Elon Musk on India

 Hangzou ali baba deepseek https://www.youtube.com/watch?v=YeuuDPBpLzI

Transcript 0:00 让我们以热烈的掌声 0:01 欢迎黄仁勋先生和王坚先生登台 0:05 现在让我们以热烈的掌声欢迎Jason 0:08 韩和王先生上台 0:17 王坚先生是中国云计算领军人物 0:20 阿里云创始人 0:21 现任之江实验室主任 0:23 王先生 0:24 季军一直是中国资本形成背后的推动力 0:27 尤其是在云计算中 0:30 王先生和王先生在炉边聊天 0:34 下来是黄先生和王先生的炉边谈话 0:39 两个最有前沿关注的视野的人 0:44 大人物将进行谈话 0:49 good morning good morning everyone 早上好大家早上好 0:51 大家好 ha 0:53 uh so this Jason 呃所以这个Jason 0:54 and welcome to supply chain expo 欢迎来到供应链博览会 0:57 and good to see you again after a long time okay 很高兴在很长一段时间后再次见到你，好吗？ 1:00 so i have written down my question on my phone 所以我在手机上写下了我的问题。 1:02 the the the first time we met in Beijing was what year 我们第一次在北京见面是哪一年 1:08 a long time ago 很久以前 1:09 yeah it's around two thousand twelve 是的，大约在两千十二岁左右。 1:13 uh thirteen at that time 呃那时候十三 1:15 uh yes about almost ten years ago i was in and 呃，是的，大约十年前，我在 1:18 and and also 而且而且 1:19 i'm really glad you know when i visit you in the silicon valley 我很高兴我去硅谷看你的时候你知道 1:23 you are really the person okay 你真的是没事的人 1:25 you are really the person 你真的是那个人 1:27 to talk about your company's technology so 谈论贵公司的技术 1:31 thank you for your time when i was in silicon 谢谢你在我硅的时候抽出时间 1:33 you showed me your company 你给我看了你的公司 1:35 and you know at that time 你知道在那个时候 1:38 i really found the founder of a company is very important 我真的发现一个公司的创始人很重要 1:42 and and you can see your passion about the the work 你可以看到你对工作的热情 1:46 you're working on that yeah 你正在努力，是的 1:47 and at the time when we first started talking 在我们刚开始谈话的时候 1:51 we were talking about computer graphics 我们在谈论计算机图形学 1:53 yes 是的 1:53 and mobile devices you're right 和移动设备，你是对的 1:56 right and so that was twenty twelve 对，那是二十二岁。 1:59 and you probably saw in the video just now yeah 你可能在刚才的视频中看到了，是的 2:02 i came to China to talk about kuda for the first time 我第一次来中国谈库达 2:06 in two thousand and seven 在两千零七年 2:08 you are yes 你是是 2:09 好久以前啊 好久以前啊 2:10 that's a long time ago really incredible and so 那是很久以前的事了，真的很不可思议，所以 2:13 it's great we've known each other a very long time 我们认识这么久真是太好了 2:16 and so thank you good to see you again 所以谢谢你，很高兴再次见到你。 2:18 nice to see you again and a story van can they hear you are 很高兴再次见到你和一辆故事车，他们能听到你吗？ 2:21 you sure 你确定 2:22 yeah remember the first time 是啊记得第一次 2:24 they might not listen your time is actually 他们可能不会听你的时间实际上是 2:27 in the in the Los Angeles 在洛杉矶 2:29 that was in the siggraph 那是在签名中 2:30 okay that was really really long time ago 好吧，那是很久以前的事了。 2:33 and so you invented you invent the gpu 所以你发明了你发明了gpu 2:39 and and change the landscape of the graphics area 并改变图形区域的景观 2:44 now we have ai so it is incredible journey yeah 现在我们有了人工智能，所以这是一段不可思议的旅程，是的 2:48 so my first question for you is really about the the technology 所以我的第一个问题是关于技术 2:53 and you know ai is a buzzword 你知道ai是一个流行语 2:57 and people have a different perspective on ai and ai computing 人们对人工智能和人工智能计算有不同的看法 3:02 so how things actually have advanced and be changed 那么事情实际上是如何发展和改变的呢？ 3:06 fundamentally in the past few years okay 基本上在过去的几年里 3:10 yeah 是呀 3:12 great question um 很好的问题嗯 3:15 first of all uh ai is a new way of doing software 首先uh ai是一种做软件的新方式 3:23 yeah 是呀 3:24 and first on first principles yeah 首先是第一原则，是的 3:26 instead of human coding describing algorithms 而不是人类编码描述算法 3:32 yeah to predict an outcome yeah 是的，预测结果，是的 3:35 we use a algorithm 我们用一种算法 3:39 to learn how to predict the outcome yes 学习如何预测结果是的 3:42 from example information 从示例信息 3:45 example data 示例数据 3:47 and and so this method of using computers to i to uh learn 所以这种使用计算机的方法让我呃学习 3:56 how to predict yeah 如何预测是的 3:57 uh has proven to be extremely scalable uh已被证明具有极强的可扩展性 4:00 yeah and as you know 是的，如你所知 4:02 we've been working on machine learning for a long time but yeah 我们研究机器学习很久了 4:05 twenty twelve was the Big Bang yes 2012年是大爆炸，是的 4:08 with Alex net yeah 和Alex net是的 4:09 and that was the time 就是那个时候 4:10 when the demonstration of deep learning 当深度学习的演示 4:13 has proven to be incredibly effective 已经被证明是非常有效的 4:16 so much better than computer scientists 比计算机科学家好多了 4:18 could do with computer vision yeah 可以用计算机视觉，是的。 4:21 and so it started from twenty twelve to the next five years 就这样从2012年开始到接下来的五年 4:25 or so you and i saw first computer vision becoming effective 或者你和我看到了第一个计算机视觉变得有效 4:30 and then superhuman yes 然后超人是的 4:32 and then speech recognition becoming effective 然后语音识别变得有效 4:36 and then superhuman yes 然后超人是的 4:37 and then shortly after that language understanding 然后在语言理解后不久 4:40 becoming effective 生效 4:41 and then superhuman yes right 然后是超人，是的，对。 4:43 so each and every one of these 所以这些中的每一个 4:46 uh different modalities 呃不同的模式 4:48 uh represented the first wave called perception ai uh代表了第一波叫做感知ai 4:52 then the second wave was generative ai yes 然后第二波是生成的ai是的 4:55 we can now translate from one modality to another yeah 我们现在可以从一种情态转换到另一种情态 4:59 from English to Chinese from English to pictures 从英文到中文从英文到图片 5:04 from pictures to English from Chinese to video yeah 从图片到英语从中文到视频是的 5:08 yeah 是呀 5:09 generative ai translation 生成式人工智能翻译 5:11 yeah 是呀 5:11 you know the ultimate translation and and so the generative ai 你知道终极翻译，所以生成人工智能 5:17 uh lasted uh really started about seven years ago yeah 持续了大概7年前开始的 5:22 and is going very strong right now 而且现在非常强劲 5:24 so now 所以现在 5:24 ai can understand information and generate information yeah 人工智能可以理解信息并生成信息，是的 5:29 the wave that we're in right now is incredible 我们现在所处的浪潮是不可思议的 5:32 yeah 是呀 5:32 and it's called reasoning yep 这叫做推理，是的。 5:34 and the reason why reasoning is so effective 推理如此有效的原因 5:37 so powerful is that the ai can understand and solve problems 如此强大，人工智能可以理解和解决问题 5:44 that it has never seen before yeah 它以前从未见过的东西，是的 5:46 just like humans yeah 就像人类一样 5:48 we can break down a problem step by step by step 我们可以一步一步地分解一个问题 5:51 and and solve problems that we have never really solved before 并解决我们从未真正解决过的问题 5:56 and so 所以 5:57 that's reasoning ai the next wave is called physical ai yeah 这就是推理ai下一波被称为物理ai yeah 6:02 when all of this capability can now go into a physical machinery 当所有这些能力现在都可以进入物理机器时 6:07 such as a robot and so this next 比如机器人，接下来就是这个。 6:10 this last twelve years 这过去的十二年 6:12 or so uh ai has moved very quickly 或者呃ai进展很快 6:15 it seems like every three or four five years three 似乎每隔三、四年、五年、三年 6:17 or four five years 或者四五年 6:18 you know we saw a big breakthrough it 你知道我们看到了一个重大突破 6:19 and i would say that we're now we're now near a time uh 我想说我们现在我们现在接近一个时间呃 6:24 when when uh ai should be able to solve uh most 什么时候uh ai应该能够解决uh大多数 6:30 cognitive tasks okay 认知任务好吧 6:32 and achieve most tests better than most humans so 并比大多数人更好地完成大多数测试 6:37 that's the level we call artificial general intelligence 这就是我们所说的通用人工智能的水平 6:41 which is the reason why now 这就是为什么现在 6:42 everybody's talking about artificial super intelligence 每个人都在谈论人工超级智能 6:45 just like in the beginning yeah 就像一开始一样，是的 6:46 we were able to achieve effectiveness 我们能够实现有效性 6:49 and then superhuman achieve effectiveness superhuman now 然后超人达到超人的有效性 6:52 for for a problem solving 为了解决问题 6:54 we should be able to achieve a superhuman fairly soon 我们应该很快就能造出超人了 6:58 it is incredible yeah 太不可思议了，是的。 6:59 incredible last last decade 难以置信的最后一个十年 7:01 and and you know that you know 你知道你知道 7:04 particularly this year the open source model is changing 尤其是今年开源模式正在发生变化 7:08 the landscape of ai 人工智能的风景 7:10 technology and the business today okay 今天的技术和业务都很好 7:12 but but Jen yeah 但是但是珍是的 7:14 which one which one 哪一个哪一个 7:15 which one of the technologies advances were you most excited by 哪项技术进步让你最兴奋？ 7:20 oh actually 哦其实 7:22 i think the the one of the things really exciting about 我认为真正令人兴奋的是 7:24 for me is actually 对我来说实际上是 7:27 computing is really the fundamental thing for everything okay 计算真的是一切的基础 7:31 so when talking your ai 所以在谈论你的ai时 7:33 is really the computing behind that 是背后的计算 7:35 and the computing is actually is changing everything you know 计算机实际上正在改变你所知道的一切。 7:38 the the ai is the something that you see 人工智能是你看到的东西 7:40 so goes back like twenty years ago 这要追溯到20年前 7:43 and we talking about the computer 我们谈论的是电脑 7:45 but very few people are talking about the computing self OK 但是很少有人在谈论计算机本身。 7:49 so they actually you know 所以他们实际上你知道 7:50 rather we says computers change the world 相反，我们说计算机改变世界 7:53 but actually the computing behind the computer that's right 但实际上计算机背后的计算是对的 7:56 actually 实际上 7:56 it's changed the world 它改变了世界 7:57 and the ai bring the computing to the next stage okay 人工智能将计算带入下一阶段，好吗？ 8:01 so it's it is incredible in general technology 所以这在一般技术中是不可思议的 8:03 and and even the way 甚至是方式 8:04 we're training the models is changing so fast right 我们在训练模型变化太快了，对吧？ 8:06 yeah the first the first decade was largely 是的，第一个十年主要是 8:11 uh occupied by pre training yes 呃，被预培训占用了，是的。 8:14 so we collected a lot of data 所以我们收集了很多数据 8:15 maybe we even use ai to prepare the data yep 也许我们甚至可以使用人工智能来准备数据，是的 8:19 and we use pre training 我们使用预训练 8:20 and then we used a a human reinforcement learning 然后我们使用了人类强化学习 8:24 right reinforcement learning human feedback 强化学习人类反馈 8:26 which is a kind of like a uh human 这有点像呃人类 8:30 coaching the ai yeah 指导ai yeah 8:32 we're we're human 我们是人类 8:34 uh aligning the ai um hmm uh对齐ai um hmm 8:36 and then now we're in this post training era 现在我们处在后训练时代 8:38 where the ai is thinking by itself and uh 人工智能自己思考的地方，呃 8:42 uh practicing uh huh 呃在练习嗯哼 8:44 uh 额 8:44 doing 做 8:45 uh reinforcement learning verifiable feedback and yeah right 强化学习反馈是的 8:49 so many synthetic data generation 如此多的合成数据生成 8:51 and uh taking test by itself and learning how to reason 自己考试，学习如何推理 8:55 so it's incredible 所以太不可思议了 8:57 the amount of computation is necessary now yeah 计算量现在是必要的，是的 8:59 you'll know actually 实际上你会知道的 9:00 i have a psychology background and so for me 我有心理学背景，所以对我来说 9:03 the the ai is not really an simulation of the human intelligence 人工智能并不是真正的人类智能模拟 9:08 it's really an augmentation of the human intelligence 它实际上是人类智力的增强。 9:12 and even more 甚至更多 9:13 so for me it's more like ai is to extend human creativity 所以对我来说，人工智能更像是扩展人类的创造力 9:17 and instead of just replacing human intelligence okay 与其只是取代人类智能，好吗？ 9:20 well our car extended our human mobility 汽车扩展了人类的行动能力 9:24 you're very right 你说得很对 9:25 airplane extended our human mobility yeah 飞机扩展了人类的流动性 9:28 and now we have ai is going to extend our human intelligence 现在我们有人工智能将扩展我们的人类智能 9:32 you're very right 你说得很对 9:32 it doesn't work the same way as our brain 它的工作方式和我们的大脑不同。 9:35 you're very right right 你说得很对，对 9:36 it doesn't work the same as our brain 它和我们的大脑不一样。 9:37 but it can perform tasks similar to what we can do yes 但它可以执行类似于我们所能做的任务，是的 9:42 okay so go back to this open source yeah 好吧，所以回到这个开源，是的 9:45 you know it's also 你知道这也是 9:46 incredible you know moment for that 难以置信，你知道那一刻 9:49 and and you know that we have deep seek 你知道我们有很深的追求 9:52 and you have we have the queen from the uh alibaba cloud 你有我们有来自阿里巴巴云的女王 9:57 and there are only a few among them okay 其中只有几个好吧 10:00 and but but my lately moon 但是我最近的月亮 10:04 moon shot Kimi oh 月亮射击Kimi oh 10:05 yes 是的 10:05 you're right Kimi is pretty good 你说得对Kimi挺不错的 10:07 so actually my question for you is really you know 所以实际上我想问你的问题是 10:10 is this the win of the disruptive 这是颠覆性的胜利吗？ 10:13 i mean the the open source model is this the win of disruptive 我的意思是开源模型是颠覆性的胜利吗？ 10:19 and driving forces for the future ai development 和未来人工智能发展的驱动力 10:23 you know 你知道的 10:24 we were just talking about how AI has progressed very quickly 我们只是在谈论人工智能是如何快速发展的 10:27 yeah and the reason for that is of course 是的，原因当然是 10:29 people say uh NVIDIA's technology is advancing very quickly 人们说呃NVIDIA的技术进步很快 10:34 and it's true yeah 这是真的 10:35 we improve the performance of computing ai 我们提高计算人工智能的性能 10:39 computing by a hundred thousand times in the last ten years yeah 在过去的十年里计算了十万次 10:44 so we can process more data learn more quickly 所以我们可以处理更多数据，更快地学习 10:49 now of course 现在当然 10:51 what is not talked about 什么不被谈论 10:53 and it should be 它应该是 10:55 is that the vast majority of ai research was done in the open 绝大多数人工智能研究都是公开进行的 11:00 the amount of archive papers yeah 档案文件的数量是的 11:03 from all over the world is incredible and in fact 来自世界各地令人难以置信，事实上 11:07 i think i saw a statistic that 我想我看到了一个统计数据 11:12 the archive papers published by research papers published by uh 呃发表的研究论文发表的档案论文 11:17 Chinese researchers is now the highest in the world okay 中国的研究人员现在是世界上最高的 11:21 and so so the thing that is happening is that in a lot of ways 所以正在发生的事情是，在很多方面 11:26 researchers are uh collaborating in open science 研究人员在开放科学领域进行合作 11:31 when they publish their science 当他们发表他们的科学 11:33 then you can read and you can contribute 然后你可以阅读并做出贡献 11:35 then i can read and contribute yeah 然后我可以阅读和贡献是的 11:36 so we are in fact 所以我们实际上是 11:38 collaborating in open science yeah 在开放科学领域合作 11:41 and the next version of that is open source yeah 下一个版本是开源的，是的 11:45 you know not not only do you do open research 你知道你不仅做开放式研究 11:48 yeah we now do open engineering 是的，我们现在做开放工程。 11:51 and so that open engineering is extremely powerful 所以开放工程是非常强大的 11:54 because then you could take my contributions 因为那样你就可以拿走我的贡献 11:57 add your contributions i add my contributions and as a result 添加您的贡献，我添加了我的贡献，结果 12:02 the the the innovation pace is not 创新的步伐不是 12:06 just the uh contribution of each company 只是每个公司的呃贡献 12:09 or each engineering group 或每个工程组 12:11 but the combined resource of an ecosystem and so 但是生态系统的综合资源等等 12:16 that's very clever about open source engineering here in China 中国的开源工程非常聪明 12:20 but don't forget that open source has many global implications 但不要忘记开源具有许多全球影响 12:26 you know 你知道的 12:27 not only did the open source models help the Chinese ecosystem 开源模型不仅帮助了中国生态系统 12:32 it's helping the ecosystems around the world 它正在帮助世界各地的生态系统。 12:35 this is the best opens you know 这是你知道的最好的开场 12:38 r one and q n r one和q n 12:40 and Kimi are the best open reasoning models in the world Kimi是世界上最好的开放式推理模型 12:45 today multi modal reasoning model 当今多模态推理模型 12:47 so it's very advanced 所以它非常先进 12:49 and so it doesn't matter who you are 所以你是谁并不重要 12:51 you could be a healthcare company or financial services company 你可以是一家医疗保健公司或金融服务公司 12:54 or robotics company 或机器人公司 12:55 you could take advantage of this 你可以利用这一点 12:57 and modify for your own own application yep 并为您自己的应用程序进行修改，是的 13:00 it is also very important to note that 同样非常重要的是要注意 13:03 open source is the safest way to advance 开源是最安全的前进方式 13:07 you know sunlight is the best disinfectant 你知道阳光是最好的消毒剂 13:11 and so when open source and all the open innovation 因此，当开源和所有开放创新 13:15 you invite global scientific scrutiny 你邀请全球科学审查 13:20 and when you have global scientific scrutiny 当你有全球科学监督时 13:23 the quality of the work is goes up 工作质量提高了 13:27 if you look at the deep seek paper yeah 如果你看看深度搜索纸是的 13:29 it is incredibly well written yeah 它写得非常好，是的 13:33 it is incredibly well written 它写得非常好 13:34 it is absolutely uh you know a plus quality 这绝对是呃，你知道的，质量很好。 13:38 uh huh science and a plus quality engineering 嗯哼科学和一个加质量的工程 13:42 and so it's they did it all completely openly 所以他们完全公开地做这一切 13:45 and it invited education and learning and sharing 它邀请了教育、学习和分享 13:50 as well as the benefit of many people scrutinizing it 以及许多人仔细检查它的好处 13:53 so it's very good good for safety 所以非常安全 13:55 yeah thank you 是啊谢谢 13:56 and by the way both deep sea and queen are from hangzhou 对了深海和皇后都是杭州的 14:00 and i'm native of hangzhou 而且我是杭州人 14:02 and i'm very proud of the city 我为这座城市感到骄傲 14:04 and you have my personal invitation to visit Hangzhou 我亲自邀请你去杭州 14:06 next time in your next trip okay 下次在你的下一次旅行中，好吗？ 14:08 Hangzhou is 杭州是 14:09 the may i dare say it's the Silicon valley of China is it 我敢说这是中国的硅谷，是吗？ 14:13 uh i would say a lot of don't say the Silicon valley of China 呃，我会说很多，不要说中国的硅谷。 14:18 it's the the hangzhou will be an an innovation hub 杭州将成为创新中心 14:22 for the rest of the world okay 对世界其他地方来说，好吗？ 14:23 all right said it's very unique okay 好吧，这是非常独特的，好吧。 14:25 uh huh 嗯哼 14:25 it's it's very again 又来了 14:27 you have a personal invitation to visit Hangzhou 你有去杭州的私人邀请 14:29 in your next trip会 你下次旅行会 14:31 会来一定会来的 会来一定会来的 14:32 thank you and you're talking about the open science 谢谢你，你说的是开放科学 14:35 the open engineering 开放工程 14:35 so in your in your in the gdc last year and you said okay 所以在你去年在GDC的时候，你说好的。 14:41 it's the first time in the human history 这是人类历史上第一次 14:44 and we have opportunity to to 我们有机会 14:47 to 到 14:48 to turn actually biology as an engineering instead of science 把生物学变成工程学而不是科学 14:54 okay it is also incredible 好吧这也太不可思议了 14:56 so what's the long term impact of ai on the scientific 那么人工智能对科学的长期影响是什么？ 15:02 discovery and technology innovation 发现与技术创新 15:05 so will 也会的 15:05 ai will change the way for the scientists to do their research 人工智能将改变科学家进行研究的方式 15:09 yeah it's you know today we're only talking about ai for human 是的，你知道今天我们只谈论人工智能。 15:13 yeah 是呀 15:14 ai for science is where we will make the greatest impact yeah 人工智能科学是我们将产生最大影响的地方，是的 15:18 now remember ai for human is relatively easier 现在记住人工智能对人类来说相对容易 15:22 and the reason for that is 原因是 15:23 because humans created the human language 因为人类创造了人类语言 15:26 yeah 是呀 15:27 and it is easy to use uh design tools yeah you 而且很容易使用uh设计工具yeah you 15:31 and i we've been 而我我们已经 15:32 we've been using design tools to make chips for a long time yeah 很长一段时间以来，我们一直在使用设计工具来制造芯片，是的 15:36 but the transistors were designed by us yeah 但是晶体管是我们设计的 15:40 so that we could use tools to manipulate the transistors 这样我们就可以使用工具来操纵晶体管 15:44 and the chip design yeah 还有芯片设计是的 15:46 but biology was created by nature yes 但是生物是自然创造的，是的 15:50 and so we have to use in order to manipulate biology first 所以我们必须首先使用才能操纵生物学 15:55 we have to understand it and finally 我们必须了解它，最后 15:57 we have a new capability called artificial intelligence and 我们有一种叫做人工智能的新能力 16:01 we can understand 我们可以理解 16:03 learn and understand the meaning of proteins 学习和理解蛋白质的含义 16:07 the meaning of chemicals 化学品的含义 16:09 the meaning of cells 单元格的意义 16:11 and uh meaning of life of course 当然还有生命的意义 16:14 and so 所以 16:15 we can even understand the meaning of the metabolic reactions 我们甚至可以理解代谢反应的含义 16:21 and actions in the human body right 和人体内的行为，对吧？ 16:23 and so 所以 16:24 if we can under 如果我们能下 16:25 if we can use ai to first 如果我们可以先使用ai 16:27 understand the structure and the meaning then 然后理解结构和含义 16:30 we can use ai to improve to configure to design design drugs 我们可以使用人工智能来改进配置来设计设计药物 16:38 and uh help people live longer 还有呃帮助人们长寿 16:42 so it's a lot of opportunities 所以机会很多。 16:43 big opportunity 大好机会 16:44 the other thing that we can use use uh ai 我们可以使用的其他东西使用uh ai 16:47 for is to uh emulate physics 就是模仿物理 16:51 you know today 你知道今天 16:52 we use uh physics equations to simulate 我们用呃物理方程式来模拟 16:55 very complicated uh interactions like weather 非常复杂的呃相互作用比如天气 17:01 you know weather is a cloud physics 你知道天气是云物理学 17:04 high cloud physics 高云物理 17:05 low cloud physics 低云物理 17:06 atmosphere of physics we have ocean physics 物理学的大气层，我们有海洋物理学。 17:09 ice physics land physics we have uh uh conduction 冰物理陆地物理我们有嗯嗯传导 17:14 we have convection um you know 我们有对流嗯你知道的 17:17 and so all of these different types of physics yeah 所以所有这些不同类型的物理学是的 17:20 has to come together right 必须走到一起，对吧？ 17:22 very very small scale physics 非常非常小尺度物理学 17:25 also to very large scale physics we call it meso scale physics 对于非常大尺度的物理学，我们称之为中尺度物理学。 17:29 yeah and the time the time domain travels from uh 是的，时域的时间从呃 17:33 probably in the case of 可能在 17:35 in the case of physics from seconds to maybe multiple years 在物理学的例子中，从几秒钟到几年 17:40 and so that range is very complicated for simulation to do 所以这个范围对于模拟来说非常复杂 17:43 i see but maybe we can teach an ai to help us predict that 我明白了，但也许我们可以教人工智能来帮助我们预测 17:49 and you know ai is much 你知道ai很多 17:50 much faster at predicting than using physical simulations 在预测方面比使用物理模拟快得多 17:54 and so i have every confidence that whether 所以我完全有信心 17:57 it's using ai to understand the laws of nature okay 它用人工智能来理解自然法则，好吗？ 18:02 or using ai to emulate the laws of nature 或使用人工智能来模仿自然法则 18:06 we could use ai to help us advance science very big deal 我们可以用人工智能来帮助我们推进科学，这很重要。 18:11 yeah it is an incredible yeah 是的，这是一个不可思议的是的 18:13 truly truly 真真 18:14 incredible yeah truly 难以置信，是的，真的。 18:15 and so my next probably is hard for you 所以我的下一个可能对你来说很难。 18:17 i'm sorry to say that you know 我很抱歉地说你知道 18:18 Jason and you know that actually Jason和你都知道其实 18:21 today's ai technology is heavy depend on Silicon technology okay 今天的人工智能技术很重，依赖硅技术好吗？ 18:26 it's depend on Silicon okay 这取决于硅，好吗？ 18:28 and you know 而且你知道 18:29 we're using the silicon to increase the computing power 我们用硅来提高计算能力 18:33 and get the gigantic memory space 并获得巨大的内存空间 18:36 and even you know 甚至你也知道 18:38 this unbelievable communication bandwidth okay 这难以置信的通信带宽好吧 18:40 it's all depend on silicon okay 这一切都取决于硅，好吗？ 18:42 so my question for you is you know 所以我的问题是 18:45 you know in the next ten or twenty years are 你知道在接下来的十年或二十年里 18:48 we still being able to rely on the silicon 我们仍然可以依靠硅 18:51 for the advance of ar technology okay 为了AR技术的进步好吗？ 18:54 yeah you know 是啊你知道的 18:55 of course a 当然a 18:56 silicon technology is already adding so many different elements 硅技术已经添加了许多不同的元素 19:00 yeah you know 是啊你知道的 19:01 it's barely silicon and so uh 它几乎不是硅，所以呃 19:03 so i think that that we will continue to advance in those areas 所以我认为我们将继续在这些领域取得进展 19:07 several areas the transistor will become three dimensional yep 晶体管将成为三维的几个领域是的 19:11 and you know we call it gate all around yep so right now 你知道我们把它叫做门，是的，所以现在 19:14 it's nanosheet the next generations called gate all around 这是纳米片，下一代叫做门。 19:17 and then 然后 19:17 after that we will have transistors on top of transistors 在那之后，我们将在晶体管之上有晶体管 19:20 you know stacking finfets 你知道叠鳍 19:22 uh instead of 呃而不是 19:23 instead of distributing power over the surface of silicon 而不是在硅表面分配能量 19:27 we distribute it on two sides we saw backside power yeah 我们把它分布在两边，我们看到了背后的力量，是的 19:31 uh instead of uh uh而不是uh 19:33 instead of uh 而不是呃 19:34 one chip at a time hmm 一次一个芯片嗯 19:35 we now have uh stacking chips yeah 我们现在有呃堆叠芯片是的 19:38 multi chips yeah 多芯片是的 19:40 and so the packaging becomes very advanced yeah 所以包装变得非常先进，是的 19:43 we call it cooss um hmm right 我们叫它cooss嗯嗯对 19:45 Nvidia was the first company to use cooss at a very large scale 英伟达是第一家大规模使用cooss的公司 19:49 and even even in the future 甚至在未来 19:51 the packages will not be this big 包裹不会这么大 19:53 but the package will be entire panels 但包裹将是整个面板 19:55 okay 好吧 19:56 so that could the size of a chip 所以芯片的大小 19:58 could be the size of this table 可能是这张桌子的大小 20:00 and then beyond that we'll use silicon photonics 除此之外，我们将使用硅光子学。 20:04 directly attach photons to electrons 直接将光子附加到电子上 20:08 with very very tight coupling 具有非常非常紧密的耦合 20:09 we call it cpo yeah 我们称之为cpo yeah 20:11 and then we can connect many of these things together wow 然后我们可以把这些东西连接在一起哇 20:15 the the number of dimensions you know 你知道的维度数 20:17 the number of dimensions of capability is incredible 能力的维度之多令人难以置信 20:21 yeah i see 是啊我明白了 20:22 the silicon technology is amazing you know 硅技术是惊人的你知道 20:24 yeah we have 是啊我们有 20:25 we have uh we have plenty of work to do for at least two decades 我们有呃我们有很多工作要做至少20年 20:29 and the reason why we kind of know 我们知道的原因是 20:30 that is 那是 20:31 because NVIDIA's roadmap is already coming up on one decade yeah 因为英伟达的路线图已经出现了十年 20:36 right so it's at least somewhere between five to ten years 对，所以至少需要五到十年的时间。 20:39 and we are already dreaming and designing 我们已经在梦想和设计了 20:42 and architecting systems that are ten years from now 并构建十年后的系统 20:45 so i i'm pretty sure 所以我我很确定 20:47 we could see ten years and the next twenty years 我们可以看到十年和下一个二十年 20:49 i'm i'm fairly certain we will have plenty of work to do 我很确定我们有很多工作要做 20:52 yeah i i truly believe you know when you have you 是的，我真的相信你知道你什么时候拥有你 20:55 and i are gonna be busy for at least twenty years 我至少要忙二十年 20:57 you're really like to eat particularly 你真的特别喜欢吃 20:59 if i have good architecture 如果我有好的建筑 21:00 when you have a good architecture and a silicon 当你有一个好的架构和一个硅 21:03 you get a much more things you can do with that okay yeah 你可以用它做更多的事情，好吗？是的。 21:05 that's right and so 没错，所以 21:07 so my last question is really about the people 所以我的最后一个问题是关于人民的 21:10 and particularly about young people 尤其是年轻人 21:12 and you ask me 你问我 21:13 what i'm doing recent years 我最近几年在做什么 21:15 you ask me before and actually every year 你以前和实际上每年都问我 21:19 i'm you know together with thousands of volunteers 我和成千上万的志愿者一起 21:24 and and and young people and and and年轻人 21:26 and we're working on the events named twenty fifty okay 我们正在进行名为2050的活动，好的。 21:31 with a very simple idea 有一个非常简单的想法 21:33 science technology brings people together 科学技术将人们聚集在一起 21:36 particularly young people so technology 尤其是年轻人，所以技术 21:38 connects people technology 连接人的技术 21:40 is that it's more than just technology 它不仅仅是技术 21:42 it's really connects people together 它真的把人们联系在一起 21:45 you have always dedicated 你一直致力于 21:47 so much of your own time to nurture and advice young people 你花了太多时间培养和建议年轻人 21:53 uh actually 呃其实 21:54 that's my passions yeah 这就是我的激情 21:55 i know ever since ever 我知道从那以后 21:56 since you and i met i knew that i got a lot of a lot of you know 自从你和我相遇，我就知道我得到了很多很多你知道的 22:00 help when i was when i was young 帮助当我年轻的时候 22:03 so i think 所以我想 22:04 it's always very exciting to talk with the people yeah 和人们交谈总是非常令人兴奋 22:07 so it's that you know to be the twenty fifty sometime 所以你知道有时候要成为二百五十岁 22:09 you can see all the young people 你可以看到所有的年轻人 22:11 and these are all the people you know 这些都是你认识的人 22:13 they don't know where their future is 他们不知道自己的未来在哪里 22:15 but they really worry about the future of the world okay 但他们真的很担心世界的未来 22:17 yeah so it's incredible young people Jen he is the hero 是的，所以这太不可思议了，年轻人，珍，他是英雄。 22:22 he's an incredible hero thank you yeah 他是一个不可思议的英雄，谢谢你，是的。 22:24 superhero so so 超级英雄如此如此 22:25 my question is is really about you know 我的问题其实是关于 22:28 do you have any specific advices and you know today 你有什么具体的建议吗？你今天知道吗？ 22:31 actually everybody knows actually 其实大家都知道其实 22:33 ai is a lifetime opportunity for most of us okay 人工智能对我们大多数人来说是一生的机会，好吗？ 22:38 and particularly for the young people 尤其是对年轻人来说 22:40 so do you have any specific advices on the youngest us 那么你对最年轻的我们有什么具体的建议吗？ 22:44 or do you have any plan to do something specifically for them 或者你有什么计划专门为他们做些什么吗？ 22:48 okay 好吧 22:48 well you know people say that ai is of course 你知道人们说人工智能当然是 22:52 solving math problems 解决数学问题 22:54 reasoning problems uh 推理问题呃 22:56 it can solve uh 它可以解决呃 22:57 uh programming problems 呃编程问题 23:00 it could even code by itself and so therefore 它甚至可以自己编码，因此 23:02 we probably don't need to learn 我们可能不需要学习 23:04 those that's exactly wrong you know 你知道那些完全错了的人 23:07 in fact that you all no matter 其实你们都无所谓 23:10 no matter as you know uh we do less programming uh 不管你知道的呃我们少做编程呃 23:13 huh as we develop our career we do less engineering but you 嗯，随着我们职业生涯的发展，我们做的工程越来越少，但你 23:17 always have to still learn how to think from first principles 总是要学习如何从第一原则思考 23:21 yeah to take a very complicated problem 是的，要解决一个非常复杂的问题。 23:25 which we've never encountered before yeah 我们以前从未遇到过 23:27 and break it down step by step by step yeah 并一步一步地分解它是的 23:30 that is built up on first principles yeah fundamental 建立在首要原则之上，是的，基本原则 23:33 knowledge right 知识产权 23:35 and so conventional wisdom is not very good to rely on 所以传统智慧不是很好的依据 23:39 you always wanna go back to first principle thinking 你总是想回到第一原则 23:42 and so 所以 23:42 we have to teach people first principle thinking otherwise 我们必须教会人们以其他方式思考的第一原则 23:45 you cannot have a critical mind yeah 你不能有批判性的头脑 23:48 and if you don't have a critical mind 如果你没有批判的头脑 23:49 you cannot tell if the answer from someone 你不知道某人的答案是否 23:53 or the answer from ai is make sense 或者人工智能的答案是有意义的 23:57 or not you need to be able to interact with the ai one 或者不，您需要能够与ai交互 24:01 you have to describe the problem for the ai to help you solve 你必须描述问题让人工智能帮你解决 24:05 two yep you have to reason about 两个是的，你必须推理 24:09 whether the ai is answering the question 人工智能是否在回答问题 24:12 properly or optimally as well 适当地或最佳地 24:15 as it can yes and so critical thinking is always very important 是的，批判性思维总是非常重要的。 24:18 whether it's critical thinking based on physics 无论是基于物理学的批判性思维 24:21 mathematics or logic 数学还是逻辑 24:24 critical thinking is fundamental to almost everything 批判性思维几乎是一切的基础 24:27 that we do and 我们所做的和 24:27 i would advise that young people 我建议年轻人 24:29 today still continue to learn math 今天还是继续学数学 24:32 and reasoning 和推理 24:33 and logic and right computer programming 以及逻辑和正确的计算机编程 24:36 even though you don't have to do it 即使你不必这么做 24:38 you should know it yes 你应该知道是的 24:39 that's number one 那是第一 24:41 the second thing i would say is almost everything 我要说的第二件事几乎是一切 24:44 every single young person today 今天的每个年轻人 24:46 uh should absolutely as fast as possible engage ai 呃绝对应该尽快参与ai 24:51 yeah 是呀 24:52 this is the new computer 这是新电脑 24:54 you know ai is makes the computer very very powerful 你知道人工智能使计算机非常非常强大 24:59 but it's very important to realize that 但认识到这一点非常重要 25:01 it has become very easy to use 它变得非常易于使用 25:05 because it understands how we interact 因为它了解我们如何互动 25:08 no matter how right 无论多么正确 25:09 and if you don't know how to use the ai 如果你不知道如何使用人工智能 25:12 you say to the ai i don't know how to use ai teach me 你对ai说我不知道怎么用ai教我 25:15 how to use ai 如何使用人工智能 25:16 and it will teach you step by step 它会一步一步地教你 25:19 and so the computer with ai 所以带ai的电脑 25:22 has been the most powerful equalizer of all people 是所有人中最强大的均衡器 25:29 and so whether you are farmer or you know elder person 所以无论你是农民还是老年人 25:34 or you don't know how to use a computer young person 或者你不知道如何使用电脑年轻人 25:37 you absolutely must engage ai as quickly as possible 你绝对必须尽快参与人工智能 25:41 i think it will really empower you 我认为它会真正赋予你力量 25:43 and then lastly 然后最后 25:44 i'm jealous of the young generation because yeah 我嫉妒年轻一代，因为是的 25:47 you know this is the generation that are being born right now 你知道这是正在诞生的一代。 25:52 where they will grow up with their own ai for their life 在那里他们将与自己的ai一起长大 25:55 yes 是的 25:57 it's like having it's like having star wars r two d two yeah 就像拥有它就像拥有星球大战r 2 d 2 yeah 26:01 yeah grow up with you 是啊和你一起长大 26:02 your whole life 你的一生 26:04 you know and having this 你知道，拥有这个 26:06 this ai companion that remembered everything your whole life 这个记得你一生一切的人工智能伴侣 26:10 and was able to advise you and you know 并且能够为您提供建议，您知道 26:14 teach you and uh huh 教你然后嗯哼 26:15 right uh huh your whole life 对吧嗯哼你的一生 26:16 it's just an amazing idea and i i'm jealous that that you know 这是一个很棒的想法，我很嫉妒你知道 26:21 i didn't have an ai uh huh 我没有啊啊哈 26:23 that remind reminded me and help me 那个提醒提醒了我并帮助了我 26:26 and remembered every everything in my life 记得我生命中的每一件事 26:28 since i was a child yeah you know 从我还是个孩子的时候 26:31 could you imagine 你能想象吗 26:32 you have a you have a ai and you say uh 你有一个你有一个ai你说uh 26:36 what was i doing 我在做什么？ 26:37 when i was one years old or two years old or three years old 我一岁两岁三岁的时候 26:41 you know tell me about where i was 你知道告诉我我在哪里 26:43 and what were the things 事情是什么？ 26:44 that i worked on and what were the things that i right 我做过什么，我做对了什么？ 26:47 talk to you about it would be i'll remember that 和你谈谈这件事，我会记住的。 26:51 and that entire journey would have been captured i 整个旅程都会被我捕捉到 26:54 i wish i wish uh 我希望我希望呃 26:55 we had that opportunity uh 我们有那个机会呃 26:57 and thank you for your encouragement and also advices actually 谢谢你的鼓励和建议 27:01 you know when i first met with you 你知道我第一次见你的时候 27:02 your passion about technology inspired me 你对技术的热情激励了我 27:06 and i i still can see the passion you have 我仍然可以看到你的激情 27:08 today okay and also 今天还可以 27:10 you're very patient thinking about you know 你很有耐心思考你知道 27:12 your company group of staff to where you are 您的公司团队到您所在的位置 27:15 you are you 你就是你 27:16 you have you know 你有你知道的 27:17 over the fortune uh 在财富之上呃 27:20 market value this incredible journey 市场价值这个不可思议的旅程 27:23 so passion and patience really important 所以激情和耐心真的很重要 27:25 for the young people and thank you for your advices for that 为了年轻人，谢谢你的建议。 27:28 thank you thank you and thank you for being such a good friend 谢谢谢谢谢谢谢谢你这么好的朋友 27:31 all these years 这些年来 27:31 and and i've enjoyed i enjoyed both of our careers you know 我很享受我很享受我们的事业 27:36 advancing together 共同前进 27:37 and and this is a this is a once in a lifetime 这是一生中仅有的一次 27:41 opportunity really yeah and in fact 机会真的是的，事实上 27:44 you know you could even say that this is a once in a generation 你知道你甚至可以说这是一代人中的一次 27:47 opportunity and ai will define the world the 机会和人工智能将定义世界 27:52 next century and so this is very important time 下个世纪，所以这是非常重要的时刻 27:56 and i'm glad that that there's so much interest 我很高兴有这么多人感兴趣 27:59 and quite frankly 坦白地说 28:00 i'm delighted to see so much advance here in China 我很高兴看到中国取得了如此大的进步 28:04 so much expertise 这么多专业知识 28:05 so much so much advance and and let's let's work 前进了这么多，让我们工作吧 28:10 let's continue to work together 让我们继续合作 28:12 and i look forward to having another sit down 我期待着再坐下来 28:15 chat in another ten years 再过十年再聊 28:16 and see where we are 看看我们在哪里 28:17 you're very tough so the best is yet to come okay yes 你很坚强，所以最好的还在后头，好吧，是的 28:20 okay yeah 好吧是 28:21 thank you thank you 谢谢谢谢 28:22 thank you thank you yeah 谢谢谢谢你耶

2025 help grok clarify ai purposes of HK, Singapore & Asean  --**Hong Kong** - **Action Plan**: During a visit where he received an honorary doctorate from HKUST, Huang pushed for AI factories to support finance and logistics, enhancing Hong Kong’s innovation hub status. - **Unique Aspect**: He highlighted its role as a bridge between East and West, advocating for urgent intelligence sharing while respecting sovereignty. #### **Indonesia** - **Action Plan**: Huang promoted affordable AI infrastructure for Indonesia, encouraging the nation to build sovereign AI systems that reflect its culture and address urbanization challenges. - **Unique Aspect**: He framed Indonesia as an emerging economy that could leapfrog development stages with AI factories and digital twins. chris.macrae@yahoo.co.uk

Singapore's founder Lee Kuan Yew determines an economics policy of jobs jobs jobs which evolved into being the most intelligent isle with great colleges and education at every ge group. In parallel, Tawain grandfather ogf tech H Li bet the country on chips (the tech kind); and H0ong Kong philanthrpoist Li Ka Shing is foi=ound cvonnecting chairs around the world with hsi support of HK colleges.

Typically all these island lead high school maths tanking making usa at 30th beyond sad; at least a decade ago all of tehse island nations were demanding partoicipation of every grade of tecaher in AI as a future affairs subject demanding curiosity - see eg ai singapore.org

For those who may argue its easier for a wealth isle to get education and health right for everyone singpaore's good neigbours commitment to asean is worth noting and takest its ai form with SEA-LION (Southeast Asian Languages In One Network): Developed by AI Singapore, it is an open-source LLM trained on 11 Southeast Asian languages including English, Chinese, Indonesian, Vietnamese, Malay, Thai, Burmese, Lao, Filipino, Tamil, and Khmer. It's designed to understand the diverse contexts, languages, and cultures of the region, according to SEA-LION.AI. SEA-LION was built prioritizing Southeast Asian linguistic and cultural contexts, using 13% regional language content compared to Llama 2's 0.5%

additionally gemini nov 2025 adds:

Singapore

 

The government actively promotes AI for citizen-centric and community-focused applications under its Smart Nation initiative: 

• Citizen services: As part of its Smart Nation 2.0 plan, 
  Singapore
   is using agentic AI to simplify engagement with government services. For example, a single portal could automatically connect a jobseeker to complementary services like benefits and grants.
• AI safety and talent: The government partners with global companies and institutions to build a network for AI safety, research, and talent development to support a thriving, community-focused AI ecosystem. 

some uae partner with west or womens empowerment subnetworks- with special thanks to friends in qataer (wise sheika moza education prize, for fazle abed, womens college main partners) dubai- legatum who at mit continued sponsorship of village phones with quadir family in bangladesh ...

welcome to this session of The Energy & AI podcast (subnetwork of DSR  sponsored by Japan's NEDO hosted Professor David Sandalow) supported by the embassy at www.uae-ebassy.org or search for UAE- US tech cooperation   --1  -- we thank them for their support  and we look forward to it developing and growing over time because the issue is so  important

in this podcast we welcome jensen huang to the AI Energy and Climate Podcast thanks for 1:19 joining us  and I'm looking forward to 1:24 discussing a range of topics including the potential benefits of AI power demand for AI your plans at Nvidia and 1:31 more and maybe let's start with the potential benefits of AI what do you see as some of the most important ways that 1:37 artificial intelligence can contribute to solving energy and climate change problems 

1:43 Jensen Huang" well  artificial intelligence has the ability to understand uh data understand 1:49 if you will information at a scale no humans can; and  multimodal information from temperature to wind 2 speed to pressure etc all at the same time; and it could study information 2 across a very large scale very large longitudinal scale and so so the ability   to understand information and and then be able to predict the future quite an extraordinary power

 but let me take a step back and and say something about about how we got here - as you know we at nvidia are now a third of a century into inventing  a new way of doing computing called accelerated computing and and we reduced  the amount of energy required to do computation  by 25X or more in  order to increase performance; the fundamental way of increasing performance is to reduce computational energy  ; together with this our partners have supported debel;oping new way of doing computation called machine (dee) learning and in turn we have 3:33 this new way of doing AI and of course a lot of the conversation we're going to have is that there's two phases of AI:

 there's the learning phase we call it the training phase of the model 

and then there's the inferencing 3:46 model which is the application phase 

it's a little bit like humans you know we we have to go spend years pre-training going to school and then afterwards we get to go apply it and inference 

 4:01 both inferencing and part training requires a lot of energy but the good news is that 4:09 most of most of the world's computation breakthroughs for humanity really going to be about inference where compuation energy can be low  so we'll be able to talk about that during this talk 

david: that's interesting your comments about driving energy consumption down are 4:28 counter to most of the narrative that we hear these days all this attention about how AI is driving up energy demand so 4:35 power demand could you just say say a word about what you mean about how AI is driving energy consumption down

jensen  yeah you 4:41 have to look at AI in its entire longitudinal aspect you know because the 4:47 goal of AI is not to train the model and these data centers that we use to train the model are quite large and you you uh 4:55 you you you apply basically three things uh uh you have to invent a model and 5:01 that model is is uh has to have the ability to uh learn the uh predictive 5:07 features from the data that you're presenting and so it could be a lot of videos or it could be um large amounts 5:14 of uh telemetry data that's taken you know about about weather or the climate um it could be it could be uh uh a whole 5:22 bunch of words that you want to learn something about the language and so you collect an enormous amount of data and 5:28 you have a you have a model that you have to train and it takes the computation takes a computer to uh 5:35 iteratively uh figure out how to predict uh uh understand the the features and 5:42 how to predict the outcome and so that takes a large computer to do but the goal of 5:48 course is not just to train the model the goal is to use the model now once 5:53 you train a large model and the model is quite large you know it's trillions 5:58 parameters large but then what you do is is you've now created if you will a professor model you take this professor 6:05 model and you distill it down into a whole bunch of little student models and so the actual model that you use is not 6:12 that big it could fit in a cell phone it could fit in a PC it could fit uh in a self-driving car for example just one 6:18 chip on a self-driving car um you know and and the response time is really good it's really fast just as you know when 6:25 we're using chat GBT or Gemini the response time is super fast and because the model has been it trained in a very 6:32 large model but it distilled it into smaller medium small tiny models now of 6:38 course these tiny models have the skills of the large model but not the general 6:43 general generalization capability um of the large models and so it tends 6:49 to be a little bit more brittle it tends to be more specific but it has the capabilities to you know to perform 6:54 whatever task that it was fine-tuned to do 

and so in the end what you have is you you train a large language model but 7:01 you inference a whole bunch of smaller models and and those small models require a lot less energy let me give 7:07 you one example so we trained uh trained a large model to to u predict uh 7:13 weather and and um uh the models that it learned from are principled simulation 7:20 models and so these are you know classical um uh physical physically based um 7:27 principled physics simulations and we would use we would use those models and observable observed 7:35 data to uh train a large langu a large simulation model large physics model physics space model but when we're done 7:43 we predict weather 10,000 times less energy instead of using a 7:50 supercompuert to predict the weather and this these supercomputers are running 247 trying to predict tomorrow's weather 7:56 and next week's weather now we could use an AI model that predicts tomorrow's weather next week's weather 10,000 times 8:04 less energy 

david: i think that's a point that's not widely understood that these new weather models actually use less 8:09 energy than some of the old physics based models do and and that's had real applications in solar and wind farms and other places ...

jensen:  yes david that's a big idea  - more at video above

===

uea major sponsor of ity plenipotentiary 4G to 5g development 2015-20 while itu led by houlin zhao, and while jim kim, jack m,a fei-fei li. melinda gates helped launch ai for good

====

the emirates minister fir ai,  alolama briliant interview - one of first data sovereignty jensen huang  

============

 

 q to grok 7/27/25 This is a transcript if csis interview with world bank economist gill. He has also just hosted the annual abcde human development economics summit out of washington dc. Can you summarise this and if you see it helping make case for india or other countries most urgent ai needs, please clarify .

....ne, Naven Girishanka here at CSIS. Welcome to betting on America. For 0:05 this episode, I'm sharing with you a conversation I had with the World Bank chief economist Indit on global growth 0:12 prospects, challenges that developing countries are facing and importantly opportunities that developed countries 0:18 have in really solving development challenges going forward. What I thought was particularly valuable 0:25 for those who care about US industrialization were Ind's insights on the sources of structural imbalances in 0:32 trade. His thoughts on populism and the future of economic reform and importantly 0:39 opportunities that advanced technologies offer not only for developing countries but importantly for US exports. It was a 0:46 really interesting conversation for anyone who cares about US industrialization. 0:52 [Music] 1:01 We are living through an era of unprecedented trade and policy uncertainty. Uncertainty that is flowing 1:07 through the global markets affecting global growth prospects particularly in developing countries. 1:13 What does the rise of populism mean for the future of economic reform? reforms that could generate growth and reduce 1:21 poverty around the world. What is the rise of new technologies and 1:26 multiple technology revolutions like AI, biotech, clean tech, nextgen telecom 1:33 mean for the opportunities for countries to leapfrog and address long-standing development challenges with new 1:39 solutions? We're going to talk about this and more with Indgill, my good friend, chief economist and senior vice 1:45 president at the World Bank Group. Indid, thank you so much for joining us. I'm delighted to have you here. 1:51 It's a pleasure to be with you, Navi. Um, it's a pleasure. You're you've recently come out with the global economic prospects report. I 1:58 wanted to start there as one of the flagship reports of the World Bank. Um, 2:03 the the team yourself and your team have put out really growth projections that are quite sobering and they project uh 2:11 global growth to deceler decelerate to the slowest rate since 2008. And for 2:16 developing countries, the picture is even starker. Growth has fallen in the last decades, in the 2000s, from 5.9% 2:23 to just under 3.7% in the 2020s. Um, a lot of this also has to do with 2:30 unprecedented levels of trade and policy uncertainty that we're experiencing now. I think I I read a blog where you say 2:38 developing countries are now a development free zone. I found that a striking statement. So help us 2:44 understand what's going on. Got it. Excellent question. So we did a 2:50 fairly big report that looked at uh the long-term growth prospects of countries around the world. And the striking 2:55 finding Naven was that across the world um the potential growth rates were going 3:01 down. So they were they were very healthy in the 2000s and then they went down in the 3:07 2010s and they ratcheted down in the 2020s. Um and so as a result of that I 3:14 think I think that that's so that's a backdrop in a sense. Yeah. Now there are some countries that are exceptions by 3:19 the way one country that is an exception is India. Mhm. Right. Right. And there are a few others 3:25 like Indonesia and others which which have tended to maintain. For the others where you haven't seen a drop they're 3:31 actually already at very low levels. Okay. Like like essentially Latin America and 3:36 Africa. Okay. Right. So that that that's a backdrop. So then beyond that if you sort of look at it 3:42 there are two other trends. The first trend is of course is a very cross-sectional one in the sense that as you go from as you go from emerging 3:50 markets like India, China, Indonesia and others their prospects are not that bad. Okay. 3:55 Then when you go to the smaller economies and the poorer economies their prospects tend to be worse. So for an 4:02 example I mean if you look at Africa uh subsaran Africa I think the nearest 4:08 approximation for their growth over the last uh decade or so is a per capita 4:15 income growth of zero essentially okay and I think that when we look at the prospects looking ahead the prospects 4:21 look like that decade is going to get repeated for many of these countries the low-income countries especially 4:28 so if if you uh so that's the D that is the second part. The third part of 4:34 course is that if you look at the most recent shocks like trade policy uncertainty and things like that they 4:39 tend to be they tend to hurt the smaller economies a lot more because smaller economies depend a lot more on trade. 4:46 Yeah. Market access market access. Uh so for all of those reasons thing now when I said 4:52 development free zone you know I guess the best way to put it is this if you 4:57 looked at the MDG period the period of the millennium development goals right. 5:02 Mhm. You know we said we made a lot of progress there etc etc but a big chunk of that progress was China Navin. 5:09 Okay. Okay. for actually for many of the indicators in in other parts of the world there was regression not 5:15 because but China so this was the first decade of the first decade when we're looking at this 5:21 decade now and look at the period of the sustainable development goals instead of 5:27 China you've got to India so India is actually making things look a lot better 5:32 for the world but actually you have a lot of other countries the the poorer 5:37 countries the countries in conflict etc. They are the ones they actually 5:43 have not they're not seeing development right that's what I meant by a development free zone 5:48 I got it so given the size of both those countries China in the first decade India in the second decade the 5:54 aggregates look better than what's actually happening in the case of say countries in Africa or countries in 6:00 Latin America absolutely and so then you say okay what happens next who's going to replace India and China 6:07 now they're gone right and the prospects don't look But right that's uh yeah 6:13 so um let's just double click on this because I want to understand the drivers. So if like I don't want to 6:20 overindex on what's happened right now with uncertainty although that's something we should get into. Yes. But when I when I when we think about the 6:27 first decade of the century and the second decade of the century, there have been some important events. The global 6:32 financial crisis, the pandemic shock, but um were those the main drivers of 6:38 reduced growth prospects during that period or was it other things? 6:44 So, so those are those are things that actually ratcheted things down rather quickly. But the things that actually 6:50 lead to declining drivers of growth are more long-term. So one of them of course 6:56 is you can sort of think about three things. The first part is is capital. 7:01 The second part is labor. The third part is energy and then the fourth part is technological progress. Right? 7:07 So if you look at all of this stuff what you find is in ter demography you find many of these countries are 7:13 aging. Right? Not all of them, not Africa, not India yet, that kind of thing. But but that's 7:18 one of them. So if you look at East Asia for example, a big part of the declining drivers of 7:24 potential growth was demography. But the second part is equally important in the sense that if you look, you know, 7:30 many of these countries based their strategies on a lot of investment investment, but investment paid off less 7:37 and less with each dollar in terms of growth and so on. So that at that part 7:43 essentially is declining productivity growth in these countries. So in a sense these countries had to 7:48 improve their the way that uh the way that they combine all of these factors 7:53 of production and things like that but with more and more efficiency because you can't rely just on more work 8:01 and more investment and so on that but that only takes you up to a point. So these countries all hit that phase where 8:07 they needed to generate greater greater efficiency and they failed in doing that or they're increasingly failing in not 8:14 all of them. And now if you say okay but what are the sort of uh reasons why the 8:21 '9s were good the 2000s were even better and then we sort of hit this declining trend. 8:27 There were these there was a consensus. The first consensus was let markets take the lead and let governments give them 8:35 support and I think that spread across the world. That but that that was a very American idea by the way. I mean it's a 8:41 great idea. The second one was macro stability is really important especially for the 8:47 poor. So inflation is slow growth is bad but inflation is a killer for for 8:53 poverty reduction. And there was a sense that don't don't spend beyond your budgets. Don't print money when you do 9:00 spend beyond your etc. And then the third part was trade. Yeah. 9:05 So countries become suppliers of goods before they become demanders. So trade really helps on that one. And the and 9:11 the flip side of trade was of course the flows of foreign direct investment. All of these things there was a 9:18 consensus about them. And this consensus one by one has frayed. 9:24 Uh so this thing about let the private markets take the lead and so on that started to fray once you had the China 9:30 model. You know there was a sense that oh well you know you can act because interest rates are 9:36 so low you can run you can run um big deficits because you can finance them 9:42 very cheaply and that got us into trouble. All right. And that that was mainly a western phenomenon but not just and then the 9:49 third part the trade part and I think this is something that I'm sure you're thinking about is that 9:55 this period of success led to these countries I'm talking about the emerging markets becoming a bigger and bigger 10:02 share of the world uh world economy. Yeah. And at some point it starts it starts to 10:08 matter in the countries that were richer. When these countries are 20% of the world economy, the flip side effects 10:16 are not that great in the richer countries. When they're 40 50% depending on how you measure it, now you're 10:23 talking about huge effects on richer countries too. Yeah. So I think that's what we are seeing 10:28 played out. So as a result of it, trade was of course the first uh uh first casualty. But very but I mean we just 10:37 finished a report on foreign direct investment. foreign direct investment has just tanked. 10:42 Okay? It's gone basically. Okay? And now then what happened and by the way foreign direct investment was 10:48 tanking at the same time that domestic investment was tank. Now why was domestic investment tanking? Domestic investment partly was tanking because 10:55 you know uh once you start to spend beyond your means, once governments 11:00 start to spend beyond the means, you accumulate a lot of debt. Okay? and countries that accumulate a lot of debt 11:06 tend to be lousy places to invest in, you know, not just for domestic investment, foreign investors as well. 11:13 So, you sort of see that happening. Yeah. Anyway, I'm sorry to depress you, but but it's a depressing picture. 11:18 It's fine. And I think part of the uh challenge is to focus on what's the 11:24 overarching goal. And I think if you think of long-term productivity growth 11:30 and you you've laid out so many of the elements at least what was the consensus on how to drive that and challenges to 11:37 that particularly let's just say a kind of statist industrial policy model which 11:43 uh many many countries even say in the around 2015 or so we're having a fairly 11:49 open re-evaluation about the role of industrial policy really compelled by 11:54 the example of China. Yes. And the thing is that they don't necessarily have the deep pockets and 12:00 the ability to absorb uh inefficiencies in the implementation of industrial 12:05 policy like China has. Yes. And so when you think of smaller countries in Africa trying to do that 12:10 and identify particular sectors where they are going to be competitive, there is the risk for example of mistakes. Uh 12:17 can you give us your thoughts on u this shifting away from what was the 12:23 Washington consensus to this kind of other approach to how to drive productivity growth and uh where are we 12:29 on that? I I think we are probably on the side of abandoning the Washington consensus too 12:35 much. Yeah. And we need to go back towards that. Right. I mean, you know, it was never seen, I 12:41 don't think the Washington consensus was ever seen as a full recipe for progress, but it was seen as the necessary 12:48 ingredients on which you could build these other things. Right. So, I think we need to go back to that 12:53 in the sense of, you know, get your fiscal deficits in order. And this is true for pretty much every country, I 13:00 think. Yeah. Right. There are few exceptions that I think and those and those exceptions by the way the uh uh 13:08 those countries that tend to have low public debt ratios and so on are countries that depend a lot on commodity 13:14 exports. Yeah. And commodity and the thing about commodity markets right now is that uh 13:20 they've also entered a phase where commodity uh prices are going to be low and commodity price volatility is going 13:27 to be high. So even for them there is this problem of managing their fiscal policies and so on. It's a different 13:33 dimension but it's an equally tough problem. Yeah. But for these other countries I think 13:39 one of the things is just let's return to the fundamentals of living within your means you know 13:45 right and I think that that's the first thing. The second one is I think that you know 13:51 what we ended up doing was we ended up um forgetting that a big part of 13:56 progress was affordable and reliable energy you know and we went away from that I 14:03 think for a good reason in the sense we were worried about climate change and things like that but I think we ended up 14:08 focusing a lot on uh where the energy was coming from rather than how uh how 14:16 countries uh were using the energy and what you really want is you want to use these energy resources frugally 14:22 efficiently instead we started to focus a lot on you know oh get rid of fossil fuels you have to do this other thing I think that 14:29 balance is coming back too it's a sensible balance and I know it's also related by the way to the other thing 14:34 that you mentioned about AI and data centers and so on because I mean you have very high energy needs for that 14:41 I want to come to that I mean you know we do a lot of work on um technology 14:46 competition techn technology competitiveness here in our shop and one of the things we think about um as you 14:53 think of technology as a driver of innovation and productivity um we think about technology enablers and energy is 15:01 one of the primary technology enablers uh obviously um skilled technical 15:06 workforce is as well um creative uh capital markets there are many elements 15:12 of that and you know that's another way to look at what are those fundamentals Right. And I'm so glad you describe the 15:20 need to maybe for the pendulum to swing back a little bit more because sometimes 15:25 I worry that um in the midst of the rise of populism, which I know you've written 15:31 about and talked about um and the re-evaluation of the role of the state in productive activities, I sometimes 15:38 think we are forgetting the timeless and universal power of markets. Yes. And I'm worried about that. And I I would I 15:44 think that's what you're saying, but I don't want to put words in your mouth. I want you to share with us your thoughts on that. So, you know, I guess this thing about 15:51 populism, right? I mean, uh so populism can rise in a lot of places and it 15:56 really doesn't change the world that much, right? When it when it happens in a big economy 16:02 like the United States, it changes the world, right? Okay. So, I think it's very important to 16:07 see what's the reason for that here. Yeah. And I think I think that when you 16:12 look at that you sort of say all right you know really it has to do with the two biggest economies in the world now 16:19 the US and China and their relations. Okay. And I think one has to look to see whether or not 16:25 these relations uh were sustainable in a sense. And I think the answer clearly is 16:32 no. Yeah. So then the then the next question is why did we get to this point where it 16:39 required such a something that looks like a rupture and I would say that's where the the uh 16:48 that's where that's where international institutions like mine the world bank 16:54 uh the IMF uh the World Trade Organizations I think that uh we didn't see all of 17:02 this coming. Yeah, I don't think we did enough to make sure that things stayed on a stable 17:08 sustainable footing. I'll give you an example. There are these uh three things that really matter a lot. The first one 17:16 is tariffs. The second one is the second one are currency values and then the 17:22 third one are non-tariff measures and so on because what you really want is I 17:27 mean again I think that the free exchange of goods and capital and so on is a very important part of that 17:33 progress that happened. So then you say what actually led to that suddenly coming to um I mean creating so much 17:39 angst it hasn't come to a stop but it's creating a lot of angst and so you say if you look at that you find the first 17:45 one of course was if you have a large economy that actually keeps currency values under undervalued you're going to 17:52 have imbalances right okay and so you saw that uh so as a referring to China 17:57 I'm referring to China as a result China consumes too little and on the US side 18:03 somebody somebody has to consume too much for them to consume too little. So I say US that actually consumes too 18:09 much. Okay. So then the second question is okay. So that was the first one. Now who's supposed to call that out? It's 18:15 supposed to be the IMF and they should call it out. It didn't. And this has been a longstanding issue. 18:22 Not it's not it didn't happen yesterday. It didn't happen. It happened between 2001 and 2009 mostly. But I would say if 18:28 you look at economies now again you know you'll actually start to sort of see maybe there is again this currency 18:35 undervaluation and so on in some economies and as a result of that one has to do something. That's the first 18:40 part. The second part is the part that actually has been a big part of the debates right now which are tariffs. And 18:47 if you look at the tariffs, I think the people just got used to the US allowing 18:52 access to its markets at much lower tariff rates than US producers and 18:58 suppliers had access in their markets. Right? I mean we just looked at this and I can cite you numbers but basically if 19:04 you go down the list of countries the groups of countries from high income economies like Europe and so on down to 19:11 upper middle inome countries like China lower middle- inome countries like India down to low inome countries like u 19:18 Ethiopia and everybody just got used to this world where they said the US would apply much lower tariff levels and so on 19:25 then I mean you can do that for a while but you can't do that forever. Right. Right. And I think we ran out of time 19:31 there. So that's the second part. And I think that was the WTO's job to call it out and so on. And then if you you know 19:38 what it really did require was not so much the WTO calling it out but changing its rules because once a country has 19:44 let's say for example it has u 40% of the world's manufacturing output. It 19:51 should not be considered a developing economy for that reason. Right? When would it be considered developing? 19:56 It gets to 100. You know, of course, at that point, so at that point, a country has to have not a definition that it 20:04 gives itself that it's a developing economy, but a definition that that is a multilaterally agreed one, which says 20:10 this is the category where you are, and as a result of it, these are the kind of additional tariffs, etc., etc., you can 20:15 have and then the third part was the non non-tariff measures and I think that's 20:21 where the world bank actually we were cheerleaders. I was I mean I was I was in the World Bank at this time and we 20:26 were we were cheerleaders for countries like China. Why? There's so many poor people in China and this this model was 20:33 great for poverty reduction. You see yeah sorry explain again that you were 20:40 advocating for reducing non-tariff measures or you didn't advocate enough. We didn't advocate enough. 20:45 Okay. I don't mean you personally but I mean the organization. Yeah. Yes. We didn't advocate enough especially in the services side because 20:51 that's where the US has comparative financial services and others. I mean those sectors tended to be 20:57 closed. Well, you know, let me just so this is such a fascinating perspective especially given who you are and the 21:04 seat that you're in and your research over the years. It's a very rich perspective and I think it resonates 21:11 with uh what I believe now is an emerging consensus on the part of the United States, bipartisan consensus that 21:18 um structural or persistent imbalances are no longer tolerable that coming off 21:25 of the pandemic and coming off of the last 40 years of I would say economic dislocation in the United States as well 21:32 as and you didn't mention this but I'll mention this a recognition of national 21:37 security risks or civil military fusion risks on the part of the PRC. All of 21:43 which contribute to what we are seeing which is the rise of economic security 21:48 policy. Yes. Not just economic policy but economic security policy. Yes. This also has a risk which is that you 21:55 overindex on the role of the state in productive activities where you might 22:00 undercut the drivers of productivity. And so like there is a very like there 22:06 you're trying to thread this needle here. Perspective on that. So my perspective on look I we don't 22:13 work on those I mean strategic issues. We work on international issues rather than like security and things like that. 22:19 Right. But even if you take those those concerns off the table it was not a 22:25 sustainable situation. Yeah that's great. That's great perspective. Yeah. Fantastic. 22:31 Let's just talk about export oriented growth because that was such an like you think about the as East Asian miracle 22:38 you think of Korea you think of number of East Asian countries Japan obviously you think of Malaysia other countries 22:44 export oriented growth winning which assumes certain access to large markets 22:50 was a huge driver of manufacturing growth yes and even I would say 5 10 years ago 22:57 there was a recognition that well is that pathway available to African countries. Is that pathway available to 23:03 India or is there something that one needs to consider around servicesled growth? Um where are we now on that 23:11 debate? Because I I just think it's an important part of the puzzle of what is a robust economic policy for developing 23:18 countries in the current world that we're in. Yeah. Before we get into the sectoral part and and we'll get into the sectoral 23:24 part like the agriculture versus services versus manufacturing. Stay stay 23:29 with me a bit on the macro part. Yeah. Okay. So you said say that I've got a um 23:35 so I have a strategy. I I uh what what I want to do is export growth. 23:40 Right. Right. So what I then do is I you know generally speaking what I end up doing is I end up keeping my currency values 23:48 lower than market. Right. That gives me this big edge. That was the East Asian strategy. Yeah. 23:53 And we never sort of called it as like a violation of international rules. we should have. Okay. 23:59 Now, when it was a small economy like Taiwan, China or when it was South Korea, it was okay. But if you remember, 24:07 uh we've we've seen this movie before at a smaller level. It was Japan, right? 24:12 Okay. Yeah. So, I I think that then what you end up doing is you keep currency values low 24:18 and then you run big current account surpluses. Now, look at the alternative. The alternative is maybe for whatever 24:24 domestic reasons and all you can't keep currency value low right let's take the case of India where I 24:31 mean overvalued currency what you then end up doing is you end up having to 24:36 have tariffs right so then these then there you have these higher tariff walls 24:42 right so as a result of it is one way or the other you end up distorting things 24:47 so you you know we say trade helps a lot and you know free trade helps even more but there's no free trade trade. So as a 24:54 result of it right now that that part need that part needed changing and it is 24:59 changing. Yeah. So one of the things that we always try to tell the countries that we work with in saying look you have to negotiate 25:08 with the country that gave access to its markets on relatively preferential 25:14 terms. You have to negotiate with them in good faith, right? And make things much more reciprocal 25:20 rather than think it's a noisy process. It's a noisy process. But that's what I 25:26 mean would say negotiate in good faith. Try to bring your tariff levels in par 25:31 there. So at least you have that's the first part. And by the way, there are winners and losers with all of this. And so that's 25:38 the political economy of this whole negotiation or set of negotiations, many of which are ongoing right now. Exactly. 25:44 Exactly. And then the other side the point that you mentioned I think the short answer to your to to your question 25:51 is you know when you have such big economies as India and China enter the 25:58 market uh so for example the the growth strategy for India could not possibly be 26:05 replicate what China did right? Okay, maybe for a small economy like Bangladesh, you could do it. But 26:12 for a big economy like India, especially if it wants to be ambitious, it wants to it cannot be the same strategy. By the 26:18 same token, now let's think of this other big economy. Let's call it subsaran Africa or Africa including 26:25 including northern Africa. It cannot be the same as what India did. Okay. It 26:30 can't be services. Yeah. because in India once it decides to get into services or modern services 26:36 and so on there's not much space for anybody else not easily see I'm not 26:42 saying that we are sure about this but what I'm saying is you better be you better keep that as a very likely 26:50 scenario where now you need to think about something else so then you say okay what's left okay of course you know 26:57 services are a very big block so you can sort of cleave out the services the India 27:03 the the uh services in which countries like India have a big advantage take out the others or you get into because many 27:11 of these countries you still have a lot of people working in agriculture. So you then you have to sort of say okay let me 27:17 try to sort of see how I can get huge increases in agricultural productivity without also getting people out of 27:25 agriculture which is which has been the model all along. Yeah, these are not easy questions, but but I 27:31 think this these new technologies that we talking about and so on applied to agriculture do have a potential for some 27:37 of that, you know. Anyway, well, let me this is very conjectural stuff. This is uh fascinating. I I want to jump 27:43 on the technology point and then I want to hit a couple of more questions. you've you've taken us through this really panoramic view growth prospects 27:50 the long-term history the short-term moment that we're in rebalancing and 27:57 really uh I think where we may have gotten our eyes off the ball both the international community but also policy 28:03 makers in developed and developing countries really fascinating let's talk about technology our department is 28:09 focused wholly uh on the economic security dimensions of uh the technology 28:15 revolutions right and So when I think of what we're going through, I often say we're not living through one Sputnik 28:21 moment. We're living through five. AI and chips and quantum that triad, 28:26 biotech, including synthetic biology, clean and climate technologies, and some 28:32 of these bleed into each other and have like an impact on each other. For example, AI and synthetic biology. So, 28:38 and they all sit on a spectrum of special purpose all the way to general purpose technologies. And so when you 28:44 think about this and you know there was a podcast uh our betting on America podcast where Brad Smith spoke 28:50 eloquently about this there are significant potential applications for development and I know 28:57 the bank world bank has been writing a lot about this as well say for example with AI. 29:02 What is really interesting here as these trade negotiations are ongoing, they 29:07 could either be around tariff concessions and maybe some NTBs that need to be reduced or they could also be about 29:14 technology. And the reason I raise this is because um US services 29:21 exports, AI enabled exports, technology exports could be a significant pathway 29:28 to achieving rebalancing. No. Yes. Am I can can you tell me if I'm right there and share your thoughts? 29:34 This is what if you if you talk like this with somebody who's from an 29:39 emerging market economy, either government or private sector, this is music to their ears, 29:44 right? They want huge technology transfers, right? Okay. And they don't have to be 29:50 technocompensated technology transfers. They they could be compensated technology transfers, but they really do 29:56 believe this. Okay. Now once you once you see that then then you start to sort 30:01 of say okay let me look at the other effects of technology let me look at the economic and the social effects of 30:07 technology and the political effects. So one of the things that we are trying to do Naven is that uh we actually have 30:13 our next world development report is on artificial intelligence. Amazing. Yeah. And we are looking at all three of these 30:18 effects. And so that's just so people know that's an annual report that you put out. Yes. It's really one of the it's the flagship 30:25 World Bank report. It's our flagship report. Yeah. And so normally we have like a one-year effort to to do it. In the case of in 30:32 the case of machine learning, artificial intelligence, so on, we found that it wasn't we didn't have we didn't have 30:39 enough evidence and and things to build on yet. So we made it a 30:45 two-year effort. So our team is actually engaged in a lot of research and outreach and so on. We're working with 30:50 people at Stanford and Chicago and Harvard and so on. Then the next year we'll actually put put all of this stuff 30:56 down. What are the implications of all of this for now? But I know that you don't want to wait until next year. So 31:03 so the question is okay now our East Asia region actually re has I mean can't 31:09 wait until next year because this is this is where a lot of these things are happening. So they put out some really I 31:16 work that was really great and I we will link that to to the YouTube post for 31:21 people to take a look at that. That is really quite interesting. Yeah. I mean this whole thing about jobs and technology and so on. One of the 31:27 nice things for example what they found is that is that if you look at not at AI 31:32 but they were looking at robotics and things like that right and they said okay what are the effects of these things? Do they displace labor 31:40 and and the picture is complicated simply because it has two effects. The 31:45 first effect is of course it displaces labor right? Okay. uh but then second effect is if 31:51 you adopt these technologies it increases your efficiency and competitiveness massively. So as a 31:56 result you get a larger share of the market the world market and that increases the demand for labor 32:02 on net they are finding that it actually for at least for these countries it is a plus 32:07 interesting a big plus and along the way it creates pressures to sort of upskill your labor 32:13 too right so you have this very you know strong demand push and so so again the early 32:19 adopters of these technologies might end up gaining and later adopters might not 32:25 be able basically. Yeah, it's really interesting. I I shared um 32:31 on another podcast a chart that looks at um surveys of optimism around AI. 32:38 Yes. Uh it's an Ipsos poll. And the IMF AI preparedness index and what you find is 32:46 many emerging markets are very very high on the optimism scale, obviously low on 32:51 the preparedness scale. Yes. And the way I think about that as as an American is that is the golden 32:57 opportunity for American AI that gap. Yes. So I mean that's that's a normative 33:04 point but I I welcome your thoughts on it. But I as you answer that consider this. 33:10 There's also like a decoupling agenda around the tariffs and that has its 33:17 impact on technology stacks. Which one are you going to buy into as a developing country? Could you just share 33:24 if you do believe there's an opportunity for American AI in developing countries, a tech offer to developing countries? 33:32 What is the nature of the challenge that a developing country faces when they're also faced with an offer from China? 33:40 Well, first one is, okay, as you were talking, I remembered a conversation I had with some business 33:47 people from India. Yeah. And they had a very uh they had a very 33:52 uh uh good example because what we you know we had done this report where we were saying look don't jump to 33:59 innovation first you know from uh from an investment strategy add on what we 34:07 called an infusion element. So this was our world development report on the mill 34:12 income trap last year. Yeah. And we said you know add on this infusion element. This infusion element 34:18 basically is just copy technologies from abroad adapt them you know rather than 34:24 putting a lot of R&D etc into new technology denovo technology and we contrasted strategies of South Korea 34:30 versus Brazil etc and we found one worked really well the other one didn't work at all you know 34:37 now when we discussed this in India they actually gave us some really good 34:42 insights about how the world of AI is changing some of this stuff it may not 34:47 change the entire fundamentals of this but it can change things enough on the margin that it makes a difference and 34:53 the example that they gave were was like uh the Uber eats kind of services and 35:01 basically what they were saying was that if when they looked at this they found that if you don't if you don't need a 35:10 human intervention to one of those transactions if somebody orders food and so on if it's if it's entirely U web 35:17 based um um you save roughly 70 rupees 35:22 right per transaction. Uh if a human intervention is needed you pay 70 rupees 35:28 extra and that 70 rupees really goes to a college educated Indian you know. 35:34 Mhm. Okay. Now what they found was with the pressure of competition they could cut 35:39 that 70 down to 10 or seven by using AI. Yeah. Okay. Now, but they said that that seven 35:47 or 10 doesn't go to an Indian company. It goes to a a company in UK. 35:53 Mhm. Or what you're saying is I want that money to come to a company in the US. 35:59 Well, they're on to that. Yeah. Because they're saying, not only am I losing demand for my college education workers, now I have a net transfer out 36:07 in terms of intellectual property payment. Yeah. So I think I think that I think that uh I still believe I still believe 36:15 that many of these people are actually part of the US technological ecosystem and so on. So as a result, the US can 36:22 actually gain a lot from that. But it requires a little bit more thinking and I think that's why I think 36:27 it's really good that you guys are thinking about it because it's not that simple a problem in the sense of saying, 36:32 you know what, we can't export uh X things anymore. Let's just export technology. Well, not so fast. 36:39 Such a great conversation. I'm going to do some quick hits. I know your time is valuable, but I don't want to lose this 36:44 opportunity. So um let's talk about innovation. For for some time uh US 36:52 economic security policy has assumed that the locus of innovation on advanced technologies is here. Hence we have 36:59 export controls not in all cases but in many cases we have to protect technology advantage. That's the imperative right? 37:07 What happens when technology innovation is located elsewhere like obviously 37:13 China but maybe not only China maybe in other countries maybe in some emerging 37:18 markets are the source of innovation particularly in this day and age um what 37:23 are your thoughts on that so you know I um I'm uh we did some work 37:31 for the Europeans uh some years ago just before the COVID crisis 37:36 about about all of this stuff and we ended up comparing Europe and US and 37:42 China. Yeah. And we said okay which of these uh places is best suit best uh best sort of 37:50 suited for like a big push on AI and sort of thing and uh the 37:58 uh the place where I concluded was it was the US right and the reason was the following. We said okay you know for for you to do 38:05 well on these things you actually need three things. You need big money. Yeah. 38:11 You need business process innovation and yet then you the third thing you you need efficient tax and transfer systems 38:17 because they have huge distribution effects. And what we found was on big money at 38:23 that point China had a big advantage. they're putting a lot of money into it 38:28 on business process innovation US easily you know bringing these the bringing 38:34 these technologies to market is the is like don't bet against the US when it comes 38:41 to that right even if that technology is developed somewhere else it'll be brought to market much quicker here than anywhere 38:47 else right that's uh the third one was uh the tax transfer systems you know how smooth are they and 38:54 so And that Europe actually had an advantage on that. Yeah. So what we were trying to tell Europe 39:00 was don't be so scared about these technologies. You are actually pretty you're actually well equipped 39:05 institutionally or or or administratively to do it. But if you sort of look at all of these 39:10 things you say okay uh who would you give the highest marks to in all of this 39:16 stuff? I' I' I'd give it to the US especially now big money is going in you 39:21 know right with just a little bit of pump priming by the government big money comes in and you have the business process 39:27 innovation these things take a long time to develop and the US already has them tax and transfer systems the US has a 39:34 tolerance for a greater amount of of social mobility in a sense I think you know and I think I think as long as it 39:40 retains that um it it'll actually have it so if I had to bet on thing I would bet on the US. 39:47 Yeah. Yeah. Interesting. Interesting. Very interesting point. Uh one more two more 39:53 quick hits. Yeah. So with the rise of the AI revolution, 39:59 the energy transition, uh obviously critical minerals have 40:04 become the thing. They are an important topic. Uh certainly a point of uh 40:10 concern for people cons who are interested in economic security particularly in the US. Um and so you 40:17 see a number of countries uh that are focusing on this. Uh there is the early 40:22 discussions about the deal in the DRC and Rwanda. There's obviously the discussion around the deal in Ukraine. A 40:28 lot of them focused on critical minerals and uh there were some African heads of state at the White House recently. They 40:35 also touched on this topic. Yes. So great. When I listen to this, I certainly see 40:41 the economic security dimensions to it. I also see uh kind of back to the future 40:48 dimension to this. You know the last 30 40 years there have been times when extractives have taken center stage in 40:54 the development agenda and there were some hard lessons learned from those experiences. Yes. 41:00 In fact this has more than a 200year history in a lot of countries. Yes. But certainly in in our lifetimes. 41:06 How do we approach this now in a way that does not repeat the mistakes of the past and tell us what you think those 41:12 mistakes may have been? Uh so you know on this one you I'm sure Naveen you know a heck of a lot more 41:18 than us. Uh so our expertise in in these kind of issues is really about looking 41:25 at the outlook for commodities markets. So we have a a team that actually the 41:31 same team that does the global economic prospects also does the commodity markets outlook. Okay. And one of the things we are 41:36 trying to have them do is exactly try to answer the question that you're asking. Not from the viewpoint of the US 41:42 necessarily, but from the viewpoint of of emerging markets and developies. And um so I don't think I'm ready to 41:50 give you an answer about this just yet. But but we know what the main problem is. You know the main problem is that we 41:56 are we are worried that I mean we were worried back then because of things like 42:02 um because of the kind of minerals we needed for EVs and solar panels and you 42:07 know those kind of things and we were worried that a large part of those technologies were concentrated in one 42:13 country or in one part of the world. Now you also have rare earths and I think that there are countries that have 42:19 invested in uh u uh they have invested 42:24 in developing these resources and they're the early developers and as a result of it they actually corner they 42:31 have a large part of the market but I think it goes back again to the thing that you mentioned because 42:39 they have invested in developing these resources I there's no reason why the US 42:45 can't invest in developing those whether they are in in the DRC or whether they 42:51 at home. Yeah. And it's really a technology issue. I think I think ultimately just like we 42:57 think that the problem of climate change is going to be solved by technology. I think that we think that the problem of 43:03 rare earths critical minerals and so on will also be solved by technology. You know so I and then again here I'm a big 43:10 believer of the US innovation system. Yeah. Once you present the US private sector a problem, whether it's flight or 43:18 whether it's something else, I think that you get the solution. I think that's fair. Let me just sort of 43:24 probe you on this one because my sense is for the United States and their allies, diversification is more 43:31 important now than ever. You have a concentration of processing on a number of these critical minerals in the PRC. 43:38 And so you don't have to go from 90% of processing in the PR PRC down to 10%. 43:44 Even if you move down to 60 and 70%, there's a degree of diversification that is important. And so that's rocks in the 43:51 ground and that's processing. And I, you know, I want to just throw this at you 43:58 and see what you say. There's an exper there's the experience of Chile. There's also the experience with I believe it's 44:03 forestry in Norway where uh these extractives these extractive industries 44:09 can also become a basis for productivity growth in countries. Yes. They don't have to they don't 44:14 automatically lead to a resource resource curse. Yes. And so my question is for for sort 44:21 of the lay people like myself, what are the ingredients of making sure that those extractives so now critical 44:28 minerals in Rwanda for example or DRC Yeah. become an engine for productivity growth? What would you advise them? 44:35 So you know I think um it's a it's it's it's it's a very good question. So one 44:40 of the things that we did was we've just finished a paper on Indonesia. Yeah. because Indonesia wanted to 44:47 develop its nickel um value added in nickel basically right 44:53 and u and the question was and it's a very reasonable thing to do saying why should we uh export ore when we can 45:02 actually develop that at home of course then you sort of say oh well you need energy for that do you have energy yeah 45:08 they can have energy as long as they use coal and that kind of a thing but but the main point is that you can't argue against keeping some of their value at 45:14 home. Yeah. But then the next question is why why do you want to keep some of that value at 45:20 home? Because it should lead to improved improved economic outcomes. 45:25 Yeah. Once you start to sort of look at that a little bit more closely, you say, "Okay, let me see that what what I'm going to 45:31 be doing is I'm going to be making this cheaper at home." So, I'm going to now take a look at all 45:38 of the industries that use this in their production. So in this case it was steel 45:44 largely. Yeah. And what you find is that once you have cheaper nickel available or cheaper 45:50 inputs available for these there are a lot of firms that would otherwise have 45:55 gone out of business because they're low value added firms now stick around. Mhm. 46:00 So as a result what you then do is that you take this good thing and you make it 46:06 a bad thing by lowering productivity levels across a huge sector. M so my own sense of it is that's the 46:13 modern resource curse that's you have fascinating that that's a modern resource now now of 46:20 course you have other countries where just these natural resources are so big and so on that they create what's called 46:28 a veracity effect and so on right I mean they weaken institutions I don't think the resource curse is 46:34 something that should worry Canada and the US for the same reason as the 46:39 resource caucus was not an issue for a country like Norway because I'm really thinking about developing 46:45 countries where we could be investors. Absolutely. Or u even vendors for those sorts of 46:50 things. Absolutely. Absolutely. And I think of these things as 100red-year relationships and you want 46:56 those relationships to be robust and benefit the population and benefit both 47:01 sides uh of the Atlantic to put it that way. So two two observations on 47:08 um uh two observations on that. If you look at uh this question about to what 47:15 extent has Chinese investment helped or hurt countries in Africa? Yes, that's a good question. Yeah, 47:20 good question. I think a very simple but not misleading answer is for uh uh based 47:29 on the analysis that that I've seen if a country had strong strong institutions 47:35 good governance basically it gained a lot from this if a country had weak governance it lost 47:42 right so then the question is for an institution like the world bank and like 47:47 what you just said if you want a sustainable relationship that leads to mutual gains. 47:52 Yeah, it can only happen if you also make sure that these countries develop their institutions. 47:58 That's a great point. I'll tell you why I think this is a great point. I think that for the United States, as we think 48:05 through what the future of foreign assistance looks like, let's not throw the baby out with the bathwater. There 48:12 is a modocum level of investment in institutions that are needed. So the ROI 48:17 of those investments including private investments are meaningful um and sustainable. So 48:23 amen what I would say to that. Great great point. Okay, I'm going to bring it home. We started out with a 48:28 sobering picture of where we are point in time and then you've taken us through and there opportunities and I think 48:34 you've like I'm excited the more you talk about it like there are still possibilities and opportunities. I think 48:41 I believe in being a hyperrealist and so I was excited to read that next week I 48:46 believe um you have a conference the annual bank conference for development 48:51 economics is that right that's exactly ABCDE for those who don't know and what I found interesting is your focus is on 48:59 u economic reform in the age of populism fascinating meaning to put it in my 49:05 words what is the future of economic reform in a world um where populism is 49:11 on the rise in developed and developing markets and maybe I would just sharpen this a little bit and ask you should we 49:18 be moving towards development policy including by the bank and others that 49:23 are robust to the political economy considerations of developed countries 49:29 particularly those that have been perceived losers from trade over the last several years you uh you're 49:36 absolutely right and uh I think I would encourage encourage uh so I would 49:41 encourage you and everybody else to actually participate in this conference. You can do it virtually, you can come, 49:47 we'll be happy to host you. I think that the short answer is that um that you 49:53 know um the short answer to your last question is that there is no way that 49:59 you can ignore political economy issues in the more advanced economies because 50:06 they may be more advanced but they are much smaller fraction of the world economy or they're not a dominant part 50:12 of the world economy anymore. They share this roughly 50/50 with the others. Mhm. 50:19 So how can you how can you ignore but the other thing that you sort of see 50:24 this no is if you really look I mean and I'm a firm believer of this if you look 50:29 at the period of prosperity that that the world economy went through it was a period in which the American 50:36 template was applied in a lot of countries one a reliance on markets not 50:43 just say okay markets are sufficient are both necessary and sufficient you said no no no you need these other 50:49 institutions like for example mass education which which I think was also I 50:55 mean it may not have been an American invention but it was definitely an American story. Yeah. Right. 51:01 And a lot of these countries did all that stuff and as a result they prospered. So when the US model needs 51:08 adjusting and so on, it should adjust because then I have a feeling that that 51:14 same template will be applied, right? And if if it and if this happens, you 51:20 all all the countries that are low-income countries will become middle- inome countries. Many middle- inome countries become high income countries 51:26 and it'll be it'll be a very different world. It'll be a much more prosperous world. It'll still be an American world. 51:33 Yeah. Yeah. Uh amen. So let me let me now bring this finally home. You have been a 51:40 world banker I think a couple of times. You've been a professor, a scholar, a researcher. I've been a big fan of what 51:46 you've done over the years. Um if you look to the next generation of economists, 51:52 you know, take take it back 40 years or whenever when you started. uh if you 51:57 look at the the generation that's coming up now, how has the field of economics 52:04 changed? What are the things that you would point to as um sort of future 52:11 directions for the field that are exciting and important and uh hopefully influential for uh tomorrow's 52:17 economists, the tomorrow's uh in their midst of tomorrow. Yeah. So, so I I think because of the 52:25 way that I was trained, I was trained by people like Gary Becker and Bob Lucas 52:30 and Chauvin Rosen. I believe that I believe that uh you have to you have to 52:37 u make economics uh work for the common good. Mhm. 52:43 You know, uh I mean if you look for example at the American model too, it 52:48 was a you know it wasn't like market based systems hadn't been tried earlier. 52:54 I mean they were all market based systems in the medieval ages etc etc but 52:59 they were not applied for the common good. M so that's the point I I say believe in 53:04 markets that's one thing for sure but make find ways to sort of make them work 53:10 for the common good right and you know one of the things that you learn of course then is that a a huge a 53:17 hugely important thing for this is competition right but you can't leave competition just to 53:22 capitalists I mean you know because people who uh who compete want to end 53:27 competition they don't want to prolong competition so That's one thing. So I 53:32 think that this whole I think that this whole No, that's Say that again. That's really important. You can't leave competition 53:38 just to capitalists. Yes. Because uh I mean um uh folks who 53:45 you make uh folks who have to compete want to end competition. 53:51 They don't want to prolong it. Right. Yeah. So somebody has to prolong competition. And I think that's the role of that's 53:58 that's the role of the state and that's a huge. Now then if you sort of look at it you say all right let's go back and 54:05 say what is it that thing I think we I would return to the fundamentals. I 54:10 would sort of say you still need uh you still need macro stability. Okay. So I 54:17 think I think that the world has to deal with the debt crisis especially the debt crisis in poorer countries because the 54:24 richer countries can you know they have the mechanisms to actually give people a second chance right but I think the world has to come up 54:31 with mechanisms to give poor countries a second chance you know uh um and then of course I think trade 54:38 uh you know trade has always been a great thing but it cannot be a trade which is imizing for one big swath of 54:47 society, even if that swath of society is in a very rich country and the rest of the country is really benefiting from 54:52 it. I think that's what we've learned. I don't see populism as a bad thing. By the way, 54:57 just like elitism, you can't say elitism was a good thing and populism is a bad 55:03 thing, you know. Uh so I think in some sense this is a correction because I think the elites 55:09 blew it. I mean I mean you could consider us as part of the elites or what but I think we we miss these big uh 55:17 big risks. Yeah. But we have a chance to fix it. So and I I don't know if I'll be able to 55:22 do it in my time but you're much younger. You should do it. Not that much. Yeah. 55:27 But I think you've uh laid out a wonderful vision for uh folks who are starting their careers 55:34 as economists. um it's given you you've laid out a purpose and I think like that's timeless 55:41 and universal but I should say particularly urgent in the moment that we're in. So with that I really thank 55:48 you for taking the time. You've been generous with your time. Such an interesting conversation. Thank you. It's been a pleasure. 55:55 Thank you for joining this insightful conversation with Indil World Bank chief economist and senior vice president. You 56:01 can find this on YouTube or csis.org. This is Naven Girish Shankar signing 56:07 off. Thank you. [Musi