |
| 180 | The issue is, if they're much smarter than us | 问题是,如果他们比我们聪明得多 |
| 181 | and they want to get rid of us, we're toast. | 他们想除掉我们,我们敬酒 |
| 182 | You've been vocal about this, but I want to give you a chance to explain | 你一直在说这些 但我想给你个解释的机会 |
| 183 | why a kill switch is not a viable solution. | 为什么开关不是可行的办法 |
| 184 | Okay. | 摆 |
| 185 | So for a kill switch, | 所以对于一个杀人开关, |
| 186 | you have to have somebody who's in charge of the kill switch. | 你得找个人负责开关 |
| 187 | Presumably, you don't want to put an AI in charge of it. | 假设你不想让AI负责 |
| 188 | You want a person in charge of it. | 你要一个人负责 |
| 189 | Now, AI is already about as good as a person at persuading you of things, | 现在,AI已经像一个人 说服你的东西, |
| 190 | and it's getting better all the time. | 它越来越好 所有的时间。 |
| 191 | A super intelligent AI will be much better than a person at persuading you of things. | 超聪明的人工智能比说服你更好 |
| 192 | So if you're the person in charge of the kill switch, | 如果你是负责开关的人 |
| 193 | the super intelligent AI will be explaining to you in very convincing ways | 超级聪明的AI会用非常令人信服的方式向你解释 |
| 194 | that you should not pull that switch. | 你不应该拉开开开关。 |
| 195 | So I don't think that that's going to work. | 所以我不认为那会有用。 |
| 196 | You've argued that AI systems may already possess subjective experiences | 您认为AI系统可能已经拥有主观经验 |
| 197 | and that there's no fundamental reason why consciousness | 而意识没有根本原因 |
| 198 | should be exclusive to biological organisms. | 应专用于生物体。 |
| 199 | Yes. | 对 |
| 200 | You don't have to be made of goopy stuff to be conscious. | 你不必被做成 低劣的东西才有意识。 |
| 201 | Okay. | 摆 |
| 202 | So take on Pope Leo, then. | 那就去对付利奥教皇吧 |
| 203 | He has a recent encyclical that takes the opposite position. | 他最近有一个通谕,采取相反的立场。 |
| 204 | He argues that AI systems do not experience anything. | 他争辩说,人工智能系统没有任何经验。 |
| 205 | They do not have feelings, they do not have inner understanding, | 他们没有感情,没有内在的理解, |
| 206 | no embodied experience, no capacity for general moral awareness. | 没有具体的经验,没有普遍道德意识的能力。 |
| 207 | And he insists that however sophisticated the systems become, | 他坚持认为无论系统变得多么复杂 |
| 208 | they will remain fundamentally different from human beings. | 他们将继续与人类根本不同。 |
| 209 | What do you think the pope fundamentally misunderstands about consciousness? | 你认为教皇从根本上 误解意识是什么? |
| 210 | Um, I think he thinks we were made by God. | 我觉得他觉得我们是上帝造的 |
| 211 | He is, after all, the pope. | 毕竟他是教皇 |
| 212 | And if he didn't think that, it'd have problems. | 如果他不这么认为,那会有问题。 |
| 213 | I don't believe we were made by God. | 我不相信我们是上帝造的 |
| 214 | I believe we were made by evolution. | 我相信我们是由进化而成的 |
| 215 | So of course the pope doesn't think these things could have consciousness | 教皇当然不会认为这些东西有意识 |
| 216 | or souls or whatever. | 或灵魂或什么的。 |
| 217 | I should say I have a lot of admiration for this pope. | 可以说我对这位教皇非常钦佩 |
| 218 | I think it's very important the work he's doing on how AI should be controlled. | 我认为他如何控制AI的工作非常重要. |
| 219 | But I have a fundamental disagreement with him about the nature of beings. | 但我与他对于生命的本质有根本的分歧. |
| 220 | I think with the product of evolution, he thinks we were made by God. | 我想用进化的产物 他以为我们是上帝造的 |
| 221 | It sounds like you're arguing that there is not something about | 听起来你好像在争论 没有什么 |
| 222 | being a living embodied human being that computation, however advanced, can't replace. | 人类是生命的象征 计算无论怎样先进 都无法取代 |
| 223 | That is what I'm arguing. | 这就是我的论点。 |
| 224 | I think consciousness is a property of a sophisticated kind of computation. | 我认为意识是一种复杂的计算方式 |
| 225 | And in fact, if you look at the scientists studying these AIs, | 事实上,如果你看看研究这些AI的科学家们 |
| 226 | when they're not thinking philosophically, when they're not thinking about the issue | 当他们不思考哲学, 当他们不思考问题 |
| 227 | of are they conscious, but just talking about what's going on, | 他们有意识,但只是谈论 发生了什么事, |
| 228 | they say things in their papers like the AI was aware that it was being tested. | 他们在报纸上说的东西 像AI知道 它正在测试。 |
| 229 | Now, normally if you say that something's aware, that's attributing consciousness to it. | 现在,通常,如果你说某件事是意识到的, 那就是将意识归于它。 |
| 230 | So in everyday life, these AI scientists are attributing consciousness to these AIs. | 因此在日常生活中,这些AI科学家将意识归于这些AI. |
| 231 | It's just when you start talking philosophy, | 只是当你开始谈论哲学, |
| 232 | they get worried because they don't know what consciousness is. | 他们担心是因为不知道什么是意识 |
| 233 | If you believe that it's possible to perhaps instill within an AI a maternal instinct, | 如果你认为有可能在人工智能中注入母性的本能 |
| 234 | could AIs not also be trained on the values of the world's great religions? | 难道不能对AI进行世界各大宗教价值观的培训? |
| 235 | Yes, they certainly could. | 是的,他们当然可以。 |
| 236 | So the way they're trained at present is they're trying to predict the next word in a document. | 因此,他们目前接受训练的方式是试图在文档中预测下一个词. |
| 237 | And the documents they're given come from all sorts of different sources. | 而他们得到的文件来自各种不同的来源. |
| 238 | They just scrape everything they can get their hands on for free on the web. | 他们只是刮了一切 他们可以得到的手 免费在网上。 |
| 239 | Now, if you were teaching your kid to read, | 如果你教孩子读书 |
| 240 | you wouldn't want to do that. | 你不会想那么做的 |
| 241 | I think the way they should be trained is they should only be trained | 我认为他们应该接受训练 只能接受训练 |
| 242 | on things that exhibit good behavior. | 对表现良好的东西。 |
| 243 | The kind of behavior we'd like these beings that we're going to share the planet with to exhibit. | 我们想让这些生物 共同展示地球的行为 |
| 244 | Once they've been trained on good behavior | 一旦他们接受了良好行为训练 |
| 245 | and they can recognize what good behavior is and what bad behavior is, | 他们可以识别什么是好行为 什么是坏行为, |
| 246 | then it's fine to show them examples of bad behavior | 让他们看看不良行为的例子就好 |
| 247 | and tell them it's bad behavior. | 告诉他们这是坏行为 |
| 248 | They will figure out for themselves it's bad behavior. | 他们会发现这是坏行为 |
| 249 | You're suggesting we should be training them the way we ideally would train our children? | 你是说我们应该以我们最理想的方式训练他们? |
| 250 | Exactly. | 没错 |
| 251 | So these AIs are not lines of code that we write. | 所以这些AI不是我们写的代码的线条. |
| 252 | They're not software like normal computer software. | 它们不像普通的计算机软件那样是软件. |
| 253 | They're things that learn from data. | 它们是学习数据的东西。 |
| 254 | And we would like them to learn to be good AIs in our definition of good. | 我们希望他们能学会成为好AI 在我们对好的定义。 |
| 255 | The control we have over them is the data we train them on | 我们控制他们,就是我们训练他们的数据 |
| 256 | and the rewards we give them, much like it is for children. | 和我们给他们的奖励, 就像它对于儿童一样。 |
| 257 | If we accept that AIs have the ability to have consciousness, | 如果我们承认AI有意识的能力 |
| 258 | what does that mean for our moral obligation towards an AI with consciousness? | 这对我们对有意识的AI的道德义务意味着什么? |
| 259 | We're people, and what we care about is other people. | 我们是人,我们关心的是其他人。 |
| 260 | I don't see any reason why we shouldn't develop advanced intelligences | 我看不出为什么我们不发展高级情报 |
| 261 | that are smarter than us and that are conscious but have less moral rights than us. | 他们比我们聪明,他们有意识,但道德权利却比我们少。 |
| 262 | I would be perfectly happy with developing these things | 我很乐意开发这些东西 |
| 263 | so that they care about us and their view of the world is that we're what's really important | 所以他们关心我们和 他们的世界观 是,我们才是真正重要的 |
| 264 | and they're not so important. | 他们并不重要 |
| 265 | So I don't see why we should give them moral rights. | 所以我不明白我们为什么要给他们道德权利。 |
| 266 | One could imagine that we are engineering a form of digital slavery. | 人们可以想象,我们正在设计一种数字奴役形式。 |
| 267 | You could imagine that, yes. | 你能想象的到,是的。 |
| 268 | So we have to make a decision about this. | 因此,我们必须对此作出决定。 |
| 269 | And I think the decision we have to make if we produce superintelligence is, | 我认为如果我们生产超级智能,我们必须做的决定是, |
| 270 | do we want to just be a passing phase in the evolution of intelligence | 我们是否只想成为 智力进化过程中的传承阶段 |
| 271 | that will eventually disappear, maybe quite soon? | 这最终会消失, 也许很快? |
| 272 | Or do we want to try and stay in control, in which case these superintelligences | 或者我们想保持控制, 在这种情况下,这些超级情报 |
| 273 | that are smarter than us have to be made so what they really care about is us? | 所以他们真正关心的是我们? |
| 274 | I'd like to turn to the news. | 转正报. |
| 275 | AI巨头NVIDIA亿万富翁CEOJensen Wong辞职. | |
| 276 | Does leaving technology in the hands of private citizens | 技术是否由公民个人掌握 |
| 277 | that concerns the national security of nations, | 事关各国的国家安全, |
| 278 | is that a risk we should be taking? | 我们应该冒这个险吗? |
| 279 | Yes, it's obviously a risk. | 是的,这显然是一种风险。 |
| 280 | I need to say a few things about Jensen Wong. | 我要谈谈黄詹森的事 |
| 281 | Please do. | 请吧 |
| 282 | I actually like him. | 其实我很喜欢他 |
| 283 | I've interacted with him a bit. | 我和他有过互动 |
| 284 | I think he's a decent guy. | 我觉得他是个好人 |
| 285 | I think he made a mistake. | 我认为他犯了一个错误。 |
| 286 | I think it was slightly stupid to think he wouldn't be caught | 以为他不会被抓到,这有点傻 |
| 287 | because obviously the security services would be looking at somebody like him. | 因为很明显安全部门会 看着他这样的人 |
| 288 | I don't put him in the same category as someone like Musk or Zuckerberg, | 我不会把他和穆斯克或扎克伯格一样 归入同一类别 |
| 289 | who I think are very self-interested and willing to do more or less anything to make more money. | 我认为他很自私 愿意做多少事情来赚钱 |
| 290 | But obviously having the future of humanity controlled by just a few men is a bad idea. | 但显然只有几个人控制人类的未来是个坏主意。 |
| 291 | The reason I ask you about Jensen Wong is there is also this question of a race between | 我问你关于Jensen Wong的原因 也是关于比赛的问题 |
| 292 | the Chinese and the Western frontier labs and whether humanity will be in better shape | 中国和西方的前沿实验室 人类是否将处于更好的状态 |
| 293 | or more secure if a representative democracy reaches super intelligence or reaches, | 如果一个代议制民主国家达到超级智能或达到更安全的程度, |
| 294 | continues to lead in the AI race. | AI赛事继续领先. |
| 295 | It's not unlike during the nuclear arms race. | 这与核军备竞赛期间的情况并无不同. |
| 296 | It seems as though there is a similar dilemma we face with AI in this national security component | 似乎在国家安全方面 AI也有类似的两难处境 |
| 297 | of who ends up leading with this new advanced technology. | 最终领先于这种新的先进技术 |
| 298 | Okay, so when it comes to super intelligence, | 好吧,所以当涉及到超级智能, |
| 299 | I think actually the interests of China and the U.S. are aligned and they will collaborate. | 我认为实际上中国和美国的利益是一致的,它们将会合作。 |
| 300 | It does not want super intelligence taking over. | 它不希望超级情报接管。 |
| 301 | Last summer I went to Shanghai and talked to a member of the Politburo. | 去年夏天我去了上海 和一位政治人物谈过 |
| 302 | I believe the Politburo understands this stuff much better than the U.S. government. | 我相信政治局比美国政府更了解这些 |
| 303 | They are very concerned about these risks and they will collaborate with the U.S. on trying | 他们非常关心这些风险 他们会与美国合作尝试 |
| 304 | to prevent super intelligence taking over. | 防止超级情报接管 |
| 305 | So I think what should be happening at present, it won't happen while we still have a, | 所以我认为现在应该发生什么 我们还有... |
| 306 | how can I say this, maybe a corrupt senile felon in the White House. | 我怎么能这么说呢 也许白宫里有个腐败的老年重罪犯 |
| 307 | So it's not going to happen for two years, but we will eventually get a sensible U.S. | 所以两年内不会发生, 但我们最终会有一个明智的美国。 |
| 308 | that I think will collaborate with the Chinese on trying to avert the risk of super intelligence | 我想与中国人合作 避免超级智能的风险 |
| 309 | taking over, much as the United States collaborated with the Soviet Union in the 1950s on trying to | 1950年代美国与苏联合作试图 |
| 310 | avert the dangers created by H-bombs. | 避免H炸弹造成的危险 |
| 311 | One of the AI executives who you seem to trust the most is Dario Amode of Anthropic. | 你最信任的AI高管之一 是雅典的达里奥·阿莫德 |
| 312 | You gave him an A for his recent essay on why we should be pacing the AI frontier. | 你给了他一个A,因为他最近的文章 关于为什么我们应该 速度AI边境。 |
| 313 | Anthropic took steps this week towards one of the biggest | 本周,Anthropic采取了步骤 走向一个最大的 |
| 314 | IPOs in history with a valuation of $1.4 trillion. | 历史上的IPO估值为1.4万亿美元. |
| 315 | Can even the most safety conscious AI companies put humanity's interest ahead of its shareholders | 即使最安全的AI公司 也能把人类的利益放在股东的首位 |
| 316 | when trillions of dollars are at stake? | 当万亿元处于危险之中时? |
| 317 | So Dario is in a very difficult position. | 因此达里奥的处境非常困难. |
| 318 | One thing I should say is don't look at what people say, look at what people do. | 我该说一件事,就是不要看别人说什么,看别人做什么. |
| 319 | And so Claude was used by the U.S. military. | 于是克劳德被美国军方使用. |
| 320 | The U.S. military wanted to be allowed to use it to spy on U.S. citizens. | 美国军方希望被允许利用来监视美国公民. |
| 321 | And Dario said no and took a big hit for that. | 达里奥拒绝了,因此大受欢迎 |
| 322 | When Dario said no, Sam Altman said, we agree with him. | 当达里奥拒绝时 Sam Altman说 我们同意他的意见 |
| 323 | And then about a day after the U.S. military had fired Claude, | 在美军解雇克劳德的一天后 |
| 324 | Sam Altman说 好吧 打开AI 我来 | |
| 325 | So Dario was willing to take a hit in order to do what he thought was right. | 于是达里奥为了做他认为对的事,愿意接受一击. |
| 326 | That's much more convincing than anything they say. | 这比他们说的更有说服力 |
| 327 | How then, you know, if a company is going to market with an IPO valuation of $1.4 trillion | 如果一个公司要上市 IPO估值为1.4万亿美元 |
| 328 | and it says in its filing papers there is a 10% chance that civilization could be | 它说,在它的档案文件中 有10%的可能性 文明可能是 |
| 329 | at risk because of the technology that is going to market, help me understand why and how | 帮助我理解为什么和如何 |
| 330 | the marketplace is willing to accept this kind of risk. | 市场愿意接受这种风险。 |
| 331 | I mean, it would have been unheard of in the nuclear age when the Manhattan | 曼哈顿的核子时代 本来是没听说过的 |
| 332 | Project was being developed that something as dire as nuclear weapons | 正在开发一个像核武器一样可怕的项目 |
| 333 | would have been treated as as incautiously as the market seems to be treating this real risk. | 市场似乎在处理这种实际风险时会视之为不合理。 |
| 334 | From Dario's point of view, he has to make a decision about whether to say | 从达里奥的观点来看,他必须做出是否说 |
| 335 | this stuff shouldn't be developed, I'm out. | 这些东西不应该开发,我退出。 |
| 336 | In which case it's going to be developed by people who are less cautious than him. | 这样的话,就会由比他更谨慎的人来发展. |
| 337 | That puts him in a very difficult situation, right? | 这让他陷入非常困难的境地,对吧? |
| 338 | He has to try and stay up with the frontier so that if it's developed by anybody, | 他必须努力守住边境 如果它被任何人开发出来 |
| 339 | it's developed by Anthropic who are going to be more cautious than the other companies. | 它由Anthropic公司开发, 他们会比其他公司更谨慎。 |
| 340 | That was the original motivation of open AI. | 这是公开AI的最初动机. |
| 341 | They said if anybody's going to develop super intelligence, | 他们说如果有人要发展超级智能 |
| 342 | it should be developed by somebody who really cares about safety. | 它应该由真正关心安全的人来开发。 |
| 343 | And open AI was set up to really care about safety. | 开放AI是为了真正关心安全 |
| 344 | Since then, it's started to care much less about safety. | 从那时起,它开始更不关心安全. |
| 345 | So I think from Dario's point of view, the reason he's in a bind is | 所以我认为从达里奥的观点来看 他陷入困境的原因是 |
| 346 | he's very worried about the safety, | 他非常担心安全 |
| 347 | but he thinks it's better if Anthropic develops it than other companies. | 但他认为如果Anthropic开发它比其他公司更好. |
| 348 | And I know quite a few people who work for Anthropic precisely because | 我认识很多为Anthropic工作的人 |
| 349 | they think it's going to be the safest company for developing AI. | 他们认为这将是开发AI最安全的公司. |
| 350 | This summer, your fellow AI pioneer and Nobel laureate, Demis Hassibis, | 今年夏天 你们的AI先驱和诺贝尔奖得主 戴米斯·哈西比斯 |
| 351 | outlined a proposal for federally overseen self-regulatory organization | 概述关于联邦监督自律组织的建议 |
| 352 | with a board that includes a set of independent technical experts. | 设有一个包括一组独立技术专家的委员会。 |
| 353 | The Wall Street Journal reported that some in the White House were receptive to the idea, | "华尔街日报"报道,白宫的一些人接受这个想法, |
| 354 | but AI executives like Mark Zuckerberg, Elon Musk, and even Jensen Huang | 包括马克·扎克伯格、伊隆·穆斯克 甚至黄詹森 |
| 355 | encouraged President Trump to reject the plan. | 鼓励特朗普总统拒绝该计划. |
| 356 | Is that a workable solution still to your way of thinking? | 你的思想方式还有这个可行办法吗? |
| 357 | I think it's a first step. | 我认为这是第一步。 |
| 358 | It's not the whole solution, | 这不是整个解决方案, |
| 359 | but I think it's crazy that at present there's almost no regulation on AI. | 但我认为这太疯狂了 目前几乎没有关于AI的规定。 |
| 360 | The regulation we have on pharmaceutical drugs, for example, | 比如,我们有关药品的条例, |
| 361 | is that if you want to release a new pharmaceutical drug, | 如果你想释放一种新药 |
| 362 | you have to convince the FDA that it's safe, | 你必须让FDA相信这是安全的, |
| 363 | and you have to do the work to convince the FDA, | 你必须努力说服FDA |
| 364 | and the FDA can consult external experts. | 林业发展局可以咨询外部专家。 |
| 365 | We need something like that for AI. | 我们需要这样的东西对AI。 |
| 366 | That's the very minimum we need, | 这是我们最需要的 |
| 367 | and it's crazy we don't have it. | 我们没有它,这是疯了。 |
| 368 | This program is the revival of William F. Buckley Jr.'s Firing Line, | 这个计划是恢复小威廉·F·巴克利的射击线 |
| 369 | which aired for 33 years on PBS. | 在PBS播出了33年 |
| 370 | In 1993, William F. Buckley Jr. hosted a discussion | 1993年,小威廉·F·巴克利主持讨论. |
| 371 | about the proliferation of nuclear weapons and the threat of nuclear destruction. | 核武器的扩散和核毁灭的威胁。 |
| 372 | Here is a physicist, Russell Seitz, | 这是物理学家罗素·塞兹 |
| 373 | talking about the challenge of containing nuclear technology. | 谈论遏制核技术的挑战。 |
| 374 | As high technology ramifies appropriate technologies | 由于高技术是适当的技术 |
| 375 | that can be subverted to the ends of proliferation multiply, | 被破坏到扩散的目的, |
| 376 | until at the margin, we may be forced to rely more on our powers of diplomacy | 在边上,我们可能不得不更多地依赖我们的外交力量 |
| 377 | than on our capacity to suppress the progress of technology globally, | 而不是我们抑制全球技术进步的能力, |
| 378 | if we are to keep under control the prospect of essentially unlimited proliferation of nuclear weapons. | 如果我们要控制核武器基本上无限扩散的前景。 |
| 379 | Okay, so there are some big differences between AI and nuclear weapons. | 好吧,所以在AI和核武器之间有一些很大的区别。 |
| 380 | The first big difference is AI can do tremendous good. | 第一个大区别是AI可以做巨大的好事. |
| 381 | We haven't talked about that, | 我们还没谈这个 |
| 382 | but it'll do tremendous good in healthcare and education, | 但它会在保健和教育方面产生巨大效果, |
| 383 | and in making almost every industry more productive. | 让几乎所有的行业都更加富有生产力。 |
| 384 | So it's going to be developed for that reason. | 因此,它会为此发展起来. |
| 385 | We can't stop it, I don't think. | 我们不能阻止它,我不认为。 |
| 386 | But the second thing is nuclear weapons aren't intelligent. | 但第二件事是核武器不聪明。 |
| 387 | AI is intelligent, and if we can educate it and train it | AI很聪明,如果我们能教育它,训练它 |
| 388 | so that it really cares about us, then proliferation is not nearly so worrying. | 这样它才真正关心我们,那么扩散就不那么令人担忧。 |
| 389 | You argue the technology that you have spent your life developing | 你争辩你一生的科技发展 |
| 390 | might have been a mistake. | 也许是个错误 |
| 391 | And you address this head on in your forthcoming book, | 在你即将出版的书中 写着这个头 |
| 392 | writing, quote, responsibility is not the same thing as regret. | 写作、引用、责任与后悔不同。 |
| 393 | Explain the difference. | 解释差别. |
| 394 | Okay, I want to distinguish actually two kinds of regret. | 好吧,我想区分两种悔恨 |
| 395 | Sort of sometimes you're mean to a friend of yours, | 有时候你对你的朋友很刻薄 |
| 396 | and you regret it. | 你却后悔了 |
| 397 | That's what I call guilty regret. | 这就是我所谓的有罪遗憾。 |
| 398 | When you did something and at the time you knew it was wrong, | 当你做了一些事情 当你知道这是错的, |
| 399 | but you did it anyway. | 但你还是做到了 |
| 400 | Then there's things that you did where at the time | 然后,有些事情 你做了在当时 |
| 401 | you couldn't have foreseen the consequences, | 你无法预见后果 |
| 402 | but they turned out to have potentially very bad consequences. | 但结果却产生了非常糟糕的后果 |
| 403 | I feel that that's what I did. | 我觉得这就是我做了什么。 |
| 404 | And if I was put in the same situation again | 如果我再次陷入同样的境地 |
| 405 | with the knowledge I had at the time, I would have behaved the same way. | 以我当时的知识,我也会采取同样的做法。 |
| 406 | Dr. Hinton, thank you so much for joining me on Firing Line. | 欣顿博士,非常感谢你 加入我Firing线。 |
| 407 | Thank you for inviting me. | 谢谢你邀请我 |
| 408 | Firing Line with Margaret Hoover is made possible in part by | 与玛格丽特·胡佛的交火线部分通过 |
| 409 | Robert Granieri, Vanessa and Henry Cornell, | 罗伯特·格拉尼埃里 瓦内萨和亨利·康奈尔 |
| 410 | the Margaret and Daniel Loeb Foundation, | 玛格丽特和丹尼尔·洛布基金会, |
| 411 | the Beth and Ravenel Curry Foundation, | 贝丝和雷文埃尔·库里基金会, |
| 412 | Cliff and Laurel Astes, | 克里夫和劳蕾尔·阿斯特斯 |
| 413 | and by the following. | 以及以下内容。 |
| 414 | 你看着PBS。 |
该视频共有字幕 414 条。解锁更多字幕为会员功能,请移动到 价格
