Generative AI Could Create Labor Growing Pains, Even in White Collar Sector

This podcast is brought to you by the Wharton School at the University of Pennsylvania.

这个播客由宾夕法尼亚大学沃顿商学院带给您。

Well, as artificial intelligence works more and more into our culture, the expectation of the level of change coming our way is being studied closely. But there will be change, no doubt about it.

随着人工智能在我们的文化中发挥越来越大的作用，人们对即将到来的变革水平的期望正在被密切研究。毫无疑问，将会有变化发生。

And our next guest suggests that labor will see some growing pains, even in the white collar sector before everything gets worked out. David Steinberg is co-founder and CEO of Zeta Global. David, pleasure to have you with us. Thanks for your time today.

我们的下一位嘉宾认为，即使在白领行业，劳动力也会经历一些不适应期，在一切问题解决之前。David Steinberg 是 Zeta Global 的联合创始人兼首席执行官。David，很高兴有你的加入。感谢你今天抽出时间和我们交流。

Thanks, Dan. I listened to you all the time. So it's great to be here.

谢谢，丹。我一直在听你的。所以很高兴能在这里。

Well, thank you. Appreciate that. Alright.

好的，谢谢。非常感谢。好的。

So the question around labor and AI, I think that's been maybe one of the most talked about components of this process really over the last few years, not even just the last few months.

所以，关于劳动力和人工智能的问题，我觉得这可能是过去几年里甚至不仅仅是过去几个月里最被议论的组成部分之一。

So how do you see the labor force being impacted by AI?

那么你如何看待人工智能对劳动力的影响呢？

Well, I mean, that's obviously a pretty loaded question, right? Because you've got multiple iterations of it. But as we think about gender to AI and moving into large language models, which actually this will sound counterintuitive, but at Zeta, we believe we're going to move into small language models.

嗯，我的意思是，那显然是一个相当复杂的问题，对吧？因为它有多个层次的迭代。但是当我们考虑到性别和人工智能以及大型语言模型时，这可能听起来有些违反直觉，但在 Zeta，我们相信我们将会转向小型语言模型。

But what you're going to see in our opinion is a massive disruption in the white-collar world, right? So if you think about technology in its evolution, so the ability to move information in real time is what drove a large offshoring of labor, right? You were able to move call centers overseas, programming, engineering was able to move overseas. You were able to move manufacturing.

根据我们的观点，你将会看到白领世界发生巨大的颠覆，对吗？所以如果你考虑技术的演化，实时信息传递能力导致大规模的劳动力离岸化。例如，呼叫中心、编程、工程等都可以被迁移到海外。甚至制造业也可以被迁移。

I always joke, right? Like 100, 200 years ago, if you had a factory in India, you would want to, you'd have to write a letter that could take months to get to the factory manager on a change to manufacturing process. And there might be a problem with what you've told him to do. You'd have to write you a letter back. It would take months. That information moves freely.

我经常开玩笑，对吧？比如100、200年前，如果你在印度有一家工厂，你会希望，也必须写封信给工厂经理，告诉他修改制造工艺的变动。这封信可能要花上几个月时间才能送到。而且你告诉他的事情可能会出现问题，他会再给你写信回复。这个过程需要花费几个月时间。现在，信息传递自由了。

And that really affected U.S. labor and highly developed industrialized nations labor by allowing that to move to lower cost labor markets.

这对美国劳动力和高度发达的工业化国家的劳动力产生了很大影响，因为它允许他们转移到成本更低的劳动力市场。

What that did to that ecosystem, I believe AI is going to do to the law profession, to the accounting profession, where creativity and, quote, making it rain will continue to be at a premium. But, you know, drafting contracts, doing tax returns, setting up a trust, all of those mundane features can easily be handled by generative AI.

我相信，人工智能将对法律和会计行业产生与之前对生态系统所产生的影响相似的效果。创造力和“取得成功”的能力仍然非常重要。但是，起草合同、编制税务申报表、建立信托基金等琐碎的工作可以轻松地由生成式人工智能来处理。

So then that would lead me to believe that we will see a level of job loss because obviously you won't need to have certain people in functions that traditionally they have been there over the last several decades.

那么，我认为这将导致一定程度的失业，因为显然你不再需要在传统上长期存在的某些职能中雇佣特定的人。

But it probably also kind of leads to kind of a shift in terms of the types of jobs that will be available and the types of jobs that people will, you know, study for in college or prepare for, you know, when they get into the workforce, that type of thing.

这可能还会导致就业类型发生一种变化，人们可能会在大学里选择不同类型的工作，或者在进入职场前做好不同类型工作的准备。

Yeah, well, there was a great Goldman Sachs study that said 300 million jobs are going to be disintermediated by AI. But the truth of the matter is there's two things happening that are going to offset that, right?

是的，有一项由高盛公司进行的研究指出，未来将有3亿个工作岗位因人工智能而被淘汰。但实际情况是，有两个因素在发生，将会抵消这种情况，对吧？

So if you look back to the what's happened in the past, I always joke that, you know, you can look back and everything you think is the first time ever often has happened before in different ways, right? So you had the Gutenberg printing press, then you had the telephone, you had the computer, and then of course the internet, and all four of those technological developments, the world was talking about how employment as we understood it would end. Nobody would ever work again.

所以，如果你回顾过去发生的事情，我常常开玩笑说，你知道的，当你回头看，你认为第一次发生的事情通常之前以不同的方式发生过，对吧？所以你有古腾堡印刷机，然后有了电话，有了计算机，当然还有互联网。而这四项科技进步，世界上都在讨论我们所理解的就业会结束。没有人会再工作了。

But at the same time, substantially more jobs were created by that technology than they destroyed, but you're exactly right. It's different jobs.

然而与此同时，这项技术创造了比摧毁的工作岗位数量多得多的就业机会，但你说得完全正确。不过，这些是不同的工作岗位。

So we might see the creation of hundreds of millions of jobs around AI programming and, you know, AI proctors and all of that type of stuff.

因此，我们可能会看到 AI 编程、AI 监考等大量就业机会的出现，你知道的，以及所有这些相关的工作。

The other thing we're dealing with from a humanity perspective is we've got a massively aging population globally. We don't have to look past Japan where the United States or now China. So look at what a large percentage of the workforce is aging out. We can't replace that at our current birth rates.

从人类的角度来看，我们还要面对一个全球大规模的人口老龄化问题。我们只需看看日本，或现在的美国和中国。看看劳动力中有多大比例的人正在变老。我们无法用我们现有的出生率来替代他们。

So AI will be able to pick up a lot of the slack for people who are just aging out of the workforce.

因此，人工智能将能够弥补工作年龄结束的人们所失去的许多工作机会。

And I think you're going to see some amazing new fields that are going to be created on top of artificial intelligence.

我认为你会看到一些惊人的新领域将在人工智能的基础上崭露头角。

Like what specifically?

具体是指什么？

Well, I mean, you could look at an AI proctor. That might sound like an incredibly stupid thing. But it's going to take a very long time for artificial intelligence to really become what I would call creative. Right?

嗯，我的意思是，你可以考虑使用一个AI监考员。可能听起来像是非常愚蠢的事情。但是要让人工智能真正成为我所谓的创造性的东西，需要很长时间。对吧？

So when you look at the publishing industry, right, this is going to hit it. This is going to hit people who write articles, right? This is going to hit people who are editors of articles.

当我们看向出版业的时候，对于写文章的人来说，这会产生冲击。同时，文章编辑也会受到影响。

But at the same time, the computer doesn't wake up in the morning and say, I want to write an article about kinkardashin today. Right? Somebody has to know how to properly prompt that algorithm to come up with an article that will be a topic that people will want to read and timely. Right? That's going to be a whole new type of profession if you look at the legal field. Right? Even if you're using AI to generate a contract, you're going to need to put the terms and conditions for that contract in to the algorithm to get it to work. And the funny thing is you're going to need really smart people to do that type of stuff so we can get the contract right on the second or third try. It's not going to do it on the first. But you don't want it taking 100 tries. That sort of defeats the purpose. You know what I mean? Right.

但与此同时，电脑不会在早上醒来说，今天我要写一篇关于金·卡戴珊的文章。对吧？有人必须知道如何正确地提示那个算法，才能生成能够吸引读者的及时的文章。对吧？如果你看看法律领域，这将成为一种全新的职业。对吧？即使你使用人工智能来生成合同，你也需要将合同的条款和条件输入算法才能让它工作。有趣的是，你会需要非常聪明的人来做这种工作，这样我们才能在第二次或第三次尝试时得到正确的合同。它不会在第一次尝试时就做到这一点。但你也不希望它尝试100次。这有点违背了初衷。你明白我的意思吗？对的。

But unfortunately, the expectation is by a lot of people that they wanted on the first try. Yes. And so that dynamic of getting it right in the quickest fashion is still something that needs to be addressed. And obviously that could be where the human component plays in. 100% agree with you again. Like, look at, I mean, OpenAI has the first mover in this sort of what I'm calling or a lot of people are now calling, the golden age of AI. Right? It's data. We've been doing this for 15 years. We started out calling it automation, then deep learning, then machine learning. Now it's now it went from AI to LLM or, you know, different type of models feeding into generative AI. And you look at sort of how this works, you have to really level set expectations. So back to OpenAI, their entire large language model is based on the internet pre-October of the year 2021. So if you want anything topical as, you know, what happened yesterday, it's very, very difficult to get that out of the current OpenAI large language model. Most people don't even realize that. Right?

但不幸的是，许多人期望他们能在第一次尝试时实现这一目标。是的，因此，在最短时间内做到准确无误仍然是需要解决的问题。很明显，这可能是人类因素发挥作用的地方。我完全同意你的观点。比如说，看看OpenAI，他们是在所谓的AI黄金时代中迈出的第一步。这是数据。我们已经做了15年。我们最开始称之为自动化，然后是深度学习，接着是机器学习。现在从AI变成了LLM，或者说是不同类型的模型输入到生成型AI中。你看看这个过程，你必须真正设定好期望。再说回OpenAI，他们的整个大型语言模型都是基于2021年10月之前的互联网数据。所以如果你想要关于昨天发生的事的任何时事信息，从现有的OpenAI大型语言模型中提取这些信息是非常非常困难的。大多数人甚至都没有意识到这一点。

At the same time, BARD is now trying to do it in real time. That's of course Google's platform. It'll be very interesting to see how that evolves. I always joke there's not a lot of first movers who ever really win. Most of the most of the big winners have always been the second or third movers. It'll be interesting to see how this plays out. I have brought this up, this question next, cushion up a couple of times and get your thoughts on it as well, of the overall impact that AI will have on the education process, on learning as you're looking towards individuals taking these positions in the decades ahead. Yeah.

同时，BARD现在正试图实时地做到这一点。当然，这是谷歌的平台。看到它的发展将会非常有趣。我经常开玩笑说，很少有先行者真正能获得成功。大部分大赢家总是第二或第三才行动的人。看看这个将如何发展会很有趣。我也好几次提起过这个问题，并想听听你对AI对教育过程和未来几十年里个体学习将产生的总体影响的看法。是的。

So listen, I first of all, I'm sort of of the school of thought. We're still educating the vast majority of people in the United States for the Industrial Revolution that we exited in the 1950s when we became the sort of information technology society we are today. And we're moving towards the artificial intelligence society of tomorrow. We have to retool our educational system to start to focus on where tech is going. Now, there's this whole narrative that AI is going to ruin learning. I don't know how old you are. I'm 53 years old, right? So I remember when I couldn't bring an HP calculator to school. Like I remember when that was a problem and when my kids started going to school, they were bringing a computer with them. So I think it's going to change the way we think about education.

所以听着，首先，我属于那种思维方式。我们仍然在美国教育绝大多数人工业革命的知识，而我们在上世纪50年代退出了工业革命，成为了我们今天这种信息技术社会。而我们正在迈向明天的人工智能社会。我们必须重新调整我们的教育系统，开始关注科技的发展方向。现在，有这种完全消极的说法，认为人工智能会毁掉学习。我不知道你多大年纪，我53岁，对吧？所以我记得以前上学不能带HP计算器。我记得当时这是个问题，而当我的孩子们开始上学时，他们带着电脑去学校。所以我认为它会改变我们对教育的思考方式。

I think societies, like you look at South Korea, they're already teaching classes on this. They're already moving to that. Like what is the next phase of education going to look like? It's going to be more of a supportive stage of how do we work with machine learning and how do we work with gender-to-day AI versus how do we subjugate it?

我认为像韩国这样的社会已经开始教授这方面的课程。他们已经在朝这个方向发展。下一个阶段的教育将更加强调与机器学习以及与当今性别相关的人工智能的合作，而不是压制它。

One of my favorite stories, I heard it recently, I was over at Viva Tech in Paris and the Minister of Artificial Intelligence from Dubai was giving a talk. I mean, I was first shocked that Dubai had a minister of artificial intelligence and he just went on to say that he'd been doing it seven years. Like it was really impressive. And his whole talk was about the Gutenberg printing press, which I have a level of fascination with as a technologist. And he said when the Gutenberg printing press came out, when it was first invented, the Middle East was by far the largest economy in the world. This was him giving the talk. I never really validated whether it was Middle East or China, but it certainly wasn't Europe. Six individuals who were ruling countries in the Middle East got together and they made the Gutenberg printing press illegal. According to him, for 182 years, the Gutenberg printing press was illegal in the Middle East. And that 182 years, the European economy, which embraced that technology, tripled in size and the Middle Eastern economy shrank. So we have to learn from that and understand that we need to embrace this technology to help us from a GDP perspective, to help us from a growth perspective. And that all starts as I think you've surmised with your question. It all starts with education.

我最喜欢的故事之一，最近听到了它，我当时在巴黎的Viva Tech会议上，迪拜的人工智能部长发表了一场演讲。一开始我很震惊迪拜竟然有一个人工智能部长，他接着说自己从事这个工作已经有七年了。这真的令人印象深刻。他整个演讲都是关于古腾堡印刷机的，作为一个技术专家，我对它有一定的着迷。他说古腾堡印刷机问世时，中东地区无疑是世界上最大的经济体。这是他演讲时说的。我从来没有真正核实过是中东还是中国，但肯定不是欧洲。六位统治中东国家的人在一起，将古腾堡印刷机定为非法。据他说，古腾堡印刷机在中东被禁止使用了182年。在这182年中，欧洲经济（采用了这项技术）扩大了两倍，而中东经济则萎缩了。因此我们必须从中吸取教训，明白我们需要拥抱这项技术，从GDP的角度帮助我们，帮助我们实现增长。我认为你的问题已经揭示了这一切的起源，那就是教育。

David, great to have you with us. Thanks for a few moments. All the best we will stay in touch. Thank you very much for having me. I really appreciate it, Dan. You got it. David Steinberg, co-founder and CEO of Zeta Global. Hello, today, acoustic tech company.

大卫，很高兴能与你在一起。谢谢你抽出一些时间。祝你一切顺利，我们将保持联系。非常感谢你邀请我。我真的很感激，丹。没问题。大卫·斯坦伯格，Zeta Global的联合创始人兼首席执行官。今天你好，声学技术公司。