人工智能正在让我们所有人都变笨

和Finn一起，我们打造了目前客户服务中心最顶尖的AI代理。

With Finn, we've built the number one AI agent for customer service.

它能解决企业高达90%的客户咨询，性能在G2排行榜上遥遥领先，还附带一百万美元的保障承诺。

It solves up to 90% of queries for businesses, tops all the performance benchmarks in the g two leaderboard, and it comes with a million dollar guarantee.

想去了解的话，上finn.ai看看吧。

Check it out at finn.ai.

本节目由Odoo赞助支持。

Support for this show comes from Odoo.

做生意已经够难了，干嘛还要用十几个互不相通的软件来给自己添堵？

Running a business is hard enough, so why make it harder with a dozen different apps that don't talk to each other?

隆重推出Odoo。

Introducing Odoo.

这是你这辈子唯一需要的商业软件。

It's the only business software you'll ever need.

它是一个全功能、深度整合的平台，帮你轻松搞定客户管理、财务、库存、电商，还有更多功能。

It's an all in one fully integrated platform that makes your work easier CRM, accounting, inventory, ecommerce, and more.

最棒的是什么？

And the best part?

Odoo 用不到几分之一的价格，就取代了多个昂贵的平台。

ODOO replaces multiple expensive platforms for a fraction of the cost.

所以有成千上万家企业都选择了切换到 Odoo。

That's why over thousands of businesses have made the switch.

那你为什么还不试试呢？

So why not you?

去 odoo.com 免费试用 Odoo。

Try Odoo for free at odoo.com.

就是 odoo.com。

That's odoo.com.

Support for this show comes from Running a business is hard enough, so why make it harder with a dozen different apps that don't talk to each other?

Support for this show comes from Running a business is hard enough, so why make it harder with a dozen different apps that don't talk to each other?

Introducing Odoo.

Introducing Odoo.

这是你这辈子唯一需要的商业软件。

It's the only business software you'll ever need.

它是一个一体化的完整平台，能让你的工作轻松很多。

It's an all in one fully integrated platform that makes your work easier.

客户管理、财务、库存、电商，还有更多功能。

CRM, accounting, inventory, ecommerce, and more.

最棒的是什么？

And the best part?

Odoo 用不到十分之一的价钱，就能替代多个昂贵的系统。

ODOO replaces multiple expensive platforms for a fraction of the cost.

所以成千上万的企业都选择了切换到它。

That's why over thousands of businesses have made the switch.

那你呢？

So why not you?

去 odoo.com 免费试用 Odoo 吧。

Try Odoo for free at odoo.com.

这就是 odoo.com。

That's odoo.com.

今天的数字是十万。

Today's number, 100,000.

据说一张梅特艺术晚宴的门票要花这么多钱。

That's how many dollars it reportedly cost to get one ticket to the Met Gala.

而一张桌子的价格是三十五万美元，参加者管这叫半次整容。

Meanwhile, the price of a table was $350,000 or as attendees call it, half a facelift.

我来给你看看这阵势。

I'll give you size.

金融市场疯了。

Money market's mad.

如果钱是邪恶的，那栋楼就是地狱。

If money is evil, then that building is hell.

演出还得继续。

The show goes on.

价格根本没人关注。

The price are never watched.

这场秀本身就在卖钱。

The show sell.

欢迎来到Prof G市场频道。

Welcome to Prof G Markets.

我是埃德·埃尔森。

I'm Ed Elson.

今天是5月12日。

It is May 12.

咱们来看看昨天市场的关键数据。

Let's check-in on yesterday's market vitals.

主要指数全线上涨，芯片股领涨。

The major indices all climbed led by a rally in chip stocks.

标普和纳斯达克都创了新高。

The S and P and the Nasdaq both hit new records.

尽管特朗普总统拒绝了伊朗提出的停战提议，但市场仍实现了这些涨幅。

Those gains came despite president Trump's rejection of Iran's proposal to end the war.

他还说，停火协议现在‘全靠生命维持系统吊着’。

He also said the ceasefire was on, quote, massive life support.

随着和平希望破灭，布伦特原油价格上涨，十年期国债收益率也上升了。

Brent crude climbed higher as hopes for peace faltered, and the yield on 10 treasuries rose.

明白了。

Okay.

还有哪些动态？

What else is happening?

从十九世纪以来，每一代人都比父母更聪明，唯独Z世代例外。

Since the eighteen hundreds, every generation has been smarter than their parents except for Gen Z.

神经科学家杰里德·库尼·霍瓦斯博士上个月在国会这样表示。

That is what neuroscientist doctor Jared Cooney Horvath told congress last month.

如今，90%的大学生和84%的高中生在课堂上或做作业时都会使用人工智能。

Today, ninety percent of college students and eighty four percent of high schoolers use AI in class or for their homework.

根据OpenAI自己的数据，AI最常见的用途之一就是写作。

And according to OpenAI's own data, one of the most common use cases for AI is writing.

与此同时，一项最新研究发现，商科学生使用AI工具与批判性思维能力下降有关。

Meanwhile, a recent study found that AI tool usage among business students was associated with weaker critical thinking skills.

这些数据引发了一个重要问题。

And this data raises an important question.

当我们把工作和思考都交给AI时，我们究竟失去了什么？

And that is what do we lose when we outsource our work and our thinking to AI?

毕竟，每周都有九亿人使用ChatGPT。

After all, 900,000,000 people use ChatGPT every week.

换句话说，AI是不是正在让我们所有人都变笨？

In other words, is AI making all of us dumber?

你可能还记得，我们上周就讨论过这个问题。

Now you might remember that we discussed this question last week.

我们最近又对这个问题做了一些更深入的调查。

We've been investigating this question a little bit more.

但今天，我们请来了两位专家，他们一直在思考这个问题，对这些议题有深入理解。

But today, we wanna bring in two experts who are thinking about this, who understand these issues.

所以今天我们要换一种方式来聊。

So we're gonna do something a little bit different.

今天不再谈市场，我们专心探讨这个问题。

We are going to move away from the markets for today and focus on this question.

我们邀请到了乔治城大学计算机科学教授、《慢生产力》和《深度工作》等八本畅销书的作者卡尔·纽波特。

So we're joined by Cal Newport, professor of computer science at Georgetown University and New York Times bestselling author of eight books, including Slow Productivity and Deep Work.

还有《纯英语》播客的主持人、《富足》一书的作者德里克·汤普森。

And we've also got Derek Thompson, host of the Plain English Podcast and author of Abundance.

卡尔和德里克，非常感谢你们两位加入我们。

Cal and Derek, thank you so much for joining us.

欢迎来到节目。

Welcome to the show.

卡尔，我先问你，因为你写过这方面内容，也谈过‘认知健康’这个概念，还提到它可能正在衰退。

Cal, I'm gonna start with you because you have written about this and you've talked about this idea of cognitive fitness, and this potential reality that it's in decline.

你觉得现实中AI的使用正在对我们的大脑产生什么影响？

What do you make of what's happening on the ground in terms of AI usage and what it's actually doing to our brains?

我觉得AI真的有让我们变笨的潜力。

Well, I think AI has the real capacity to make us dumber.

它还太新，使用也还在增长，所以大影响还没显现，但我担心迟早会来。

It's new enough, and usage of it is still growing that we're not seeing the major effects yet, but I fear that we are gonna see it.

我理解认知健康这个概念时，觉得是社交媒体和手机上那些让人上瘾的工具最先开启了这种趋势。

And the way I conceptualize this world of cognitive fitness is that social media and highly engaging tools on our phones started this trend.

它们让我们远离了那些需要持续专注的活动，而这些活动本来能锻炼我们的大脑。

It moved us away from more sustained concentrated activities through which we strengthen our brain.

现在，AI又把目标对准了另一个让我们变强的核心认知活动——写作。

AI is now taking target on the other main cognitive activity that makes us stronger, which is writing.

很多人用AI来缓解面对空白文档时那种无从下笔的压力，这正成为它最主要的应用之一。

This is emerging as one of the major uses of this tool is to alleviate the strain you feel when you look at a blank page and have to, fill that blank page.

如果AI真的大幅减少了我们的写作量——不管是重要的文章还是简单的备忘录，我认为这会延续十年前智能手机过度干扰带来的智力衰退趋势。

So if AI does, in fact, significantly reduce the amount of writing we do, whether it's super important or just a memo, I do think we're gonna see a continued diminishment of our intelligence that began with, highly distracting phones about ten years ago.

我们一会儿再聊该怎么应对这个问题。

We'll get into what we do about this.

不过，德里克，你同意卡尔的说法吗？

But, Derek, do you agree with Cal?

是的。

Yeah.

当然。

Of course.

当然了，他说得没错。

Of course, he's right.

也许过几分钟，我会谈谈我和卡尔在某些观点上的不同。

You know, maybe we'll explore some disagreements between me and Cal in a few minutes.

但就这一点来说，他说得非常到位。

But on this, I think he said it right on the money.

如果你对卡尔的说法有怀疑，又在用AI，那就留心一下自己的生活吧。

I mean, like, if you doubt what Cal is saying and you use AI, pay attention to your own life.

留意你自己是怎么用时间的。

Pay attention to your own use of time.

当你让人工智能帮你总结一篇文章、一篇论文，或者天啊，甚至整本书的时候，你真的觉得你对这些内容的理解，能跟自己读完一样深吗？

When you ask artificial intelligence to summarize an article or to summarize a paper or, God forbid, to summarize an entire book, do you understand that article, that paper, or that book as well as if you'd read it?

当然不可能。

Of course not.

好吧。

Okay.

也许你会说，没错，我确实省了时间——本来读一本书要十个小时，现在我能用同样的时间总结十五本书。

Now maybe you could argue that, all right, well, I saved time because now rather than read that one book, which might have taken me ten hours, I can summarize 15 books, and that'll take me sort of ten hours to process or something.

但即便如此，你对每本书的接触都太浅了，我怀疑你根本没法真正看懂它们之间是共识多还是分歧多，更别提你因此失去了长时间专注阅读的能力。

Well, even there, you're engaging at such a shallow level with each book that I'm not sure you really understand the degree to which they agree and disagree with each other, but also what you're depriving yourself from the inability to read anything for more than five or ten minutes at a time.

而这种能力，长期积累下来，才是形成深刻洞察、真正有见地的思考的基础。

And that is a skill that leads over time to the ability to make the sort of deep connections that I think are the basis of all true insightful thinking.

所以我觉得这里的风险，确实像我之前说的，至少有两层。

So I absolutely think that the risk here is really, I guess, as I described it, sort of of at least two layers.

第一，你正在剥夺自己真正理解某件事的体验，而你本以为自己是在努力理解它。

One, that you're depriving yourself of the experience of truly understanding something that you think you're trying to understand.

第二，你会失去一种对未来深度思考至关重要的习惯。

And number two, that you fall out of a habit that's that is necessary to think deeply in the future.

至于卡尔的第一个观点，我就先说到这里，虽然顺序有点打乱——我们现在看到的，比如弗林效应，还有随着时间变化的考试成绩。

And to Cal's sort of maybe first point to end there because of going achronologically, you know, we're looking at things like the Flynn effect, and we're looking at things like test scores over time.

如果我们正在削弱五年级、六年级孩子思考的能力，而他们到了七年级、八年级，一直到十二年级、大学，都继续依赖AI，那就不只是失去了一年深度阅读和深度思考的机会。

Well, if we're depleting the inability of fifth graders and sixth graders to think, and they continue to use AI in seventh grade and eighth grade and through twelfth grade and through college, that's not just one year of losing the practice of doing deep reading and deep thinking.

我们现在谈的是整整十年——一个关键的成长期，你却选择不去锻炼那种我常比喻为肌肉的能力，不去培养长期真正理解事物、变得睿智所必需的思维能力。

Now we're talking about a decade, a formative decade, that you've chosen to essentially not work on the kind of muscles I do like this fitness metaphor, not work on the kind of muscles that are so necessary in the long run for understanding something deeply to be smart about it.

所以，我完全同意卡尔的看法。

So absolutely, I think the cal's onto something.

我觉得我们在这里其实识别出了两种主要的力量。

I mean, it seems like there are two main forces that we're kind of identifying here.

一种是屏幕本身，以及我们对屏幕日益加深的依赖。

One is the screen in general and our increasing addiction to those screens.

另一个则是人工智能，以及我们对AI解决更复杂、更微妙问题的依赖。

And then the other is AI and our dependency on AI to solve harder problems, more nuanced difficult problems.

卡尔，回到你刚才说的，现在哪个因素影响更大？这个问题现在还成立吗？

Cal, just going back to you, which is the more dominant force or is that even a relevant question right now?

到目前为止，影响最大的其实是过去十年里，那些高度优化的移动设备互动体验。

Well, the biggest impact so far has come from a decade of hyper optimized engagements on a portable device that we have with us at all times.

这种影响非常巨大。

That has had a massive impact.

本质上，背后那些机器学习算法，尤其是短视频平台的，已经模拟出了我们大脑中短期奖励系统的运作方式，专门推送能强烈刺激这些神经回路的内容。

Essentially, what happened is the machine learning algorithms behind, especially short form video platforms, built an approximation of our short term reward centers in our brain so that it could give exactly the signal that's going to resonate strongly with those particular circuits.

这让手机变得根本无法抗拒。

This makes the phones essentially irresistible.

只要手机在我身边，我就忍不住要拿出来。

When it is with me, I have to take it out.

我非得看上一眼不可。

I have to look at it.

所以过去十多年里，这种现象已经严重损害了多代人保持专注的能力，更不用说建立那些在需要时用于深度思考的神经回路了。

So that over the last decade or so has done substantial damage to multiple generations' ability to actually not just sustain attention, but again, to build those circuits you can use to think deeply when the time is required.

这些回路是通过阅读和写作培养出来的。

These circuits are built through the activities of reading and writing.

在现代人类历史上，阅读和写作曾经是极为珍贵的活动，

These are privileged activities in the history of modern humanity,

后

post

旧石器时代的人类。

Palaeolithic humanity.

人工智能虽然才刚出现，但我总觉得，它会成为我们早已面临的屏幕高吸引力内容的灾难性延伸。

AI is new on the scene, but I really feel like it's gonna be a catastrophic cousin to what we already were encountering with hyper engaging content on a screen.

因为如果真的专注于阅读，我们现在早已不再能静下心来，像过去那样专心读完一本书，从而建立深刻的理解。

Because if that really focused on reading, we no longer sit and concentrate on a book in a way that could build deep understanding.

写作则是它的配套活动。

Writing was its partner.

写作是阅读的搭档。

Writing is the pair to reading.

写作就是把我们通过深度阅读培养出来的思维模式反过来运用，创造出属于自己的原创想法。

Writing is where we take the circuits we etched with deep reading, and then we apply them in reverse to create original thoughts of our own.

我们同样需要锻炼这种能力。

We have to practice that muscle as well.

而如今，我们第一次可以大规模地把这项活动外包出去。

And now for the first time, we can begin to substantially outsource that activity.

所以我把阅读和写作看作是两种实实在在的活动。

So I really think about reading and writing as activities.

这并不是怀旧。

This is not nostalgia.

也不是说，我们如今在汽车时代还念叨着马车。

This is not, oh, we're talking about horse buggies in an era of automobiles.

我真心认为，正是阅读和写作塑造了后旧石器时代现代人类的大脑——那个孕育了神学、政治、哲学，以及我们一切珍视之物的大脑，它们从根本上依赖阅读与写作来成型。

I really do think those are the activities on which the post paleolithic modern human brain were built, the brain that gave us theology, that gave us politics, that gave us philosophy, that gave us theology, the brain that everything we hold dear was built around substantially dependent on reading and writing to shape it.

所以我特别担心，我们已经失去了阅读，现在又出现了一个新工具，可能会连写作也一并取代。

So I'm really worried about what we already lost with reading, and that we have a new tool that's gonna start to take writing off the table as well.

听起来你是在说，有人可能会反驳说，历史上每一种新技术不都让生活变得更方便了吗？

It sounds like sort of the argument you're making because, you know, someone would say in response or if someone were to push back, the argument would be, well, every technology in history has made our life easier in some capacity.

比如你发明了发动机，造出了汽车，出行就方便多了。

Like, you know, you invent the engine, you invent the car, it makes it easy to get around.

他们可能会说，这个工具也是同样道理，让批判性思考变得更轻松，就像其他技术让其他工作变得更简单一样。

And the the argument would be this makes it easier to do the job of critical thinking in the same way that other technologies do make other jobs easier.

卡尔，你的意思似乎是，这次不一样。

Cal, it sounds like what you're saying is that this is different.

批判性思考这种能力，是属于更高层次的。

The brain critical thinking is on a different level.

它已经深深融入了人性的本质，所以这真的不是好事，和其他技术不一样。

It's so endemic to what it means to be a human to the point where this is actually a bad thing unlike other technologies.

这样理解对吗？

Would that be the right characterization?

不是。

No.

我觉得你说得对。

I think that's right.

如果用健身来打比方，阅读是一种很好的技术，能帮我们提升批判性思维。

If we use the fitness analogy, reading was a great technology to make us better at critical thinking.

写作也是一种很好的技术，能帮我们提升批判性思维，但用AI就像把叉车开进健身房，说：‘我们在这儿练了好几年了。’

Writing was a great technology to make us better at critical thinking, but to use something like AI is like bringing a forklift into the gym and be like, you know, we've been in here for years.

我们一直靠举重来让自己变强。

We've been using weightlifting to try to get stronger.

结果我发现，用叉车省事多了。

Well, I figured out with a forklift, it'll be a lot easier.

我不用自己动手举重了。

I don't have to lift the weight myself.

但这么做其实背离了真正目标——锻炼认知能力，让大脑真正变强。

You're actually being counterproductive to the actual goal, which is strengthening the cognitive muscle to get stronger.

所以，我认为这根本不是一种能提升我们批判性思维能力的技术。

So no, I do think this is not a technology that's making us better at critical thinking.

它只是让我们绕开了过去那些真正锻炼大脑的艰难过程。

It's allowing us to sidestep the hard activities that previously we used to make our brain stronger.

这款产品所宣传的卖点，是当下的便利，而不是让你的大脑更强大、思维更敏锐。

The product, the benefit being sold by this product is convenience in the moment, not a stronger brain or stronger ability to think.

广告插播后，继续收听本场对话的后续内容。

Stay tuned for more of this panel right after the break.

顺便说一下，我们月底就要开始巡演了。

And by the way, we are heading out on tour at the end of the month.

想了解更多信息，或者购买附近场次的票，欢迎访问 profgmarketstour.com。

So for more info and to get tickets to a show near you, head to profgmarketstour.com.

本节目由 Hostinger 赞助支持。

Support for the show comes from Hostinger.

对大多数创业者来说，最大的障碍并不是资金不足。

The biggest barrier to entry for most entrepreneurs isn't a lack of capital.

最难的是迈出第一步。

It's the friction of starting.

你可能会花上几个月在策划阶段，可这些时间本可以用来真正行动起来。

You can spend months in the strategizing phase, which is precious time that could instead be spent actually making moves.

但如今，规则已经变了。

But these days, the rules have changed.

人工智能正在重新定义谁能创办企业。

AI is redefining who gets to build a business.

所以，当你打造下一个大项目时，用Hostinger，几分钟就能上线，不用等几周。

So when you're building the next big thing, go live in minutes, not weeks, with Hostinger.

Hostinger是一个一体化平台，把域名、网站、邮件营销、AI工具和AI助手全都整合在一起。

Hostinger is an all in one platform that brings everything into one place, your domain, website, email marketing, AI tools, and AI agents.

这样你就不用到处拼凑五个不同的服务了。

So you can launch online without stitching together five different subscriptions.

从一个提示开始，再加入你自己的创意。

Start with a prompt and add your personal touch.

你可以不用写代码或懂设计，就能创建网站、在线商店和自定义地图。

You can create websites, online stores, and custom maps without coding or designing skills.

再用AI代理自动处理繁琐任务，帮你扩大业务。

Then use AI agents to automate tedious tasks and grow your business.

Hostinger 支撑了上千万个网站，它能拿到CNET编辑推荐奖，绝不是偶然。

Hostinger powers over 10,000,000 websites, and there's a reason it earned a CNET Editor's Choice Award.

把你的‘某一天’，变成真正的‘第一天’。

Turn your one day into day one.

前往 hostinger.com/theprofg，每月不到三美元，就把你的想法上线。

Go to hostinger.com/theprofg to bring your ideas online for under $3 a month.

使用优惠码 The Prof G，还能再享八折优惠。

Plus, get an extra 20% off with promo code The Prof G.

这还不到一杯咖啡的钱。

That's less than the price of a cup of coffee per month.

记住，是 hostinger.com/the Prof G，优惠码 The Prof G，额外再打八折。

That's hostinger.com/the Prof G, promo code The Prof G for an extra 20% off.

本节目由Odoo赞助支持。

Support for this show comes from Odoo.

经营企业已经够难了，干嘛还非得用十几个互不相通的软件来给自己添堵？

Running a business is hard enough, so why make it harder with a dozen different apps that don't talk to each other?

介绍下Odoo。

Introducing Odoo.

这是你这辈子唯一需要的商业软件。

It's the only business software you'll ever need.

它是一个全功能、一体化的平台，能让你的工作轻松不少——客户管理、财务、库存、电商，样样都行。

It's an all in one fully integrated platform that makes your work easier CRM, accounting, inventory, ecommerce, and more.

最棒的是什么？

And the best part?

Odoo用不到几分之一的价格，就取代了多个昂贵的系统。

ODOO replaces multiple expensive platforms for a fraction of the cost.

所以，成千上万家企业都选择了换用Odoo。

That's why over thousands of businesses have made the switch.

那你为什么不来试试呢？

So why not you?

去 odoo.com 免费试用 Odoo 吧。

Try Odoo for free at odoo.com.

就是 odoo.com。

That's odoo.com.

本节目由 Odoo 赞助支持。

Support for this show comes from Odoo.

做生意已经够难了，干嘛还用一堆互不相通的软件给自己添堵？

Running a business is hard enough, so why make it harder with a dozen different apps that don't talk to each other?

介绍下 Odoo。

Introducing Odoo.

这是你这辈子唯一需要的商业软件。

It's the only business software you'll ever need.

它是一个一体化、完全集成的平台，能让你的工作轻松不少。

It's an all in one fully integrated platform that makes your work easier.

客户关系管理、会计、库存、电子商务，还有更多功能。

CRM, accounting, inventory, ecommerce, and more.

最棒的是什么？

And the best part?

Odoo 用不到原来几分之一的价格，就取代了多个昂贵的系统。

Odoo replaces multiple expensive platforms for a fraction of the cost.

所以才有成千上万的企业选择了切换到 Odoo。

That's why over thousands of businesses have made the switch.

那你为什么还不试试呢？

So why not you?

去 odoo.com 免费试用 Odoo。

Try Odoo for free at odoo.com.

就是 odoo.com。

That's odoo.com.

我们回来了，继续带来 Prof G Markets 的内容。

We're back with Prof G Markets.

所以，如果我们大家都认同这项技术正在让我们变笨，我觉得这已经很明显了。

So if we all agree that this technology is making us dumber, and it seems that it's I mean, I I think that.

目前我真想不出还有谁会不同意这一点。

I I I'm not I'm not sure who disagrees with that at this point.

对我们来说，这已经很清楚了。

I think it's pretty clear to us.

德里克，咱们来用博弈论分析一下吧。

I mean, Derek, let's, like, model this out, game theory it out.

从经济角度看，这会走向什么方向？

Where does this go in terms of the economy?

如果我们完全依赖人工智能，却再也提不出原创想法，也无法对问题进行批判性思考，你认为这会不会改变经济的发展轨迹？

I mean, if we are dependent on AI, but none of us can really come up with original ideas and we can't think critically about issues, do you think that that steers the trajectory of our economy in perhaps a different direction?

让我从一个非常宏观的角度来回答这个问题，然后再具体展开。

Let me try to take this question at a really high level of abstraction, and then I'm gonna zoom in on some specifics.

我觉得技术本身只是工具。

I think that technology is use.

人工智能的影响，完全取决于我们怎么使用它。

How the effect of AI is exquisitely dependent on how we use it.

你看，梅奥诊所最近用人工智能在放射科检测胰腺癌，平均比医生早两年半就能在扫描图里发现。你根本没法说这叫人工智能让人变笨。

If you look at how artificial intelligence was recently employed by the Mayo Clinic in radiology to see pancreatic cancer on average two point four years before a doctor could see it in a scan, You cannot possibly argue that that is AI making people dumber.

是的。

Yeah.

这明显让我们在识别胰腺癌方面变得更聪明、更有能力。

That is clearly making us better, smarter as a species at seeing pancreatic cancer.

在这里，科技和人工智能的作用是辅助放射科医生，帮他们更准确地发现胰腺癌。

The use of technology, the use of artificial intelligence there is to supplement the human radiologist's eye to see pancreatic cancer.

所以，我觉得这显然是好事。

So I don't and that is obviously good.

我并不想让人觉得我的观点是——所有人工智能都是坏的，也许卡尔也这么想。

So I don't want to represent my opinion here as being and maybe Cal agrees, as being like, oh, all AI is bad.

但人工智能在中学和大学里的用法，根本不是这么回事。

Like, But that's not the way that artificial intelligence is being used in high schools and college.

它被用来作弊，而且作弊的规模大到让学生失去了学习如何学习的能力。

It's being used to cheat and to cheat at a scale that is keeping students from learning how to learn.

我对这项技术在某些行业中的应用非常乐观，但与此同时，我认为卡尔说得对——如果看看人工智能在高中和大学里的使用方式，我对这一代学生毕业时能否学会思考、深入钻研、独立写作，几乎看不到任何希望。

So I am very optimistic about how this technology is being employed in some industries, while at the same time, I think Cal is absolutely right that if you look at the use of artificial intelligence in high school and college, I see practically no reason to be optimistic about that generation's ability to learn, to think deeply, to write by the time they graduate.

技术本身是中性的。

So technology is use.

人工智能确实有一些非常出色的应用场景。

There are some wonderful use cases of artificial intelligence.

但在今天的教育体系里，它基本上成了一种大规模作弊的工具，长远来看，这剥夺了学生独立思考的能力。

But within the education system today, like, I think it is basically a tool for mass cheating that is in fact cheating students out of the ability to think in the long run.

是的。

Yeah.

你提到了一个关键点，那就是我们得弄清楚，到底是谁因为人工智能变得更笨了。

You you bring up an important point here, which is we should probably distinguish who is getting dumber because of AI.

事实上，我们主要讨论的是孩子。

And the reality is we're we're mainly talking about children here.

我们说的是那些还在上学的人，甚至高中生，他们用人工智能来写作业、作弊。

We're talking about people who are in school or even high schoolers who are using AI to do their homework, to cheat.

正如你之前提到的，我们看到数学成绩在下滑，科学成绩在下滑，所有这些标准化考试的成绩都在下滑，连阅读能力也在下降。

And we're seeing, as you mentioned earlier, that math scores are going down, science scores are going down, all of these standardized test testing scores are going down, Even literacy rates are going down.

所以，我想我们大家可能都认同的一点是，人工智能已经彻底改变了上学的意义。

So, I mean, it sounds like maybe the the point on which we would all agree is that AI has fundamentally transformed what it means to go to school.

这一点尤其值得我们进一步深入探讨，展开更深入的讨论，甚至需要制定一些监管措施。

And that is the point that perhaps needs further and deeper exploration, deeper deeper discussion, and perhaps some regulation.

德里克，如果现在人人都在作弊，那我们该怎么办？

Derek, if this has meant that everyone cheats now, what do we do?

是的。

Yeah.

如果我要写一篇关于这个话题的杂志文章，我会这样来定位，我很认同卡尔的视角，所以我直接借用了他的说法。

If I was gonna write like a like a magazine piece about this, I think the the way that I would frame it and I really like Cal's framing, so I'm I'm borrowing this this from him.

但我想说，过去这二十年里，我们的学校一直在进行一场关于分心的实验。

But I would say that for the last ten to twenty years, we've been running this experiment of distraction in our schools.

我们有非常明确的相关性数据，我认为更有说服力的是因果证据，表明手机是导致全球数学成绩、阅读能力和其他注意力相关指标下滑的主要原因，这不只是美国的问题，而是全球性的。

Like, we have very clear correlative, but I think causal evidence that suggests that phones are an enormous distraction that's responsible for the global, not just US, but global decline in math scores, in literacy scores, and in other measures of one's capacity to maintain attention.

在这一巨大的干扰源之上，再叠加人工智能——这种能高效整合信息的工具，让学生们得以大规模作弊，这一点在大学和高中里已经显而易见。

Now on top of this weapon of mass distraction, you add artificial intelligence, which is this extraordinary tool for synthesizing information, which allows students to cheat at an extraordinary scale that we know is happening in colleges and and and high schools.

如果你想解决这个问题，想应对这种由干扰引发的作弊浪潮，就必须直接下手，从根本上处理。

If you wanna fix that, if you wanna fix this weapon of mass distraction followed by this weapon of mass cheating, you have to solve it directly.

把手机从教室里拿走，放进专用收纳袋里，做个实验，看看效果如何。

Take the phones out of out of the classrooms, put them in pouches, run that experiment, certainly, to see if it works.

在考核知识掌握程度时，必须放弃那些容易被作弊的测试方式，转而采用无法被作弊的评估方法。

And then when it comes to testing knowledge, you just have to move out of the modes of testing knowledge that can be cheated toward modes of testing knowledge that can't be cheated.

比如，牛津模式就挺不错，大部分成绩取决于课堂上的口头答辩。

So one thing that can't be cheated is something a little bit more like the Oxford model, where most of the grade is dependent on in class oral exams.

你有一种体系或者文化，比如说你上历史课。

You have this system or culture of you know, you take the history class.

你学习哈布斯堡王朝的历史。

You learn about the Habsburg empire.

与其写一篇关于哈布斯堡王朝的论文，还不如直接让杰克帮你写。

Rather than write an essay about the Habsburg empire, much more likely, just ask Jack to write it for you.

你站起来在全班面前讲哈布斯堡王朝，讲神圣罗马帝国，别人提问，你当场回应，用你的理解证明给老师和同学看。

You get up in front of the class and talk about the Habsburg empire and talk about the Holy Roman empire, and people ask you questions, and you defend and prove your intelligence to the classroom, to the teacher.

这就有点像我妻子刚完成她的临床心理学博士论文。

So it's a little bit like like my my wife just finished her PhD in clinical psychology.

博士毕业时，我们用什么词来形容这个过程？

At the end of the PhD, what's the verb that we use to describe the end of the PhD?

你要答辩你的论文。

You defend your dissertation.

你不是光把论文交给一群专家，说‘你们读完给我学位就行’。

You get up in front of a group of experts, and you don't just give them the paper and say, read it and then, you know, give me my degree.

你是要亲自答辩。

You defend it.

他们会向你提问。

They ask you questions.

他们会问：这个研究方法怎么解释？

They say, what about this methodology?

那图一又是怎么回事？

What about figure number one?

你得回答：我之所以用这个方法，是因为这样，而图一之所以长这样，也有它的原因。

And you say, oh, well, here's why I did the methodology, and here's why figure one looks the way it does.

你要当场证明，这篇论文确实是你写的，你真正理解自己做的工作。

You prove in real time that you are the author of that paper, that you understand the work that you did.

我觉得，要真正解决作弊泛滥的问题，教育方式很可能得采纳牛津那种模式，或者论文答辩这种形式，因为口头考试要作弊可难多了。

And I just think that more education, if we really wanna get around the cheating epidemic, probably has to slurp in this Oxford model or this or this dissertation model because it's much harder to cheat in an oral exam.

这个观点确实很有意思。

It's a really interesting point.

卡尔，你同意吗？

Cal, do you agree?

不同意。

No.

这肯定是对的。

That has to be right.

我的意思是，现在学术界就是这样。

I mean, this is what's happening in academia right now.

这是牛津模式和我一直提倡的理念的结合，那就是把培养专注思考复杂问题的能力作为教育的核心目标。

It's a combination of the Oxford model and what I've long been advocating for, which is the explicit discussion and promotion of the ability to aim your mind's eye towards complicated topics as the goal of school.

这种理念应该从中小学就开始推广，一直延续到大学阶段。

And it's something that we should be talking about starting at grammar school and moving all the way through the university system.

我们在这里不只是为了掌握知识、在考试中复述内容，而是要教会大脑享受思考的过程，这种视角能帮我们理解几乎所有活动。

That we are here not just to get content and reproduce content on test, but to teach our mind to be comfortable thinking, and that's a frame through which to see almost every activity we do.

我再补充一点。

I would also throw into this.

我觉得我们之前提到的‘具体性’是个非常重要的观点。

I think specificity is a really important point we made earlier.

所以我再加一个具体性的限制条件。

So I'm just gonna throw in a a sort of specificity constraint here.

如果我们用我的说法，我们真正讨论的其实是，‘人工智能’这个词用错了。

What we're really talking about, if we're gonna use my terminology, AI is the wrong term.

这范围太广了。

That's way too broad.

它甚至包括德里克提到的克利夫兰诊所或梅奥诊所那种模式。

That includes things like the Cleveland Clinic or Mayo Clinic model that Derek was talking about.

那种模式跟前沿AI公司推出的那种大型语言模型根本没关系。

That model, for example, has nothing to do with a large language model like the type you would see produced by the Frontier AI companies.

对吧？

Right?

这是一种专门在放射科影像数据集上训练出来的预测模型。

This is a prediction model that's custom trained on labeled data sets of radiology scans.

这种技术我们从九十年代就开始做了，一直稳步前进。

We've been doing this since the nineties and been making slow and steady progress.

这类实用性强的AI模型其实并不新鲜，也没有正在经历爆炸式的突破，但前沿AI公司常常把这类非大语言模型的成果包装一下，混进他们自己的东西里。

Like, these sort of AI models that are very utilitarian and useful aren't new, aren't currently experiencing a massive exponential takeoff and capabilities, but often the frontier AI companies will launder the results from these non LLN models and sort of mix in with what they're doing.

但我们真正讨论的是基于大型语言模型的工具，特别是用它们来生成书面文本，或者在某种程度上辅助思考。

But what we're really talking about here is large language model based tools, and in particular, using those for the production of written text or in some sense to sort of aid thinking.

而这正是我们之前一直在讨论的学术环境中出现所有问题的地方。

And that's exactly where we get to all the problems in the academic setting that we've been talking about.

从国家经济的层面来看，你认为这个问题有多严重？

How big of a problem do you consider this in terms of like a national economic scale?

因为，我的意思是，一方面我们想保护孩子。

Because, I mean, there's one side of this which is like, you know, we wanna protect our kids.

让孩子们在学校里拥有充实、有趣的经历，这很重要。

It's important that our kids have fulfilling, interesting school experiences.

他们能接受良好的教育等等，我相信我们都同意这一点。

They get a good education, etcetera, which I'm sure we all agree.

但另一方面，我们其实也需要孩子们拥有健全的头脑，毕竟他们将来要领导这个国家。

But then there's also another side to it, which is like, we kind of need children to have functioning brains for when they eventually lead the nation.

这里甚至可能牵扯到中美之间的竞争。

And there might even be, like, a China versus USA argument here.

比如，如果地球另一端的学生接受了良好的训练，他们的AI聊天机器人也得到了有效监管。

Like, if students over on the other side of the planet are being trained properly, their AI chatbots are being regulated properly.

他们懂得如何动脑思考。

They know how to use their brains.

那难道不会意味着，五十年后，他们在各项指标上都全面超越我们吗？

Doesn't that mean that sort of fifty years down the line, they're going to beat us and outperform us on every which metric?

我的意思是，卡尔，你觉得这个观点有道理、重要吗？

I mean, is that an argument that you see as relevant or important, Cal?

你在讨论这个问题的时候，会提到这一点吗？

Is that something that comes up in your conversations when you discuss this topic?

我对这个问题有个相当激进的看法。

Well, I have a relatively radical view on this.

我很想听听德里克的意见，毕竟他是这方面的经济学专家。

I'll be interested you know, Derek is the economics expert here, so he's interested in on his take on it.

但我认为，我们已经看到了这种认知能力下降带来的经济影响。

But I argue we have already seen the economic impact of this reduced cognitive fitness.

过去十到二十年，这已经是个非常明显的问题了。

This has already been a major storyline of the last ten to twenty years.

想想看，我们有这么多数字技术的进步，尤其是在办公数字化方面，总生产率理应大幅飙升，尤其是在知识型行业这类非工业领域，但事实并非如此。

I mean, given the technological advancements we've had in the digital the intersection of the digital and the office, we should be seeing exploding total factor productivity, especially in nonindustrial sectors like the knowledge sector, and we haven't.

对吧？

Right?

背后有各种因素在起作用。

There's been a lot of different things have been playing on that.

经济危机和其他问题当然也有影响，但非工业领域的总生产率一直停滞不前，或者起伏不定，远不如预期。

We have the economic crisis and other things going on, but a total factor productivity in non industrial sectors has been more flat or uneven than you would expect.

我认为，这在一定程度上是因为我们生活中和工作中，分心和频繁切换任务的情况已经严重加剧了。

And I would argue this is in part a result already of massively increasing the distractions and context switching that happens in our lives and in the workplace.

现在这个世界，有一个特别能说明问题的数据：普通员工平均每三分钟就要查看一次邮件或聊天窗口。

We're in a world now, I think one of the most telling statistics of the current office is now the average worker is gonna check an email inbox or chat channel once every three minutes on average.

在这种认知环境下，想靠大脑为信息创造价值——而这正是知识型工作的核心——根本是不可能的。

That is a disastrous cognitive context to use your brain to add value to information, which is the core activity of knowledge work.

所以我觉得，生产力停滞的局面已经出现了。

So I already think we're seeing a flat lining.

这有时被称为2000年代的生产力悖论，原因就在于它对认知能力的削弱。

This is sometimes called the productivity paradox of the 2,000, 2,000 tens because of this impact on cognitive fitness.

如果继续沿着这条路走下去，让基于大语言模型生成的写作取代教育体系中原本重要的写作训练，问题就不只是关乎孩子的大脑发育，或把‘聪明’简单等同于好结果了。

So, yes, if we go farther down this road and using LLM based produced writing, take that that important strengthening activity off the plate in our educational system, it's not just about kids' brains and some sort of abstract notion of smartness equals good.

我认为，过去十到二十年我们可能已经感受到的经济影响，只会变得更糟，这确实是我们从国家层面必须认真对待的问题。

I think the economic impact that we may already have been feeling for ten to twenty years is just gonna get way worse, and it's it is something we do have to really care about from a national perspective.

德里克，你对这些经济影响有什么看法？

Derek, what are you your views on these economic impacts?

对。

Yeah.

我听你和卡尔聊的时候，突然想到两个数字，放在一起挺有意思的。

You know, as as as I was listening to you and Cal talk, sort of these two two different statistics sort of pops in my head that I think juxtaposed together interestingly.

一个迹象是，Z世代可能是美国历史上物质主义最强的一代。

One is that there are a lot of indications that Gen Z is the most materialist generation that we've ever seen in American history.

如果你问不同世代群体，他们在美国家认为多少财富才算成功。

If you ask various groups, bucketed by generational cohort, how much money they consider success in America.

大多数世代都把十五万美元当作标准，但到了Z世代，他们说要四五十万美元。

You tend to have about a $150,000 be the norm in most generations until you get to Gen Z, and they say it's 4 to $500,000.

家庭研究学院最近分析了一项名为《未来追踪》的调查，该调查询问了高中生男女对物质主义的态度。

Institute for Family Studies recently looked at a monitoring the future survey that asked various questions about materialism among young boys and girls in high school.

这种物质主义倾向一直在持续攀升。

And that line of materialism has just gone up and up and up.

我认为，过去三十年来，第一次有更多女性在某项物质主义指标上超过了男性。

And I think for the first time in the last thirty years, women are now higher on a certain measure of materialism than men.

一方面，年轻人对成功有着前所未有的渴望。

So on the one hand, you have this extraordinary desire among young people to be successful.

他们一打开手机。

They open their phones.

就去看那些网红。

They look at influencers.

他们看到那些富有、成功、漂亮的人过着富足、成功、光鲜的生活。

They see rich, successful, beautiful people living their rich, successful, beautiful lives.

所以这是眼下正在发生的一条路径。

And so that that's one train track that's coming along here.

但还有一条并行的路径，那就是高中生和大学生们不断作弊。

But there's this other parallel train track, and that is students cheating constantly in high school and college.

短期来看，如果你每次考试都作弊，那你只是骗过了考试。

In the short run, if you cheat on every test, you're cheating the test.

但从长远看，如果你每次考试都作弊，你就是在骗自己。

In the long run, if you're cheating on every test, you are cheating yourself.

没错。

Yeah.

你是在剥夺自己承受压力、锻炼能力的机会。

You are removing from yourself the ability to lift the weight.

如果你想富有、想成功，我敢说，我认识的每一个真正富有又成功的人，无一不是付出超乎寻常的努力，都极其擅长在高强度认知状态下持续投入。

And if you want to be rich and if you wanna be successful, I myself certainly know of absolutely no individual who is rich and successful who doesn't work unbelievably hard, who isn't very good at what I think of as cognitive time under tension.

打个比方，如果你在卧推时只用一秒钟举起150磅，这种阻力是有限的。

That is to extend the fitness metaphor, this idea that if you do sort of one rep of a 150 pounds on, you know, a bench press and it takes one second, that's a certain amount of resistance.

但如果你把上推和下放的过程延长到五秒甚至十秒，肌肉承受的张力就会大得多。

But if you make that a five or even ten second up and down, it's much more tension on the muscles.

这就叫时间张力。

That's time under tension.

我觉得思考也是一样，真正了不起的想法，往往需要你花很长时间去琢磨，去找到那种简洁性——就像奥利弗·温德尔·霍姆斯说的，简洁就在山的另一边。

And I think thinking has a similar principle, that really great ideas benefit from the ability to sit with those ideas for a long period of time to figure them out, to find the simplicity that I think, as Oliver Wendell Holmes said, is on the other side of the mountain.

你先去学习一件事，然后通过理解它，把它简化、变得实用。

You learn about something, and then you are able, through your learning about it, to make it simple and make it effective.

如果你连这些考验都逃避了，那你等于是在剥夺自己变得富有和成功的能力。

If you are cheating yourself out of all these tests, you're cheating yourself out of the ability to become rich and successful.

我担心的不只是这些作弊的人会输给中国人、芬兰人或者丹麦人——也许会，也许不会。

And so one thing that I'm afraid of is not just that these people who are cheating are gonna lose out to the Chinese or whatever, the Finnish or the Danes, maybe they are, maybe they aren't.

他们真正会输给的是那些会思考、肯下功夫、能静下心来琢磨想法、真正培养出认知张力的人。

They're gonna lose out to people who can think, who are doing the work, who can sit with ideas, who do have and are building cognitive time under tension.

所以我觉得，如果一代人对物质成功抱有极高的期待，却缺乏真正实现成功的能力，那简直就是为他们铺好了一条通往巨大失望、焦虑和抑郁的路。

And so I just think that a world in which you have a generation of people with extraordinary expectations of material success, but underdeveloped abilities to actually achieve that success, that just seems like you're setting up a generation for unbelievable disappointment, anxiety, and depression.

这不仅仅关乎国家荣耀，比如美国和中国之间的竞争，虽然可能也沾点边。

So, you know, this goes, I think, not just to, you know, the concept of national greatness, US versus China, although maybe it touches on that.

更重要的是，我们到底想要什么样的人生？

It goes to, like, you know, what do we want from our life?

那些想变得富有和成功的人，他们真正该追求的是什么？

Like, what do people who wanna be rich and successful, what should they want from their life?

他们应该学会忍受不适，努力工作，享受复杂问题带来的挑战，热爱深入思考，因为真正的财富就藏在这些过程里。

They should want the ability to sit the the the ability to sit with discomfort, to work hard, to enjoy complicated problems, to love thinking through them because that's where your money is made.

如果你失去了这些，那就等于放弃了实现‘新美国梦’的能力。

If you lose that, you really lose out on this ability to achieve, like, what is the new American dream.

我之所以对经济角度这么感兴趣，是因为我觉得，反对我们观点的人往往会说，你们这是反技术、反进步的卢德主义。

I guess the reason that I'm so interested in the economics angle is because I feel like the argument against what we're saying is that it's sort of this Luddite argument that you're anti technology, anti progress.

但我觉得，德里克，你说到点上了：如果一代人从小时候就被训练成总想走捷径，不愿静下心来思考，不肯下苦功，只会不停地刷手机，活在一个虚幻的成功假象里，却从没真正培养出实现目标的能力，那最终，我们整个一代人、整个国家都会变成懒惰、不会思考、一事无成的群体，连基本的共识和决策都做不了，更别提建设什么了。

And I think the thing that really resonates for me to your point, Derek, is if you have a generation of people who have been trained since their infancy to take shortcuts, to not sit with ideas, to not work hard, to just scroll, scroll, and kind of like live this sort of fleeting imaginary version of success, and you never actually build the tools or the abilities to actually go out there and achieve it, then ultimately, we'll have an entire generation, an entire nation of basically lazy, non thinking losers who can't really get anything done, who can't really come to a consensus and make decisions and build things.