全球世代：驾驭变革中的世界，第1集

讨论让世界持续运转。

Discussion keeps the world turning.

这就是RoundTable。

Is RoundTable.

从北京的心脏到全球舞台的边缘，你就在RoundTable。

From the heart of Beijing to the edges of the global stage, you are at RoundTable.

我是林业火。

I'm Yeoho Lin.

在当今世界，年轻人不仅是全球变革的观察者，更是塑造变革的参与者。

In today's world, young people are not just observers of global change, they are participants in shaping it.

为了在国际重大事件中凸显他们的声音，RoundTable很高兴推出《世代全球：应对变化的世界》这一新系列，致力于这一紧迫主题。

To highlight their voices in international milestones, RoundTable is happy to present Generation Global navigating a changing world, a new series dedicated to this urgent premise.

我们在全国人民代表大会上推出了这一系列，中国全国人大第四次会议于3月5日在北京人民大会堂开幕。

We launched this series at the National People's Congress, China's national legislature opens its fourth session at the Great Hall Of The People in Beijing on March 5.

随着中国两会全面展开，国家步入第十五个五年规划的首年，人工智能依然是这一新篇章的核心主题。

As China's two sessions gear up on full swing and the country steps into the first year of its fifteenth five year plan, artificial intelligence keeps being one of the defining themes of this new chapter.

中国在人工智能领域的快速发展正吸引着全球关注，对于在华学习和工作的外国青年来说，这既令人兴奋又充满复杂性。

China's rapid AI progress is attracting global attention and for foreign youth studying and working in China, it is both exciting and complex.

今天的对话参与者包括：在北大研究人工智能与伦理的学者兼学生Ace Asim，亲自报道两会的Yushun，以及我们自己的Steve Hatherley。

Today's conversation is between Ace Asim, a researcher and student focusing on AI and ethics at Peking University, Yushun, who's covering the two sessions himself, and our very own Steve Hatherley.

现在请坐好，一起来聆听他们的对话。

Now pull up a chair and enjoy their conversation.

欢迎来到《世代全球：应对变革中的世界》。

Welcome to Generation Global navigating a changing world.

这是一个特别系列，聚焦来自中国及全球各地的年轻声音。

This is a special series featuring young voices from China and around the globe.

我是来自RoundTable的Steve Hatherley。

I'm Steve Hatherley from RoundTable.

我也和来自RoundTable的Yushun在一起。

I'm with Yushun from RoundTable as well.

下午好。

Good afternoon.

你最近怎么样？

How are doing?

你好。

Hi.

我很好。

I'm good.

这个环境真不错。

And this is a good setup.

这环境太棒了。

It's a great setup.

我喜欢我们的演播室，也喜欢我们今天的话题。

I love our studio and I love our topic today.

而人工智能这个话题，你不仅最熟悉，而且也最热爱。

And this AI is a topic that you are number one familiar with, and number two, you love.

当然。

Definitely.

我们已经多次讨论过人工智能这个话题了。

And we've been talking about the topics of AI for so many times.

当然，在这个人工智能飞速发展的时代，我们每次谈到这个话题都必须提及伦理问题，因为几乎每周我们都会被震惊一次。

And of course, ethics is the points that we need to mention when we are in this era that AI is developing so fast that we can get shocked basically every week.

就在我们说话的此刻，人工智能正处在一个关键的十字路口。

We're at a critical juncture with AI as we speak.

今天，我们邀请到了北京大学的研究员兼学生艾斯·阿西姆。

Well, today we're joined by Ace Asim, a researcher and student at Peking University.

这是一个有趣的故事。

An interesting story.

他原本来自新泽西。

He's originally from New Jersey.

现在他常驻北京，为当今最紧迫的对话之一带来了跨大陆的视角。

He now bases himself in Beijing and he brings across continental perspective to one of the most urgent conversations of our time.

人工智能如何重塑各行各业和不同文化中的社会。

How artificial intelligence is reshaping societies across industries and cultures.

他的研究位于人工智能、伦理、政策与文化视角的交汇点，探讨新兴技术如何塑造我们的未来、影响公平性，并反映不同社会的价值观。

His work sits at the intersection of AI and ethics and policy and cultural perspective exploring how emerging technologies shape our future, influence equity, and reflect different society's values.

从算法偏见和媒体表达到AI工具如何重塑跨文化的叙事、创造力与获取方式。

From algorithmic bias and media representation to how AI tools are reshaping storytelling, creativity, and access across cultures.

艾斯确实提供了全球视角，这种视角源于他的亲身生活经历——不仅来自美国，还生活在中国，并在各大洲之间学习。

Ace definitely offers a global and shaped by living, not only being from The United States, but living here in China and studying across continents too.

另外，玉顺，我们与艾斯的对话恰好与中国两会同期举行。

Also, Yushun, our our conversation with Ace happens to coincide with China's two sessions.

当然，这是每年国家规划政策方向的重要会议。

That, of course, is the annual meetings where the nation charts its policy direction.

今年的讨论尤为重要，因为人工智能治理和新兴技术的伦理边界，已成为中国在创新与责任之间寻求平衡时对话的核心内容。

And this year, it's an important discussion because AI governance and the ethical boundaries of emerging technologies are very much a part of the conversation as China balances innovation and responsibility.

你知道，我们生活在一个AI正在迅速改变我们创造、消费和连接方式的世界中。

You know, we're living in a world, of course, we know where AI is transforming rapidly how we create and consume and connect.

我们即将与艾斯进行的这场对话，将帮助我们不仅理解AI能做什么，更理解它应该做什么。

Conversations like the one that we're just about to have with Ace will help us understand not just what AI can do, but what it should be doing.

Ace，下午好，非常感谢您今天加入我们。

Ace, good afternoon, and thank you so much for being with us today.

下午好。

Good afternoon.

非常感谢史蒂夫和玉顺邀请我。

Thank you so much, Steve and Yushun, for having me.

我非常期待和你们交谈。

I'm very excited to talk with you guys.

你从新泽西搬到北京已经有一段时间了。

So it's been a little while since you've relocated from New Jersey to Beijing.

对吗？

Correct?

已经过去几年了吧？

It's been a couple of years?

是的。

Yeah.

我第一次来中国是在2024年1月。

It's been so I first got to China in January 2024.

我之前在上海和苏州一带生活了大约两年，直到今年秋天，我来到了北京的北京大学。

I lived around Shanghai, around Suzhou, Shanghai area for about two years until, you know, this past fall where I am now at Peking University in Beijing.

从美国来到中国，再从上海搬到北京，这确实带来了很大的节奏变化，因为这两座城市各有其独特的魅力。

So definitely a change of pace with regards to going from The US to China, but also from Shanghai to Beijing, two very beautifully distinct cities.

我想你从中国南方来到北京的经历会和我相似，因为我实际上就出生并成长在中国南方那个区域。

Then I think you will have the similar experience of traveling from that southern part to Beijing as I did because I am actually born and raised in that area that that southern part of China area.

离上海很远。

Far from Shanghai.

对吧？

Right?

是的。

Yes.

离那里真的很近。

It's really close to there.

你现在住在北京，体验这里的气候等各种情况，感觉怎么样？

And how are you like feeling right now living in Beijing and experience all the climate and everything?

是的。

Yeah.

我的意思是，没错。

I mean, I would say yeah.

就在今天，我上中文课的时候，老师问我对北京天气的看法。

Just today, actually, I was in Chinese class and my Lao Shi asked me what what I think of Beijing's weather.

我第一个想到的就是‘干’。

And the first thing that comes to mind is Gan.

这里非常干燥。

It's it's incredibly dry.

所以，当我还在南方的时候，那里很潮湿，但这里冬天简直太残酷了。

So, you know, when I was in the South, it was humid and but here, it's just the the winter is just brutal.

又冷、又刮风，还很干燥。

It's cold, it's windy, and it's dry.

所以我得不断保湿，涂润唇膏，不然我可能会……不知道，直接干没了。

So I have to keep moisturizing myself and putting Chapstick on or else I'm gonna, I don't know, you know, disappear.

我笑了，因为这正是我初到北京过第一个冬天时的经历。

I'm laughing because that's exactly the same experience I went through with my first winter in Beijing.

好吧。

Alright.

那我们深入聊聊吧，Ace。

Well, let's dive into it, Ace.

我们先谈谈在美国生活的经历吧，毕竟你来自那里，现在又在中国学习和生活，这和人工智能有什么关联？

Let's talk first perhaps about the experience of living in America, of course, being from there and now studying and and being in China and how that relates to AI.

到目前为止，你的经历，尤其是你最近最熟悉的这两种文化，是如何帮助你理解人工智能在社会中的角色的？

Your experiences thus far, especially dealing with the two cultures that you're most familiar with most recently, how has that helped to shape your understanding of AI's role in society?

是的。

Yeah.

就像你说的，我是在美国新泽西州长大的，就在费城外面。

So, like you said, I grew up in The States, in New Jersey, right outside of Philadelphia.

我直到上大学之前都是在美国接受教育的，但也曾短暂地去西班牙和巴塞罗那留学。

I have been, you know up until university, was educated in The States, but I also had some brief study abroad in Spain and Barcelona.

所以我会说，中国的生活与我在美国的生活截然不同，这里存在语言障碍，而一旦你开始跨越或甚至只是接触这个语言障碍，方言的问题就会浮现。

And so I will say that life in China has been, like, the furthest from furthest from my life in The US, where, you know, there's a language barrier and, you know, once you cross or begin to even cross that language barrier, dialects come into play.

你知道，我在美国会开车，但在这里我主要依赖公共交通。

You know, I drive in The States, but here I rely on public transportation.

在美国，我们有句俗话，叫‘现金为王’。

And, you know, in in The States, we have a saying, you know, cash is king.

而在这里，我们生活在一个无现金、完全依赖手机的社会。

And here, we live in a cashless, you know, phone based society.

所以这里有很多与美国不同的地方。

So there's so much that's different here.

因此，如果我从这些经历中对人工智能和社会的关系有什么领悟，那就是人工智能的成功取决于它能否适应本地环境并保持敏感性。

So if there's really anything that I have taken away from that with regards to AI and society, it's that AI's success depends on its ability to adapt to local context and be sensitive.

在某些方面，可能并不存在一种放之四海而皆准的解决方案。

And where there may not necessarily be such a functional or there's there's, you know, one one size fits all solution may not be the best way to go in some aspects.

你知道，有些技术会让人感到强大，而在其他情况下却让人感到疏离。

You know, some technology can feel empowering while alien in others.

因此，有很多差异和细微之处需要被考虑到。

So there are a lot of differences and a lot of nuances that need to be taken into account.

你知道，我亲身经历过一些文化差异。

You know, I've had my experiences with just cultural differences.

有时候，我说中文时会误解别人的意思。

You know, sometimes I misunderstood when I speak Chinese.

现在，这很可能只是因为我的中文不太好，发音有问题，但确实也有过既面临语言障碍、又遭遇文化差异的时候。

Now, could it just very well be because my Chinese is not that good with my pronunciation, but, you know, there are times where I've dealt with language barrier, but also just cultural difference.

所以通过亲身体验，我意识到，如果我们即将拥有像AI这样渗透到生活方方面面的技术，它就应该更具适应性，并能敏锐地感知其运作的环境。

So experiencing it firsthand, I realized that if we're gonna have something as, you know, permeating as AI, dealing with almost every aspect of life, it ought to be more adaptable and sensitive to the context within which it it operates.

你刚才说的某句话让我豁然开朗，我以前甚至没想过AI要适应社会这个问题。

Do you think that you you said something that kind of gave me a light bulb moment that I I don't think I've even thought about before about AI adapting to society.

但当你这么说的时候，我的第一反应是：我觉得反而是社会在适应AI。

But my first reaction when you said that was, I feel like society is adapting to AI.

技术正在不断涌现。

The technology is coming out.

技术进步发展得非常迅速。

The advancements are coming out so quickly.

是的。

Yeah.

我一直在努力弄清楚，该怎么使用手机上最新的应用或者AI的最新更新。

I'm constantly trying to figure out, you know, how to use the latest app on my phone or the latest update in AI.

你觉得这方面还有讨论的余地吗？

Do you think there's any debate to be had there?

到底是谁在适应谁？

What what is adapting to what?

我完全相信这是一个循环。

I fully believe that it's a circle.

这是一个相互持续影响的循环。

It's a cycle where they are constantly influencing each other.

它们有能力做到这一点，我们看到，是的，我们的生活现在很大程度上被AI所主导，我们正在适应AI。

And they have the ability to do that, where we see, yes, we live our lives that are very much, you know, now dictated by, you know, we are adjusting to AI.

而这正是所谓的AI素养，我们可能在接下来的讨论中会深入探讨这个概念。

And that's, you know, that's a that's a, you know, it's its own term of AI literacy that we'll probably get into later in this discussion.

但我的意思是你对AI的熟悉程度或了解有多深。

But, you know, how well or how acquainted you are with AI is is what I mean by that.

而其他方面，也就是人工智能的价值观、优先级或输出结果，其实反映了其输入内容。

And the others are are basically the values or the priorities or the outputs of artificial intelligence can reflect the inputs.

对吧？

Right?

这不是简单的‘输入什么就输出什么’的纯净过程。

It's not a clean whatever goes in goes out.

所以我将这视为一个相互影响的循环。

So I think of this as a cycle where they both influence each other.

我们过着这样的生活，某种程度上是在适应最新的AI，或者不得不使用它，我们不断学习如何使用它，这就是所谓的AI素养。

So we live our lives, you know, sort of adjusted to the newest AI or or we have to use it, We are constantly learning about how to use it, you know, that's, you know, AI literacy.

还有另一个方面，即文化价值观、某些优先事项或偏见会进入这些人工智能模型。

And there's the other aspect where the cultural values or the certain priorities or biases go into these AI models.

因此，它可能会把这些东西反馈出来。

And as a result, it's you know, it can it has chances of spitting those things back out.

所以，人工智能在某种程度上实际上强化了文化价值观、文化认知或优先事项。

So AI in a way actually reinforces cultural values and cultural understanding or priorities.

但与此同时，另一方面，它也有这种强化事物的能力。

But at the same time, on the other hand, it has this ability to reinforce things.

强化可能意味着它强化了文化，但也可能在某种程度上导致文化侵蚀，因为我们正在应对这些算法中呈现的趋势或某些内容，而这些正是我们所消费的信息。

And reinforcement can mean that it it reinforces culture, but it also can reinforce or it can it can move towards a cultural erosion in some ways, where we are dealing with trends or certain things that are in the these algorithms that are put out, and as a result, that's the content or that's what we are feeding on.

所以我认为两者彼此产生深远影响。

So I think that they both very much impact one another.

因此，任何输入人工智能的东西，都会以某种形式输出，而我们会据此调整，然后再将这些内容反馈回人工智能系统。

So whatever goes into AI, you know, has it comes out, and we adjust to that, and then we put it back into the AI machine.

我觉得我观察到了一些差异，比如中国的人工智能发展与全球或国际人工智能应用之间的区别。

I feel that I've observed some of the differences, I think, comparing to the development of AI from China and the global or international AI applications.

因为我觉得近年来，我们看到了大量新的AI应用涌现，它们聚焦于不同的领域，无论是大型语言模型、聊天机器人，还是生成视频和图像。

Because I feel that in recent years, of course, we're seeing a lot of these new AI applications coming out, and they are focusing on different areas, no matter it's about large language model, they can they can be chatbots, or they can generate videos or images.

我认为，许多国际AI模型都专注于提升生产力。

And a lot of these, I think, international AI models, they are focusing on enhancing the productivity.

而最近，在中国非常流行的那些AI模型，则更关注如何让生活更轻松，或者无缝融入日常生活，努力将使用AI的门槛降到极低。

While in recently, a lot of these AI models that are really, really popular here in China, they are more focusing on how can make your life easier, or they are just seamlessly integrating into your lives and trying to making the threshold of using an AI really, really low.

在最近这种情境下，我观察到的差异是：在中国的这些AI应用正试图真正实现我们所说的AI素养——当然，我们有能力、也应该有能力学习并使用AI来让生活更轻松或更高效。

And I observed that kind of differences in this kind of scenario or in recent days, is I think these AI applications here in China, they are trying to like, really, what we're talking about, the AI literacy, of course, we have the ability or we should have the ability to learn, to use this AI to make our lives easier or more productive.

但问题是，中国的一些AI应用根本不需要你去学习，就能帮你轻松完成任务。

And the thing is that some of these AI application in China, they are just helping you to make your live user without even you to learn it.

这也是这些公司努力实现的目标之一。

That's also some of the purposes or aims of these companies they're trying to do.

而

Where

AI变得如此出色，以至于它在我们的生活中已经不再显而易见了。

AI becomes so good to the point where it's not even that recognizable in our lives anymore.

它无时无刻不在我们身边，但我们却不知道自己是否正在使用它。

It's around us all the time, but we don't know whether we're using it or not.

我觉得这是一个很中肯的观点。

And I think that's a that's a fair point.

让我们谈谈伦理方面的问题，因为我觉得，我读到过很多内容，认为我们现在正处在一个非常重要的转折点。

Let's get to the ethics side of things because I think, you know, I've I've read a lot that we're at a really important inflection point right now.

我们正处在一个十字路口，新技术层出不穷，如果我们不时刻提醒自己AI的伦理重要性，可能会陷入为时已晚的境地。

We're at a we're at a we're at a crossroads where things are coming out so quickly, and if we don't if we're not careful about reminding ourselves of the ethical importance of AI, we might get to a point where it becomes too late.

但我想谈谈娱乐业和媒体业。

But I wanna talk about the entertainment industry and the media industry.

这是如今一个主要的讨论话题。

This is a major talking point these days.

它正在越来越多地塑造人们在社交媒体动态中观看的内容、聆听的音乐。

It's it's increasingly shaping what people watch in their social media feeds, what they listen to.

所以当你思考AI在娱乐领域的伦理问题时，你看到了哪些风险？

So when you think about the entertainment side of AI and ethics, what risks do you see?

我们生活在一个由AI算法驱动的文化中。

We're living in an AI algorithm driven culture.

我们生活在一个被推荐系统、信息流等事物不断包围的环境中。

We're living in a constantly surrounded by recommendation systems, feeds, things like that.

你认为那里存在哪些风险、问题或危险，艾斯？

What risks, what problems, what dangers do you see there, Ace?

是的。

Yeah.

这些推荐算法，你知道，它们的目标是预测我们会与什么互动，或者什么能让我们持续关注某个内容。

So those, you know, recommendation, those algorithms, you know, they aim to sort of predict what we'll engage with or, you know, what will keep us engaged with that particular thing.

这可能基于我们的观看时长、点击次数，或者我们在应用上花费的时间。

So it could be based off of our watch time, how many clicks, you know, we've had on it, or how much time we've spent on the apps.

从某种意义上说，这并不仅仅关乎这些，因为这种算法的不完整性会带来一些伦理风险。

And it's not necessarily about much more than that, in a sense, where there are a few ethical risks that come with that sort of incomplete nature of algorithms.

第一个风险，实际上，《哈佛商业评论》上有一篇文章对此进行了很好的阐述。

The first one actually, there's an article on Harvard Business Review that covered this really well.

但它实际上有助于区分习惯与真实偏好。

But it basically differentiates it helps differentiate habits between true preferences.

因此，这些系统通常只是观察我们的行为，却未必帮助我们实现某些目标或价值观。

So oftentimes, the systems, these recommendation systems simply just watch our behavior and don't necessarily help us to accomplish certain goals or values we may have.

举个例子，你每周一点披萨，只是因为这样方便。

So an example is, you know, you order pizza every Monday out of the fact that it's convenient for you.

但算法并不知道你是出于便利才这么做的。

But the algorithm doesn't know that you did it out of convenience.

它只是假设你非常喜欢披萨。

Rather, it just assumes that you love pizza.

因此，无论你是否想吃得更健康，它都会不断向你推荐披萨。

So it'll keep pushing it back, you know, into your system regardless of whether you want to eat healthier.

这就像是一个非常机械的助手，仅仅观察你的行为，却从不问你的感受。

So it's like a very literal assistant that doesn't really that just watches what you do and doesn't necessarily ask how you feel.

因此，这个问题至关重要。

So that issue is very paramount.

这一点在讨论算法未能真正展现对你有益的内容时经常被提及。

That's that's one that's talked about a lot with with regards to algorithms not exactly showing what is good for you.

你知道，'有益'的定义各有不同，但真正重要的是你实际想达成的目标，而不是你表面上的行为。

You know, and good is defined, you know, in its own way, but it's it's what is actually what you are actually trying to get at versus what you may do.

对吧？

Right?

有时候你可能会用鼠标在垃圾食品上停留一小时，但最终你点的却是更健康的食物。

Sometimes you may hover over, you know, junk food for like one hour with your mouse, but you actually might end up clicking on healthier food.

你知道，这些算法目前还有很多没有覆盖到的地方。

You know, so there is a lot that is not being covered yet in these algorithms.

因此，我们有时是基于不完整的信息集获得推荐，这可能导致后续出现伦理或其他问题。

So we are getting recommendations based off of incomplete data sets at times, which, as a result, can have ethical or, you know, these other issues down the line.

另一个担忧是，这些算法只会向你展示你喜欢或之前互动过的内容，这样一来，你更难偶然发现那些不熟悉的小众或具有挑战性的内容或领域。

The other one the other concern is, would these algorithms show you what you like or have previously engaged with, it's it's in that way harder to stumble into that unfamiliar niche or challenging content or those territories.

这不仅会把你屏蔽在某些领域之外，还阻止你接触可能喜欢的其他类型音乐或食物——你很难突破这些算法的限制，去偶然发现那些新领域。

So not only does this, you know, sort of shield you or not only does this prevent you from, you know, going into those areas that may you know, you may like a different type of music, or you may like a different type of food, but you're not able to you know, it's it's harder to get to those you know, to get past these algorithms to then stumble into that territory.

对吧？

Right?

这也可能伤害艺术家、餐厅或任何商品生产者，他们可能销售一些不为人熟知或小众的产品，但由于现状、趋势等原因，算法会引导你远离这些，而趋向于熟悉的东西。

And that also can can harm the an artist or a restaurant or any producer of a good, where maybe they sell something or they providing something that they be unfamiliar or maybe niche, and because there may be a status quo or or a trend or something, that it can steer you away from that and towards what is familiar.

我们都明白，当我们一直接触熟悉、不具挑战性的事物时，很容易陷入回音室效应。

And we all know that when we are dealing with things that are familiar all the time and not challenging, it can become somewhat of an echo chamber.

而回音室效应本身并不是完全没有问题的。

And, you know, echo chambers aren't necessarily the most unproblematic of things.

第三点是算法偏见的存在。

And the third thing is there's algorithmic bias that happens.

对吧？

Right?

任何基于人类数据训练的算法，都会吸收其中可能存在的偏见。

So anything that's trained you know, when when algorithms are trained on human data, they also absorb the prejudice that may be in it.

比如，一家公司在招聘工具中训练出倾向于男性员工的模式，尤其是在男性主导的行业，这可能会以规模化的方式复制这些不公平的模式。

You know, an example could be a company that hires hiring tools that learn to favor men in male dominated industries, which, you know, can reproduce these unfair patterns at a scale.

因此，算法偏见只是另一种方式，它排除了某些群体或那些使这一领域在伦理上充满挑战的因素。

So algorithmic bias is just another way that it cuts out a certain portion or a certain group or those things that that create that that make it an ethically challenging sort of landscape.

是的。

Yes.

这些所谓的信息茧房或回音室，当我们频繁在日常生活中使用人工智能时，已经被反复提及多年了。

These, like, echo chambers or information cocoons are the words that has been, like, pointed out for so many times or so many years when we're using AI so frequently in our daily lives.

当然，它们向我们推送的是我们已经关注过的信息，而你很难判断这是好事还是坏事，因为你喜欢它们提供的内容。

And, of course, they are feeding us the information that we've have been looking at, and you cannot really tell it is a good thing or bad because you like that content that they feed you.

对吧？

Right?

但问题是，我们不断重复地看到自己想看的信息，这真的好吗？

But the thing is that we are constantly and repeatedly seeing the information that we want, is it really a good thing?

也许这也是我们需要思考的另一个观点。

Maybe that's also another point that we need to think about.

另外，这类问题也出现在一些伦理讨论中，当然，今年的两会也提到了。

And another thing is that these kind of things are also being mentioned in some of the ethical problems and also, of course, this year's two sessions.

许多代表也在这一领域提出了建议。

A lot of these deputies, they are also giving out proposals in this kind of field.

例如，全国人大代表陈玉佳关注算法歧视，比如大数据杀熟。

One example is that National People's Congress deputy Chen Yu Jia focused on algorithmic discrimination, such as big data price discrimination.

我想我们之前也讨论过，如果你是某个平台的频繁用户，平台可能会给你比新用户更高的价格。

I think we also talked about it before that maybe you if you are, like, frequent user in that certain platform, and then maybe the price given to you even higher to a new user.

因此，这类场景以及低质量AI产品对弱势群体造成的伤害，比如农业AI工具因误判作物状况而带来的问题——因为在许多领域，人们也在使用AI。

So these kind of scenarios and also harm caused to vulnerable groups by low quality AI products, such as agricultural AI tools that misdiagnose crop conditions because in many fields, they are also utilizing AI.

对吧？

Right?

这位代表还呼吁对此加强监管。

So and these the deputy is also calling for stronger oversight in this sense.

我们稍后会谈到相关政策。

We'll get to policies in in just a little while.

但我想问问你，Ace。

But I wanna ask you about that, Ace.

我想我的问题有两个方面。

There's I guess my question is twofold.

首先，当我们看到所谓的算法推荐内容时，这些内容似乎是被推送到我们面前的。

One, if we compare because when we see you know, we talk about algorithmic feeds and and things that are pushed on us, so to speak.

而它们之所以被推给我们，是因为正如你之前所说，我们曾经点击过某些内容，让算法认为‘你喜欢这个’，就像你周一点了披萨，于是它又推荐给你。

Well, the reason that they're pushed on us is because, like you said, at some point before, we clicked on something that made the algorithm think, oh, you like that, just like you, you know, the pizza on a Monday, so it recommends it to you again.

所以从某种意义上说，正是我们自己创造了这些推荐内容。

So in effect, we are the ones who create the feed in a sense.

从市场营销或广告的角度来看，这与传统形式的广告或营销有何不同？

And then from a from a marketing point of view or from an advertising point of view, is this different from traditional forms of advertisement or marketing?

我想知道，你知道，Ace，你是个足球迷。

I wonder because if you watch I know you're a bit of a football fan, Ace.

如果你在周日下午看超级碗，你肯定会看到鸡翅、奶酪玉米片、不健康食品和啤酒的广告，因为广告商知道，在这个时候，这些产品会卖给正在观看比赛的人。

If you watch the Super Bowl, you know, on a Sunday afternoon, well, you're gonna see advertisements for things like chicken wings and nachos and unhealthy foods and beers and things like that because the advertisers know that's gonna sell at that particular time to the people that are watching.

这里有没有可比性，还是说我们面对的是完全不同的东西？

Is there any comparison comparison to be made there, or are we dealing with something completely different?

不知怎的，我觉得它们有点相似。

Somehow, it feels a little bit similar to me.

是的。

Yeah.

我认为它们非常相似，因为它们都利用了庞大的受众来促使人们想要他们的产品。

I would say that they are quite similar in the fact that they are certainly leveraging a large audience to create those to to make people want their product.

但与此同时，你知道，算法采取了更个性化的做法，基于你所表现出的偏好。

But at the same time, you know, algorithms take a more personalized approach where it's it's it is based off of indicated preferences you have.

对吧？

Right?

它知道你想要或者你曾经点过，抱歉，是点过。

It knows that you want or that you have ordered excuse me.

它知道你以前点过披萨，所以会推荐你五个不同的披萨店。

It knows that you have ordered pizza before, so it's gonna recommend you five different pizza places.

但超级碗的广告，你知道，它们有各种各样的广告，不只是食品，还有其他产品，比如服装，等等，它们的目标我认为是迎合如此庞大的观众群体，因此它们更注重让尽可能多的人看到自己的产品，而算法则可能只是更精准地向那些更有可能购买的人推荐产品。

But a Super Bowl ad, you know, they have a wide variety of of advertisements, not just about food, but they have other products, you know, maybe clothes, maybe, you know and so many different that that are there to, I think, cater I think what they are more focused on is the fact that there is such a large audience, and so they are pushing for their product to be seen by as many people versus an algorithm maybe simply or maybe recommending the products to somebody they are more confident will buy it.

因此，在这个方面它稍微更专业化一些。

So it's a little bit more specialized in that area.

是的。

Yeah.

这是你一个人对抗四亿人，或者不管有多少人观看超级碗。

It's you versus the 400,000,000 people or however many people watch the Super Bowl.

是的。

Yeah.

是的。

Yeah.

我还觉得，关于超级碗广告，它是一种一次性的事情，这个事件只在特定时间内发生，而你却一直在使用这些应用和工具。

I I also think that with regards to the Super Bowl ad, it's it's a one and done sort of thing where that event takes place, you know, for a certain amount of time, whereas, you know, you're always using these apps and these things.

因此，算法会持续更新以适应你的偏好。

So algorithms are constantly being updated and used for your preferences.

你提到了关于偏见和强化偏见的潜在风险，但它也能提升可及性。

You mentioned about the potential dangers when it comes to bias and and reinforcing bias, but it can also improve access.

它能提升代表性。

It it can improve representation.

你能想到一个例子吗？在创意产业、一般行业或招聘流程中，AI如何促进公平？

Can you think of an example where you might see AI helping equity when it comes to creative industries or just industries in general or hiring processes?

它表现不佳的领域是非常明显的。

The the areas where it falls short are are are quite obvious.

有没有什么方式或方法，它也能带来一些积极的作用？

Is there any way or any how that it can be doing something good as well?

是的。

Yeah.

当然。

Definitely.

教育是AI被用来提供帮助的一个绝佳例子。

Education is a great example of how AI is being used to help.

显然，它并非没有某些问题，但有人认为，AI在某种程度上能够使学习民主化。

Obviously, it it doesn't come without certain issues, but in a way, it has been said that AI can democratize learning in some ways.

对吧？

Right?

学生可以使用聊天机器人或AI导师，在教师资源匮乏的地方提供额外的个性化帮助。

Students can use chatbots or AI tutors to offer some extra personalized help where teachers may be scarce.

而且，你知道，这种帮助是24小时全天候的，更加个性化，因此学生可以获得他们需要的帮助，或者得到更有针对性的指导。

And, you know, that is twenty four seven, more personalized, and as a as a result, students can get the help that they need or can get more targeted help.

对吧？

Right?

所以，现在一个负担不起家教的学生，只要能使用ChatGPT的免费版本，就可以向它提问，比如：‘你能帮我理解光合作用是怎么回事吗？’

So now a student who couldn't afford a tutor, but who can afford the free version of ChatGPT can ask it questions and and maybe ask it, oh, can you please help me, you know, understand how photosynthesis works?

这样一来，这个孩子就有更大的机会通过这门课程，达到他应有的学习目标。

And now that kid can have a better chance at passing that class and getting where he needs to go.

因此，从某种意义上说，人工智能可以实现学习的民主化。

So in a way, artificial intelligence can democratize learning.

当然，很明显，它也有许多可能带来负面影响的方式。

Now, obviously, there are a lot of, you know, ways in which it can it can affect it negatively.

但人工智能在教育中的民主化特性确实值得肯定。

But there is something to be said about the democratizing nature of AI in education.

这是一个非常有趣的观点，因为当我们谈论这些人工智能发展得多么迅速、多么优秀时，我们总是感到焦虑。

That is really a interesting perspective because when we are talking about how fast, how good these AI are developing, we are always feeling anxious about it.

对吧？

Right?

一个例子是，最近非常著名的中国模型叫做C-Dance 2.0。

One example is that the recent, very, very famous Chinese model is called C Dance two point o.

人们正在利用新闻里提到的一切。

People are using whatever the news Right.

C-Dance。

C Dance.

是的。

Yeah.

他们用这些模型生成逼真甚至具有电影质感的视频片段。

They're using them to generate that realistic, even cinematic vibe video clips.

然后，学习插画或动画的学生说，我们的工作可能会被这类工具取代。

And then students majoring in illustration or animation are saying that, okay, our jobs can be taken by these kind

的

of

人工智能。

AIs.

但换一个角度看，如今的学生实际上是站在巨人的肩膀上。

But from another perspective, you can also think that students right now, they are standing on the shoulder of the giants.

互联网上提供了大量由人工智能提供的资源。

You can have so much resources on the Internet offered by AI.

你可以直接生成这类图像或视频片段。

You can just generate these kind of images or video clips.

也许当你需要逐帧绘画，花费数月时间才能制作出两分钟的动画时，一个提示词就能帮你实现。

Maybe when you are working on drawing frames by frames, spending months of maybe creating two minute of that animation, and then only one prompt can give you that.

但关键是，我们现在确实能获得这些资源，可以从中学习，从而对他人——甚至国际上的创作——有更深入的理解。

But the thing is that, of course, we now can have the resource, and you can learn from these kind of resources and have an also better perspective on what people are doing, maybe even internationally.

所以我认为它也为学生提供了更多渠道，让他们能够从这些不同的资源中学习。

So I think it is also offering more, I think, channels for students that they can actually learn from these different resources.

是的。

Yeah.

我经常用它来简化概念、理解某些东西，以便更好地表达我的想法。

I use it myself all the time to simplify concepts and to understand some things, to better help me express an idea as well.

我认为在学习方面，这确实有其积极的一面。

And I think there is definitely a an upside there with regards to learning.

对于创造力和教育来说，通常它们是两个不同的领域，但在这里由于对它的过度依赖而产生了一些重叠。

Well, for creativity and for education, I mean, usually, those are two separate fields, but somehow they overlap here because of the perhaps over reliance on it.

我知道当AI刚进入课堂时，老师们都很恐慌，因为学生们交上来的论文里每一篇都用了破折号，还有‘Yeah’。

I know that when AI first became a thing in the classroom, teachers were freaking out because students were handing in essays, and every essay had an em dash and Yeah.

在结尾提出一个问题

Stated a question at the end of

不仅，而且。

the Not only, but also.

是的。

Yeah.

对。

Yeah.

没错。

Exactly.

所以这一点变得非常明显，老师们都非常不安。

So it became blatantly obvious obvious, and and teachers were up in arms.

然后，你知道，再过一两年，突然之间，又出现了一场关于教师过度依赖课堂工具的新争论。

And then, you know, fast forward maybe a year or or two later, and all of a sudden, there's a new debate about teachers' overreliance on the use of of the tools in the classroom.

我们可以像玉顺说的那样，只需几条提示、几分钟就能制作出一部完整的电影，而演员、导演、配音演员和制片人却需要数月甚至数年，还要巨额预算。

We can make like Yushun just said, we can make probably full length movies in a in a matter of prompts and and minutes, whereas actors and directors and voice artists and producers would take months and years in a large budget.

我认为很容易看到双方的观点，我们可以说这些工具是用来辅助的。

I think it's easy to see the arguments on on both sides where we can say the tools are there to assist.

为什么不加以利用呢？

Why not make use of them?

但另一方面，过度依赖这些工具会削弱我们自身的学习能力和创造力。

And the flip side of that coin is, but, yeah, but overreliance on the tools takes away something from our own ability to learn, our own ability to create.

你在这场争议中持什么立场？这基本上是如今每一场人工智能对话的核心问题。

Where do you where do you lie on this issue, which is basically the crux of of what every AI conversation is is about these days?

是的。

Yeah.

我尽量坚持一个观点，那就是人工智能是一种工具，而不是替代品。

I try to keep it centered on the idea that AI is a tool, and it's not a replacement.

所以我通常把它用在一些繁琐的工作上，而把判断和决策的部分留给自己。

So I like to think of it as I use it for grunt work sometimes, and I keep, you know, my my contribution as far as the judgment aspect.

你知道的。

You know?

例如，你可以用它来起草论文、翻译或构思想法，但最终的结构、情感核心和语气还是要由你来决定。

For example, you can draft a paper or translation or ideas or something, but you will decide the structure, the emotional core, the tone of it.

我经常把它当作合作者，让它向我提出一些具有挑战性的问题，以检验我论文中可能要论证的观点，或者提供替代方案。

And I use it a lot as a collaborator too, to ask me, you know, challenging questions to sort of stress test whatever it is I may need to argue in a paper or something or suggest alternatives.

通过这些方式，我们能够将其作为工具使用，同时保持一种不可替代的地位，坚守那些属于人类独特优势的领域——作为情感生物，能够感知、体验这些事物，我们依然掌控着核心部分，同时让工具为我们提供辅助。

And in those ways, are able to use it as a tool, but also maintain a position that is irreplaceable, and to maintain a strong position that allows you to remain intact with things that are, our strong points as human beings, as emotional beings, as beings that feel, that can sense, you know, these things that we allow ourselves to remain in control of, but allowing the tool to help us.

因为当今世界的发展趋势，每个主要国家和超级大国都希望将人工智能融入其运作体系中。

Because the way that the world is going right now, every major country superpower wants to incorporate artificial intelligence in their operations.

人工智能是一项极其重要的资产。

It's an incredibly important asset.

因此，人工智能在我们生活中渗透的程度，并不完全由我们控制，毕竟它已经无处不在；但我们能做的，是明确自己如何使用它，始终清楚地认识到：我使用它作为工具，而非替代自己。

So the permeability or how how much AI is involved in our lives is not necessarily something we can control in this way of just because how much it's around us already, but what we can do with it and how we can maintain a sort of position and keep it very clear that I use this as a tool, but not as a replacement for me.

对吧？

Right?

我曾和一些律师讨论过这个问题，他们担心会被人工智能取代，但我觉得不必如此害怕人工智能会取代你。

And I've had some discussions with I was actually talking to some lawyers who were afraid of being replaced by AI, and I I'd think of it as not necessarily being afraid of of AI replacing you.

问题是，许多公司或学校严格禁止使用人工智能，有些甚至如此。

The thing is that a lot of these companies or schools, they strictly prohibited the use of AI, some of them.

但现在，一些小学也开始教授学生如何使用人工智能。

And now, also, in some primary school, they also teach they started to teach students how to use AI.

所以我们并不是说任何一方是错的。

So, we're not saying that, you know, either side is wrong.

但关键是，我们需要改变对人工智能的看法，以及在时代变化时如何使用AI的理念。

But the thing is that we need to change our mindset or the idea towards the AI, and the idea of how to use AI when the time is changing.

是的。

Mhmm.

我只是担心公平问题。

I just worry I just worry about equity.

你知道，Ace提到过，这是一个很好的观点，即那些课外负担不起老师或辅导班的学生。

You know, you said, Ace, about and and it's a great point about how, you know, students who might not be able to afford a teacher or a tutor outside of class time.

现在，他们只要握住手机，就能拥有一个随身的辅导老师。

Now they have a tutor in the palm of their hand when they when they hold their phone.

但与此同时，我们看到越来越多的私立学校兴起，这些学校完全以AI为核心。

But at the same time, we see private schools popping up, which is fully AI focused.

我们最近讨论过的一所叫Alpha School的学校。

One that we were discussing recently is called Alpha School.

我不确定你是否熟悉这个学校。

I'm not sure if you're familiar with that one.

它最近在美国出现了。

It's popped up in The States.

每年的学费大约是65,000美元。

It's something like $65,000 a year.

当然，不是每个人都能负担得起每年65,000美元的学费。

Now Well Obviously, not everybody can afford to go to school for $65,000 a year.

所以这是一个例子。

So that's that's one example.

但如果我们从更宏观的角度来看，不是每个国家都能像富裕国家那样平等地获得AI工具。

But then if we look at it from a broader context, not every country will have equal access to AI tools that, you know, a wealthier country might have.

所以这又是一个平衡的问题。

So again, it's the balancing act.

但让我们不要谈教育方面，转而聚焦于国家层面。

But let instead of the education side, let's focus on the countries.

我们如何确保未来每个人都能平等地使用相同的工具和相同的AI，从而避免在全球范围内造成更大的差距？

How can we ensure that that we can all have access to the same tools, the same AI in the future so it doesn't create an even greater divide around the world?

这个问题要求很高。

That's a tall order, that question.

我认为，如果我们想推动人工智能的公平性，就必须克服一些实际障碍，但其中非常重要的一点是人工智能素养。

I think that if we wanna move towards equity in artificial intelligence, I think that we have to overcome a few logistical steps, but also one thing is that AI literacy is incredibly important.

一方面，存在所谓的数字鸿沟，即有些人根本无法接触这些技术，而有些人则可以。

So there's one aspect of, you know, what we call the digital divide as people who simply don't have access to it and people who do have access.

再进一步，是如何正确地利用这些技术。

And the other, you know, a few steps further is being able to leverage those things properly.

至于各国该如何迈出获得访问权限的第一步，我很难多做评论。

Now, I can't speak so much as to how countries can begin to get to that first step of getting access, you know.

我没有资格就这个问题或相关领域发表意见。

I'm not qualified to speak in that way or about that subject.

但我可以说的是，那些至少已经能接入互联网或初步接触到这些工具的国家，应当鼓励提升人们对人工智能的理解和数字素养，教导学生和公众如何正确使用这些技术。

But what I can say is that, you know, countries who do have at least access to the Internet or or these tools, the beginning phases of these tools even, is to encourage more understanding and more digital AI literacy, to teach students, to teach people as to how to use it properly.

正确使用它意味着知道如何向它提问，以及对其回应保持怀疑。

How to use it properly means how to prompt it, how to be skeptical of its responses.

怀疑的意思是不要轻易相信它表面所说的内容。

Skeptical meaning don't take what it says for surface value.

就像你在互联网上阅读任何内容时一样，你可能想要，或者确实应该去核实信息，尤其是当你在做研究或撰写文章时，必须进行事实核查。

Just like any time you read something on the Internet, you might want to, or you do want to, even especially if you're researching something, if you're writing about something, you need to fact check.

你需要确保自己了解它是如何思考的，或者它是如何得出答案的。

You need to make sure that what this you know, you need to follow the way that it's thinking or the way that, you know, it's getting to its answer.

这些方面绝对是各国可以重点关注并鼓励学生理解的，要以一种视角看待人工智能——它是一种工具，但同时也要以批判性的眼光看待它。当然，'怀疑'这个词可能不是最贴切的，因为它自带负面含义，更准确的说法是'批判性思维'。

So those sorts of things are definitely something that countries can really hone in on and to try to encourage students to understand and look at AI from a a lens that it's a tool, but also to be skeptical of it in ways skeptical, you know, I I maybe skeptical isn't the best phrase because it automatically has a negative connotation, but a more critical lens.

听起来你像是在谈论一个你不太确定的表亲。

Sounds like you're talking about a cousin that you're not quite sure of.

你知道的？

You know?

我该相信他说的话吗？

Like, do I trust what he's saying?

我该相信她说的话吗？

Do I trust what she's saying?

或者在相信之前，我应该先问问别人。

Or maybe I should ask somebody else before I take it to be true.

但我的意思是，这正是关键。

But the I mean, that's Exactly.

这正是我们在这里所面对的情况。

Kinda what we're dealing with here.

是的。

Yeah.

对。

Yeah.

好的。

Okay.

那么让我们转向商业方面。

So let's turn to the to the business side of things.

当我们谈论人工智能伦理时，对话通常会转向科技公司、电影、游戏、音乐和娱乐行业。

When we talk about AI ethics, the conversation often goes to tech companies, film, gaming, music, entertainment industry.

我想回到我之前提到的一个观点：我们现在正处在一个非常重要的十字路口，我听到有人说，如果我们现在不将伦理融入人工智能，那么我们只有大约五年的时间来做这件事，否则就太晚了，无法事后补救。

And I wanna come back to the point that I that I brought up before about how we're we're at a really important junction right now, and I've heard people say that if we don't put the ethics into the AI right now, we have about a five year window to do so, that it'll be too late, and we can't we can't retrofit it.

我看到的一个比较，我想是在弗吉尼亚教育委员会的网站上之类的，他们将其比作在汽车上加装安全带。

The comparison I saw, I think it was on the the Virginia Education Board website or something like that where I saw this, they compared it to retrofitting seatbelts in a car.

嗯。

Mhmm.

你不会把车开上路之后，才去考虑车是否安全，然后才说：哦，我们得之后再装安全带。

You don't put the car on the road and then figure out whether the car is safe, then go, oh, we have to put the seatbelts in later.

他们用这个来比喻人工智能，说我们需要在人工智能中植入伦理，因为不能事后才添加，这一点适用于各个行业。

They were comparing that to AI saying, we need to instill the ethics in the AI because we can't just attach it later on, and that applies to various industries.

你对此有什么看法？

What are your thoughts on that?

而且，我的意思是，你之前用了‘艰巨的任务’这个说法。

And, I mean, again, you you used the expression tall order before.

确保我们在正确的时间将伦理融入这些不同行业，这同样是项艰巨的任务吗？

Is that an equally tall order to make sure that we get ethics into these different industries at the right time?

是的。

Yeah.

我的看法是，最难的任务是及时将伦理融入人工智能系统。

I mean, I think the tallest order to me is getting ethics into the AI systems on time.

但关于这种不可逆转的临界点，或者说我们必须迅速制定伦理规范、对这些人工智能系统实施严格监管的观点，我完全同意，我们确实需要非常迅速地行动。

But with regards to, you know, this sort of point of no return or this idea that we need to be very prompt with our putting ethical regulations, you know, putting very stringent regulations on in these AI systems, I agree with it 100%, where we do need to be very fast with it.

因为我们都清楚，任何人都能查到人工智能在各个行业中的普及速度有多快。

Because it is I mean, we we all know and anybody can look up how fast AI is, you know, becoming prevalent in every industry.

因此，明确我们何时该使用它、如何使用它、为何使用它、何时不该使用它，这些问题就显得尤为重要。

And that's why it's important that we have clear understanding of what you know, when we should use it, how we should use it, why we should use it, when we shouldn't use it, these sorts of questions.

但那个安全带的例子非常好，说明我们确实需要及时纳入这些伦理考量。

But and that seat belt example is a very good one, where we need to be prompt with putting in those ethical concerns.

但同时，我更愿意将其视为一种持续的对话或持续的试错过程：我们设立伦理规范是为了提供一个框架，但这些规范也应根据我们的需求、本地化需求、未来需求等变化而不断补充和调整。

But I think at the same time, I would like to view it more as a constant conversation or a constant trial process where we have ethical regulations that are put there in place to provide a structure, but they are also there to, you know, be added on or to be adjusted as our needs, as localized needs, as the future needs, whatever these things may be different.