人工智能正在颠覆法律领域。这对我们其他人意味着什么？——与梅莉亚·拉塞尔对话

自从一名律师因使用虚假ChatGPT生成案例而被处罚后，人工智能与法律领域已取得长足进展。

AI and law has come a long way since a lawyer was caught using fake ChatGPT generated cases and was sanctioned.

这项技术正在颠覆整个领域，可能预示着我们将面临的未来。

The technology is upending the field and it may be a sign of what's to come for the rest of us.

相关内容稍后为您呈现。

That's coming up right after this.

事实上，人工智能安全就是身份安全。

The truth is AI security is identity security.

AI代理不仅仅是一段代码。

An AI agent isn't just a piece of code.

它是您数字生态系统中的一等公民，需要得到相应的对待。

It's a first class citizen in your digital ecosystem, and it needs to be treated like one.

这就是为什么Okta正在引领保护这些AI代理的潮流。

That's why Okta is taking the lead to secure these AI agents.

开启这层新防护的关键，是一套身份安全架构。

The key to unlocking this new layer of protection, an identity security fabric.

企业需要一种统一全面的方法，通过一致的政策和监督来保护每个身份——无论是人类还是机器。

Organizations need a unified comprehensive approach that protects every identity, human or machine, with consistent policies and oversight.

不要等到发生安全事件才意识到您的AI代理是一个巨大的盲点。

Don't wait for a security incident to realize your AI agents are a massive blind spot.

了解Okta的身份安全架构如何帮助您保护下一代身份，包括您的AI代理。

Learn how Okta's identity security fabric can help you secure the next generation of identities, including your AI agents.

访问okta.com。

Visit okta.com.

网址是okta.com。

That's okta.com.

Capital One的技术团队不仅在多代理AI领域高谈阔论。

Capital One's tech team isn't just talking about multi agentic AI.

他们已经部署了一个。

They already deployed one.

它被称为聊天礼宾服务，正在简化购车流程。

It's called chat concierge, and it's simplifying car shopping.

通过实时API检查的自我反思和分层推理，它不仅帮助买家找到心仪的车辆。

Using self reflection and layered reasoning with live API checks, it doesn't just help buyers find the car they love.

还能协助安排试驾、获得融资预批，并估算以旧换新的价值。

It helps schedule a test drive, get preapproved for financing, and estimate trade in value.

先进、直观且已投入实际应用。

Advanced, intuitive, and deployed.

这就是他们的技术堆栈。

That's how they stack.

这就是Capital One的科技实力。

That's technology at Capital One.

欢迎收听《大科技》播客，这是一档以冷静视角探讨科技界及更广阔领域细微话题的节目。

Welcome to Big Technology Podcast, a show for cool headed and nuanced conversation of the tech world and beyond.

今天，我们将讨论一个关于AI如何颠覆法律领域的精彩故事。

Today, we are gonna talk about a fascinating story about how AI is upending the legal field.

一个鲜为人知的故事，我私下听闻已久，但很高兴能公开与您探讨——今天我们请到了最合适的嘉宾共同展开这个话题。

A hidden story, one that I've been hearing about privately, but I'm excited to speak with you about publicly, and we have the perfect guest to do it with us today.

玛利亚·拉塞尔来了。

Malia Russell is here.

她是《商业内幕》的高级记者，负责报道AI如何改变法律领域。

She's a senior correspondent at Business Insider, and she covers how AI is changing the legal field.

玛利亚，欢迎你。

Malia, welcome.

谢谢邀请，亚历克斯。

Thanks for having me, Alex.

让我们回到2023年6月。

So let me take you back to June 2023.

当时有篇路透社的新闻标题。

There's a Reuters headline.

标题是'纽约律师因在诉讼摘要中使用ChatGPT案例而受到制裁'。

It says New York lawyer sanctioned for using Chatuchupti cases in a legal brief.

简单来说，这位律师向法官提交了论证，声称这些案例支持他的观点，但这些案例完全是ChatGPT编造的。

Basically, the lawyer put this argument out to a judge and said, these cases support my argument, but those cases were completely fabricated by ChatchiPT.

他受到了制裁，全世界都在说，如果律师使用生成式AI，他们就会编造事实，不值得信任。

He got sanctioned, and the whole world said, if lawyers use generative AI, they're going to include hallucinated facts, and you can't trust them.

但两年后，这个领域已经取得了长足进步

But two years later, the field has come very far since

那时。

then.

请带我们回顾过去几年，AI技术如何成为律师非常有用的工具。

So talk us through the last couple of years and how AI tech has become a very useful tool for lawyers.

那个案件让全美律师心生恐惧，成为了噩梦般的场景，我认为这很大程度上影响了公众对律师使用AI后果的看法。

That case struck fear in the hearts of lawyers across America and became the nightmare scenario that I think colored a lot of public perception of what happens when lawyers use AI.

现实情况是律师每天都在使用AI，虽然仍有问题发生，但并非每次都会登上路透社或商业内幕的头条。

The reality is lawyers are using it every day without ending up in Reuters headlines or Business Insider headlines, although it does still happen.

根据我读到的最新统计，已有超过600份诉讼文件被法官发现包含来自通用聊天机器人或其他工具的幻觉内容。

I think the latest count I read was like over 600 identified pleadings where lawyers had included content caught by a judge that contained a hallucination from a general purpose chatbot or other tool.

自2023年以来发生的变化是，我们迎来了为法律专业人士量身定制的应用程序大爆发。

What has shifted since 2023 is we have this Cambrian explosion of applications that are legally tuned for the legal professional.

这些工具能理解律师的工作内容。

These are tools that understand the work they do.

它们具有智能结构，知道在何时调取正确的文件。

They have structures so that they know the right documents to pull at the right time.

在这种情况下应该启用哪些模型才能获得最佳结果？

What are the models to turn on for the best outcome here?

我们如何在事务所内部保护保密规则？

How do we protect confidentiality rules within the firm?

现在有针对律师的Chat2PT、Claude或Gemini，但也涌现出一批全新的针对性解决方案，正在为律师带来实际成效和成果。

There is Chat2PT for lawyers or Claude or Gemini, but there's also a completely new wave of point solutions that is delivering real results and real outcomes for lawyers.

是啊。

Yeah.

我们经常讨论AI的投资回报将来自哪些方面。

And we talk often about where the ROI from AI is going to come from.

目前已经有几个领域是人们经常提及的。

And there have been a couple of areas that people have talked about.

编程显然是首当其冲的领域。

Coding is obviously the top one.

客户服务是另一个应用场景。

Customer service is another.

但问题在于，当你进入像法律这样高度监管的领域时，容不得错误、虚假案例或幻觉的情况会发生什么。

But one of the issues has been what happens when you get into a highly regulated field like law that can't afford mistakes, that can't afford fake cases, that can't afford hallucinations.

这期节目的起因是我和一位律师朋友在看足球赛时，他开始向我讲述他们事务所运用人工智能的各种方式。

And I was this episode came about because I was with a friend who's a lawyer at a football game, and he started telling me about all the different ways that his firm is using artificial intelligence.

这真的让我震惊，因为它清晰地展示了这项技术如何带来投资回报并改变整个领域。

And it really stunned me because it was a clear case of how this can have a return on investment and how this can change the field.

我想我在开头已经有所暗示——通过观察法律界的现状，你将预见到其他领域即将发生的变化。

And I think I sort of teased this in the beginning, but it is a way that if you look at what's happening in law, you're going to see what's going to happen everywhere else.

让我举个具体例子，我也很期待听到你是否了解更多案例。

Let me give you one example, and I'm curious to hear if you hear more.

显然，律师工作的重头戏是检索——查阅文件，寻找可能适用于辩护或起诉的相关案例。

So obviously, a big part of a lawyer's job is search, searching through documents, trying to find other cases that may or may not be applicable for defense or prosecution or whatever it might be.

在生成式AI出现之前，他们采用的方法是使用不同的术语和连接词进行搜索。

The way that they've done it before generative AI has been using different terms and connectors.

你需要搜索可能与案件相关的不同术语。

So you search for different terms that may be applicable for your case.

然后阅读部分内容，判断是否可以引用。

Then you read some and you see, okay, is this something I can cite or not?

这位律师告诉我，现在他们不再需要耗时费力地进行关键词检索，而是能将完整的复杂段落输入Westlaw等生成式AI工具，系统会自动返回关联性强、相关度高的重要研究成果，实质上是替他们完成了研究工作，极大减少了他们在这项工作上花费的时间。

What this person told me was instead of like basically doing this like keyword search, painstaking, time intensive, what they are able to do now is drop full sophisticated paragraphs into generative AI tools like Westlaw and it will return connected, relevant, and important research basically, by doing the research job for them effectively and fully minimizing the amount of time that they spend on this work.

这里的变革在于，过去律师们需要挤在会议室里翻找成箱的文件，就像寻找关键的'冒烟枪'证据一样。

And the change here is that whereas before you had lawyers going into conference rooms, digging through filing boxes of papers, looking for the firing gun, if you will.

现在他们使用的新工具可以上传案件所有相关文件、证词和访谈记录，不仅能搜索关键词，还能找出所有证明X事件发生的佐证材料，系统会自动呈现相关内容，可能为他们节省数周甚至数月的工作量。

Now, they're turning to tools where you upload all of your relevant documents and depositions and interviews on a case and you're able not to just search for keywords, but you're able to search for, you know, find all of the supporting evidence that X happened and it's going to turn up what is related and saves them potentially weeks, months of work.

没错。

Right.

顺便说一句，那都是很昂贵的工作。

That's expensive work, by the way.

我认为这实际上是在进行压缩。

I mean, what this what I think this is doing is it's compressing.

我是说，我们讨论过这个。

I mean, talked about it.

原本需要数周的工作，现在基本上可以压缩成一个提示词就能完成。

Weeks you could take weeks of work and basically condense it into a prompt.

这会对经济产生什么即时影响？

Like, what are the economic implications of that right away?

当然。

Sure.

我先给你讲点背景故事。

And I'm going to give you a little bit of backstory.

我报道法律科技领域已经有七个月了。

So I've been covering legal tech for about seven months now.

这源于我的主编在成为记者之前，曾是耶鲁法学院毕业生并担任过法律助理。

And it came about because my editor in chief at Business Insider, long before she was a journalist, was a Yale Law School graduate and a law clerk.

她的社交圈子里有很多律师。

And she ran in social circles with a lot of lawyers.

她不断听到的是对计时收费模式消亡的恐惧和偏执。

And what she kept hearing was this fear and paranoia around the death of the billable hour.

律师是按小时收费的。

Lawyers get paid by the hour.

当然也有例外情况，比如采用固定费用模式的工作。

There are definitely exceptions to that where the work is more flat fee based.

但如果是企业客户，通常还是按小时计费。

But if you're working with enterprise clients, you typically charge by the hour.

那些承诺节省时间的软件，在帮他们算账的同时，也可能会削减他们年底的收入。

And software that promises to save them time, that math does the math for them when the risk is that it cuts into their revenue at the end of the year.

通过与律师交谈，我发现他们仍然担心这会改变他们的商业模式。

What I found talking to lawyers is that there's still paranoia that this changes the business model for them.

这将促使他们考虑转向固定费用等替代商业模式。

It will nudge them to look at alternative business models like switching to flat fee.

我经常听到的另一种说法是，他们仍会按小时计费，但按小时收费的工作类型会发生变化。

Another one I hear a lot is that they're still going to charge by the hour, but the kinds of work they charge a billable hour for changes.

比如昆鹰律师事务所的高级合伙人，现在可能每小时收费1万美元，因为过去由初级律师处理的大量机械重复性工作，现在可以用生成式AI工具完成。

You, a top partner at Quinn Emanuel, might be billing $10,000 an hour because there's a lot of rote, repetitive work that your class of associates used to do that you're doing with a generative AI tool now.

这样他们就能把过去收费的部分工作剥离出去，转而将真正的溢价放在人类判断上——即在生成式AI完成初步工作后所需的人工判断环节。

So they can push aside some of the work they used to bill for and instead put the real premium on the human judgment that comes in after that kind of first pass work that happens with generative AI tools.

要知道，从宏观角度看，律师群体是出了名的技术保守派，这背后有几个原因。

You know, to take a step back, lawyers are notorious Luddites, and there's a few reasons for that.

首先，他们主要处理文档工作，长期以来，软件在从文本中提取含义方面表现并不出色。

First, they work largely in documents, and for a very long time, software wasn't especially good at extracting meaning from text.

此外，律师们还将文件存储在本地物理服务器上，以增强数据安全性和控制力。

You also had lawyers storing files on physical servers on premises for increased security and control over that data.

保密性就像是

Confidentiality is like the

在那个领域相当重要。

Pretty important in that field.

在该领域确实非常重要。

Pretty important in that field.

变化在于，ChatGPT向律师事务所展示了软件史上最伟大的演示——它能对白领专业人士带来怎样的改变。

What changed is that ChatGPT offered law firms the greatest demo in the history of software for what it could do for the white collar professional.

我认为很多律师在那个时刻后从屏幕上抬起头来说：哦，这将彻底改变我们的工作方式。

I think a lot of lawyers looked up from their screens after that moment and said, oh, this is going to transform us.

我们看到的情况是，客户现在主动找到律所表示：我们正见证AI如何提升内部效率。

What we saw was that the clients are now coming to the law firms and saying, we're seeing how AI is improving our efficiency internally.

你们的策略是什么？

What's your strategy?

这很重要，就像你之前说的，律师们要么抵制这种变化，要么坚持按小时计费。

Well, that's important because, like you said earlier, that lawyers have resisted this or they they bill by the hour.

我还听到另一种说法：客户不仅询问策略，而且正在强制要求。

And that's another thing that I've also heard is that the clients have said, not only what's your strategy, but they're demanding it.

他们要求了解所使用的工具明细，因为如果你按小时支付高额费用，而服务提供商却故意低效，你肯定会愤怒。

They want to see the breakdown of the tools that they use because, again, if you're paying this fee per hour and your service provider is being intentionally inefficient, you're going be pissed.

所以，好吧，也许按小时计费的方式会改变，但客户现在正找律师要求他们这么做。

And so, Okay, maybe the fee per hour changes, but the clients are now coming to the lawyers and saying, do this.

法律领域在这里可能别无选择。

So the law field might not even have a choice here.

可能直接就朝这个方向发展了。

It may just immediately go this way.

是的。

Yeah.

我认为律师们正在意识到，如果他们不采用这项技术，客户就会去找其他采用技术的律所。

I think lawyers are realizing if they don't adopt it, their clients will find other firms that do.

我们能暂停片刻，单纯欣赏一下这里的技术故事吗？

Can pause for a moment just to appreciate the technology story here?

当然，这项技术确实存在很多问题。

Now, look, there's definitely lots of problems with the technology.

但今天的人工智能已经足够优秀，以至于我们正在进行这场真实的讨论——它确实能将工作从数周压缩到几分钟或一小时，这相当了不起。

But the fact that AI today is good enough that we're having this discussion for real, like a for real discussion that it can condense the work down from weeks to a couple of minutes or an hour, it's pretty remarkable.

我是说，再想想这个人尽皆知的故事。

Mean, think about again this story that everybody heard.

2023年律师们还在使用虚构案例提交法庭而遭到制裁，如今技术已经进步到客户要求律师必须使用它的程度。

Twenty twenty three lawyers like using a hallucinated case and sending it to a judge and getting sanctioned to now the technology is good enough that it is something clients are demanding their lawyers use.

是啊。

Yeah.

我认为他们看到的是，一个变化在于通用工具已经足够胜任大量法律工作了。

I think they're seeing that I mean, one thing that's changing is that the basic, the general purpose tools are becoming good enough for a lot of legal work.

法律工作的核心就是检索和撰写文本，而AI...

The bread and butter of legal work is searching and writing text and AI.

这正是AI最完美的应用领域。

That is the perfect surface area to apply AI.

所以对客户来说这根本无需犹豫。

So it's kind of a no brainer on the part of the client.

而且他们看到自己的法律费用逐年膨胀。

And they're seeing their legal bills balloon year after year.

这是他们每年预算中较大的支出项目之一。

It's one of the bigger line items their budget every year.

所以他们表示想知道——我最近其实参加了一个法律会议，台上是一位七大科技公司（Mag-seven）的总法律顾问。

And so they're saying, we want to know I was actually at a legal conference recently, and on stage was the general counsel of a Mag-seven company.

有趣的是，他正在接受自己外聘律师的采访。

And interestingly, he was being interviewed by his own outside counsel.

是Meta公司吧？

It was Meta.

对吧？

Right?

我不能透露。

I can't say.

那是非公开谈话。

Was off the record.

好吧。

Okay.

或者是遵循查塔姆研究所规则。

Or it was Chatham House rules.

但现场有人提问，你预计明年能在法律费用上节省开支吗？

I But someone in the audience asked, do you expect you'll be saving money next year on your legal bills?

他回答说我当然希望如此。

And he said, I absolutely hope so.

而且你看，他的外部法律顾问主席就坐在他旁边。

And like, here, the chair of his outside counsel is sitting right next to him.

会后我和另一位七大巨头公司的总法律顾问交流，他说我们一个月前刚把所有外部顾问召集起来开了个峰会。

Did talk to, after his panel, another MAG-seven GC who said, you know, we brought in all of our outside councils for a summit about a month ago.

我站在他们面前明确表示：我需要知道你们正在使用什么工具、如何追踪效果，以及我能预期多少成本节约。

And I stood in front of them and I said, I want to know what tools you're using, how you're tracking it, and what kind of cost savings I can expect.

从技术角度看，你认为是什么关键因素让AI从几年前的胡言乱语进步到如今的高水平？

What do you think has helped on the technology standpoint to get the technology from the point where it was hallucinating a couple of years ago to now it's so good?

是否单纯因为OpenAI的GPT模型进化了所以减少了幻觉现象？

Is it just that OpenAI's GPT models have gotten better and hallucinate less?

还是说像Westlake（抱歉，是Westlaw）这样的公司决定采用这项技术，从而使其表现良好？

Or is it the fact that companies like Westlake or sorry, Westlaw have decided to implement it, and that's getting good?

你认为技术前沿正在发生什么变化？

What do you think is happening on the technology front?

在法律科技领域，有两家堪称行业恐龙的企业——LexisNexis和拥有Westlaw的汤森路透。

So in legal tech, you have two companies that are the dinosaurs of this industry, LexisNexis and Thomson Reuters, which makes Westlaw.

他们向律师事务所和总法律顾问推销的理念是：你应该购买我们的软件，因为在生成案例引用时，你可以将这些引用链接到我们的海量判例法数据库这一权威来源。

And the pitch they make to law firms and general counsels is that you should buy our software because when you're generating case citations, you can link those citations in a source of truth, which is our own vast repositories of case law.

因此他们的论点是：使用我们的产品能减少幻觉，因为我们拥有数据所有权。

So their argument is that use us because we're going to limit hallucinations because we own the data.

这促使众多法律科技挑战者提出：我们可以与数据提供商合作来解决你们的幻觉顾虑。

That has left a long trail of legal tech challengers to make the case that we can partner with the data providers to get around your hallucination concerns.

他们正在不断加强功能开发。

They're increasingly working on features.

想象一下类似拼写检查的功能——比如'我已经完成了这份案情摘要'。

Imagine like a spell check for, okay, I've completed this brief.

我希望AI能扫描并核实这些案例是否匹配真实案例。

I want the AI to scan it that these cases match real cases.

我的助理在这份简报中写的内容，与你认为原始案例文件意图的匹配度是否超过85%？

Does what my associate wrote here in this brief match more than 85% what you think the intent of the original case file says.

所以它不仅仅是检查案例是否存在，还能验证内容是否与判例法的实际表述相符？

So it's not just checking that it exists, but it can even check, does it match what the case law actually says?

因此技术发展非常迅速。

So the technology is moving really quickly.

我认为律师事务所对选择哪家供应商感到完全困惑，他们正在尝试各种方案。

Law firms are, I think, completely puzzled what is going to be the vendor of choice here, and they are trying everything.

LexisNexis和汤森路透就像不可谈判的巨头。

LexisNexis and Thomson Reuters are like non negotiators.

基本上，如果你是一家大型律所，就必须订阅其中一种或两种软件，因为这对他们从事的法律研究工作至关重要。

They pretty much if you're a big law firm, you have to have one or both of those software subscriptions because it's just like mission critical for the work that they do, that legal research.

但他们同时也在购买许多其他AI平台和专业解决方案的许可，进行全面测试。

But they are buying licenses to so many other AI platforms and point solutions, testing all of them.

这导致那些初创公司的收入虚增，估值被推高。

And that has this effect of inflating those startups' revenues and driving up their valuations.

我们稍后再深入讨论这一点。

And we can get into that in a bit.

对。

Right.

所以你的意思是，关键在于将技术建立在数据基础上，对吧？

So basically, what you're saying is the key here is grounding the technology in the data, right?

汤森路透拥有这些数据。

So Thomson Reuters has the data.

律商联讯也拥有这些数据。

LexisNexis has the data.

如果能将这些数据与模型连接起来，就能减少幻觉现象，使其真正有用。

And if you're able to connect that with the models, you'll have less hallucinations and make it useful.

是的。

Yeah.

今年法律科技领域的一大突破是Harvey宣布与LexisNexis建立战略合作伙伴关系。

And one of the BFDs in legal tech this year was Harvey announced a strategic partnership with LexisNexis.

Harvey是法律科技领域的巨头企业。

So Harvey is the goliath of the legal tech space.

他们开发的软件旨在帮助律师更快速、更经济、更高质量地提供法律服务。

They make software that helps, you know, that aims to help lawyers deliver legal services faster, more affordably, and at a higher quality.

而且他们是原生AI企业，彻头彻尾的AI原生。

And they're AI native, AI native.

由

Funded by

2023年就获得了OpenAI创业基金的首批投资，并提前获得了GPT-4的使用权限。

were one of the first checks out of OpenAI's startup fund back in 2023 and had early access to GPT-four.

他们今年宣布与LexisNexis达成战略合作。

They announced this year a strategic partnership with LexisNexis.

现在如果客户同时拥有LexisNexis和Harvey的订阅，就可以直接在Harvey应用内访问LexisNexis的内容。

So if a customer has a LexisNexis subscription and a Harvey subscription, they can now access LexisNexis within the Harvey app.

这有力地证明了我们的结果是可信的。

It made a very strong case that you can trust our results.

我要说的是，律师们仍在告诫初级律师必须核查工作成果。

I will say that lawyers are still hitting home to their junior lawyers, you need to check the work.

我觉得这些制裁案件最耐人寻味的是，律师们正因未核查AI的工作成果而承担个人责任，这或许本应是理所当然的事。

I think what's so interesting about some of the sanction cases we've seen is that the lawyers are taking personal accountability for not checking the AI's work, and that kind of feels maybe an obvious thing to do.

但今年美国十大律所之一的瑞生律师事务所就发生了这样的事。

But, you know, there's a firm this year, one of the top 10 law firms in The United States, Latham and Watkins.

今年早些时候，他们代表Anthropic处理一起版权诉讼案。

And earlier this year, they were repping Anthropic in a copyright lawsuit.

其中一位助理承认，他们使用Claude生成了专家证词中的部分引文，这些证词被提交用于这起版权案件。

And one of the associates said that they had used Claude to help generate citations within an expert testimony that they submitted for part of this copyright case.

而Claude在其中一处引文中虚构了论文作者和标题。

And Claude had hallucinated the authors and title of a paper in one of the citations.

这颇具讽刺意味——代表Anthropic的律所使用了Anthropic的产品，结果却制造出虚假案例。

So it's very meta in that you have a law firm representing Anthropic that used Anthropic's product and produced a hallucinated case.

我的意思是，这对所有相关方来说都非常尴尬。

I mean, it was hugely embarrassing for all parties involved.

这正是当前让大型律所感到恐慌的原因。

And that is the terror striking big law right now.

我认为律所正在重新思考他们培训律师的方式。

I think firms are really rethinking the way they train lawyers around this.

这要从入职培训开始。

And it starts at the onboarding.

我知道Latham and Watkins最近为整个一年级助理律师推出了AI学院，他们将400名律师召集到华盛顿特区的会议厅。

I know Latham and Watkins rolled out an AI academy recently for their entire first year associate class, where they brought 400 lawyers into a conference hall in Washington, D.

特区。

C.

在为期两天的活动中，他们观看了产品演示，听取了客户总法律顾问的分享，并参与了关于律所合伙人如何使用AI的讨论会。

And for two days, they sat through product demos and they heard from general counsels at their clients and they did listening sessions on how partners at the firm were using AI.

贯穿整个演讲的核心思想是：我们Latham and Watkins有着极高的标准，最终成果仍然是你们的责任，你们必须保持掌控权。

And the through line in this talk was that we have exceptional standards here at Latham and Watkins and the final product is still your responsibility and you've got to keep your hands on the wheel.

是啊。

Yeah.

不，这是个有趣的观点，某种程度上让我想起朋友告诉我他正在使用的那些工具——它们会出错，但同时也能帮律师完成部分工作。

No, it's an interesting point and sort of brings up another thing that my friend told me about the tools that he's been using, that they get things wrong but they all they also will get him or can get a lawyer a certain portion of the way there.

就像我之前说的，它做了所有研究，检索了相关案例，却得出了错误结论。

And you know again like it's doing all this research is pulling the relevant cases and coming to the wrong conclusion.

但如果是个经验丰富的律师，复查这些工作成果，实际上反而节省了时间。

But if you're a lawyer who has some seasoning and double checks the work, that actually that becomes the time save.

所以你不必完全相信它的输出，但确实可以依赖它完成大量基础工作。

So you don't necessarily trust it for the output, but you do rely on it for a lot of the work that it does.

关于法律AI，我听到的一个有趣问题是：谁最能从中受益？

One of the interesting questions that I hear around legal AI is who gets the most use out of it?

是初入行的律师还是资深合伙人？

Is it the early career lawyer or a more senior partner?

退一步说，初级律师的工作与生成式AI工具最擅长的领域最为相似。

And to take a step back, the work that an early career lawyer is most similar to what the Gen AI tools are very good at.

它负责撰写那些引文并进行核对。

It's writing those citations and checking them.

它负责起草备忘录的初稿。

It's taking first passes at writing memos.

它负责在数字档案箱中筛选，试图找到关键证据。

It's sifting through digital filing boxes trying to find that firing gun.

所以，是的，他们可以利用AI来加速这部分工作。

So there's, yes, they can use AI to speed up that work.

但真正有趣的是，我听说经验更丰富的律师实际上从使用这些工具中获得了巨大价值，因为他们能更快识别出哪些内容不达标。

But what's really interesting is that I hear the more seasoned attorneys are actually getting huge value out of using these tools because they can recognize more quickly looking at it what 's not up to snuff.

比如，他们具备辨别真伪的能力，可以说‘好吧，我只需要用生成式AI过一遍初稿，但我能更快分辨真假’，这让他们能更快得到更好的结果。

Like, they have a BS radar that might allow them to say, Okay, I just want to do a first pass with the Gen AI, but I can recognize more quickly what's real and what's not and allows them to just get to a better result faster.

是的，AI的使用方式非常有趣，因为工程领域一直存在这个问题，对吧？

Yeah, it's very interesting in terms of the way that AI is used because there's been this question think about engineering, right?

它是让每个程序员都变成10倍效率的程序员，还是让10倍效率的程序员变成100倍效率的程序员？

Does it make every coder a 10x coder, or does it make a 10x coder a 100x coder, right?

这是否会让超级明星变得遥不可及，因为他们掌握了这个强大的工具并真正懂得如何运用它？

Does it make the superstars sort of unreachable because it puts this powerful tool at their hands and they really know how to use it?

比如在写作领域就一直存在这样的争论。

And there's been this debate like for writing for instance.

AI似乎只能产出中等水平的写作，这很棒因为它把很多作者提升到了平均水平——对许多作者来说已经是巨大进步了。

Like you know AI seems to just produce average writing, which is great because it brings up a lot of writers to average, which is a massive improvement for many writers.

但另一方面是它究竟能如何帮助专家？

The other side of it is what can it really help the experts?

特别是在法律领域，那些经验丰富的从业者似乎能从中获益良多，以前所未有的方式最大化他们的工作效率。

But in law in particular, it seems like those those that have experienced or seasoned are going to get a lot from this and really be able to maximize their productivity in a way they couldn't previously.

我认为确实如此。

I think that's right.

这个问题最近又被提出来了。

And this came into question recently.

几个月前法律科技界曾有过一个病毒式传播的爆点事件。

There was kind of a super viral moment in legal tech a couple of months ago.

在面向法律科技爱好者的这个Reddit子版块上，曾有一篇匿名帖子提问：'真的有人在用这些工具吗？'

There was an anonymous post on this subreddit for legal tech aficionados that asked, Is anyone really using these tools?

发帖人自称是Harvey公司的前员工——虽然无法核实其真实性

And this person claimed to be a former employee of Harvey, which who knows if that's really the case.

他们暗示只有初级律师愿意使用Harvey，因为按照固有印象，年轻律师通常更懂技术，且个人生活中就在使用类似工具，所以接受度更高

But they were suggesting that only junior lawyers wanted to use Harvey because you figure, you know, the tropes that the younger lawyers are going to be more tech savvy and more they're using these tools personally, so it comes more naturally to them.

但Harvey公司坚决反驳称，他们观察到各年龄段律师都有很高的使用率和参与度，并且年度周活跃用户数据呈现大幅增长

But Harvey has certainly pushed back that they are seeing high adoption and engagement kind of across generations of lawyers and that they're actually seeing huge gains in usage, weekly active usage, year over year.

让我们最后再讨论下投资回报率的问题

So let's end this section just talking again about the ROI question.

法律行业的有趣之处在于，你很难用传统ROI公式计算——比如你可能每月只为ChatGPT支付20美元，但使用Westlaw的话显然不止这个价

It's interesting with law because you can't really, like, back it into a traditional ROI calculation because you could be paying, like, dollars 20 a month for ChatGPT, but probably you're not if you're using Westlaw.

虽然需要支付订阅费，但它能大幅减少你的工作量，从而腾出更多可计费时间

But you could be paying, let's say, a subscription, but that brings your work down by such a dramatic amount that you get to bill us.

这就形成了看似负ROI的局面——即便它确实提升了你的工作效率

So like you actually have a negative ROI even as it's making you more productive.

但我其实很好奇想听听你的观点，这些公司将如何进行投资回报率计算，以及我们是否会看到这种模式不断重复——技术以令人印象深刻的方式运作，但对于行业尤其是企业而言，它彻底改变了他们长期的业务方式，导致在推广应用中会遇到各种阻碍。

But I think that I actually would be curious to hear your perspective on how these companies are going to do the ROI calculation and whether we're just going to see this type of pattern repeat itself where the technology works in very impressive ways but for industries and enterprises in particular it changes the way that they've done business for so long that there will be just these speed bumps to try to get it working and rolled out.

嗯，我时不时听到的一种说法是，现在更难衡量那些在生成式AI出现前根本无法完成的工作。

Well, something I hear from time to time is that it's harder to measure the work you could now do that wasn't possible before Gen AI.

所以没错，也许你确实看到了可计费小时的减少。

So yes, maybe you do see a reduction in billable hours.

但律师们告诉我，由于AI帮助他们提高效率，现在能更深入地研究案件了。

But lawyers are telling me that they're able to dig deeper on cases because AI helps them be more efficient.

这种现象甚至发生在小型和中型律师事务所层面。

And this is happening even at the really small and mid sized law firm level.

我采访过一位独立执业律师。

I talked to a solo practitioner.

他是圣地亚哥的一位民权律师。

He's a civil rights lawyer in San Diego.

他曾代理过一个家庭案件，该家庭的孩子们曾被非法拘留在大学里。

He repped a family where the children had been unlawfully detained at The U.

美墨边境。

S.-Mexico border.

他向我讲述了如何利用工具节省生成引用或总结证词的时间。

And he spoke to me about how using tools, he was able to save time on generating citations or summarizing depositions.

他可以利用这些时间更好地思考法庭策略，花更多时间跨境与家属会面，从他们那里获取更多信息。

And he could use that time to think through his strategy better for the courtroom, to spend more time crossing the border, to go meet with the family and glean more from them.

这种无需埋头翻阅文件箱就能完成更多工作的理念，我认为将成为律所投资回报率计算的一部分。

This idea that there's so much more work to be done when you aren't just sifting through boxes, I think will be part of that ROI calculus for firms.

我确实认为我们将会看到法律领域的变化同样发生在会计和咨询行业。

I do think we're going to see what's happening in legal happen in accounting, happen in consulting.

法律领域的变革似乎来得特别快，虽然我不确定具体原因。

It seems to be happening in legal very quickly because I guess I'm not sure why exactly.

不妨猜猜看。

I mean, have a guess.

无非是因为它处理的就是文字，对吧？

It's just that it's words, right?

这是文字工作。

It's words.

会计是数字工作。

Accounting is numbers.

这些模型处理起来更难，但文字是语义搜索。

Harder for these models to do, but it's words, semantic search.

这就像是生成式AI唾手可得的果实，而咨询等其他领域可能需要工具辅助。

Like, it's sort of it's the low hanging fruit for generative AI, where the others will I think consulting, you know, it probably requires like some tool use.

需要编写代码进行计算，会计也是如此。

You know, you go in, you do calculations with code, same with accounting.

但这些技术在各类专业服务任务上正变得越来越强。

But these things are just getting better and better at all of these professional services tasks.

如果当前法律界的趋势能说明什么，那就是未来几年（甚至现在）许多行业都将展开类似讨论。

And if what we're seeing now in legal is any sign that this is a conversation I think many industries will be having over the next couple of years, if not having now.

更乐观地看，或许律所会放弃计时收费模式，转向更多服务的固定收费制。

And also, a more optimistic spin on this is that, sure, maybe firms slough off the billable hour model and they switch to flat fee billing for more services.

但他们也可能通过节省人力成本来获益，如果开始用软件工具替代律师，因为更好的软件工具能完成更多工作。

But they're also potentially going to save on labor costs if they start to displace lawyers because they can do more with the better software tooling.

未来我们可能会看到由更精简团队运营的、利润更高的律师事务所。

We might be looking at more profitable law firms in the future that run on leaner teams.

这就像我们在科技行业看到的讨论，那种一人独角兽公司的概念。

And, you know, this is like I feel like we see this conversation playing out in tech, the idea of the, you know, one person unicorn company.

我认为这种模式也将在法律领域得到呼应。

And I think we're going to see that model echo in law as well.

好的。

All right.

我想休息一下，因为，嗯，我们不得不休息。

I want to take a break because well, because we have to.

不过等我们回来后，我确实想谈谈就业问题。

But also, when we come back, I do want to talk about the employment question.

还想多聊聊个体律师如何获得赋能，以及使用这些工具的客户可能会如何改变法律行业。

Want to talk a little bit more about how individual lawyers are empowered, and then also how the clients using this stuff might change things law.

所以这个话题我们稍后继续。

So that's coming up right after this.

Capital One的技术团队不仅仅在讨论多模态AI。

Capital One's tech team isn't just talking about multiegetic AI.

他们已经部署了一个。

They already deployed one.

它叫做聊天礼宾服务，正在简化购车流程。

It's called chat concierge, and it's simplifying car shopping.

通过自我反思和分层推理结合实时API检查，它不仅帮助买家找到心仪的车辆。

Using self reflection and layered reasoning with live API checks, It doesn't just help buyers find a car they love.

还能协助安排试驾、获得融资预批以及估算置换价值。

It helps schedule a test drive, get preapproved for financing, and estimate trade in value.

先进、直观且已投入使用。

Advanced, intuitive, and deployed.

这就是他们的技术实力。

That's how they stack.

这就是Capital One的科技实力。

That's technology at Capital One.

如今，似乎每一分钱都应该更努力地工作，但弄清楚把钱放在哪里可能会让人困惑。

These days, it feels like every dollar should be working a little harder, but figuring out where to put your cash can be confusing.

这正是Wealthfront发挥作用的地方。

That's where Wealthfront comes in.

Wealthfront是一个以科技驱动的金融平台，旨在帮助你将储蓄增长为长期财富。

Wealthfront is a tech driven financial platform built to help you grow your savings into long term wealth.

截至2025年11月7日，他们通过合作银行提供的高收益现金账户提供3.5%的年化收益率（APY）。

Their high yield cash account through program banks offers a 3.5% APY on your uninvested cash as of 11/07/2025.

而且没有月费，没有最低或最高余额要求即可获得该利率，甚至可以在几分钟内24/7随时向符合条件的账户进行免费即时提款，让你的资金触手可及。

And there are no monthly fees, no minimum or maximum balance to earn that rate, and you can even make free instant withdrawals to eligible accounts in just minutes, twenty four seven, so your money can always be within reach.

目前，Wealthfront为新客户提供额外0.65%的年化收益率优惠，在最高15万美元的余额上享受三个月的基础利率加成。

Right now, Wealthfront is offering new clients an extra point 65% APY over the base rate for three months on up to a $150,000 balance.

当你首次开通现金账户时，总计可获得4.15%的浮动年化收益率。

That's a total of 4.15% variable APY when you open your first cash account.

立即访问wealthfront.com/bigtech进行注册。

Go to wealthfront.com/bigtech to sign up today.

这是Wealthfront的付费代言。

This is a paid testimonial for Wealthfront.

内容不代表他人体验，也不保证未来表现或成功。

It may not reflect the experience of others, and there's no guarantee of future performance or success.

Wealthfront经纪服务并非大额利率。

Wealthfront brokerage is not a big rate.

利率可能随时变动。

It's subject to change.

促销条款与条件适用。

Promo terms and conditions apply.

更多信息请查看节目描述。

For more information, see the episode description.

欢迎回到Big Technology播客，我们继续与Business Insider高级记者Malia Russell探讨AI如何改变法律领域。

And we're back here on Big Technology podcast with Malia Russell, senior correspondent at Business Insider, talking about how AI is changing law.

让我问你这个问题。

Let me ask you this.

我是WebMD的粉丝。

So I'm a fan of WebMD.

虽然不是最好的网站，但它确实能让我在感觉不舒服时去诊所，并说出我认为自己可能存在的问题。

Not the best website, but it does definitely allow me to go to the doctor's office when I'm feeling bad and say, I think that this might be the problem with me.

有一次我搜索喉咙发痒，它竟然建议我可能得了埃博拉病毒，这显然不准确，特此声明。

Now, I did one search about scratchy throat and it suggested that I had Ebola, which was inaccurate, just for the record.

它可以

It That's can

好消息。

good news.

随着生成式AI的出现，这种自我诊断对我来说变得更加频繁了。

It's even become with generative AI, that exercise has become even more intense for me.

我一定会和医生谈谈。

I definitely will speak with Doctor.

在去看医生前，我会先查查资料。

Chachapiti before I go into the doctor.

然后我会复述我认为的问题所在。

And then I'll recite sort of what I think is going on.

医生会说：‘你是世界上最烦人的病人，但你可能说对了。’

And the doctor says, You're the most annoying patient in the world, but you might be on to something.

所以我很好奇想听听你的看法：生成式AI如何改变律师与客户关系？也许客户会带着自己写的案情摘要来，或通过生成引擎做些研究，向律师提供类似案例或想法。这是否会让咨询律师变得更像顾问模式，而非现在常见的‘每分钟15美元，你坐下听我说’的情况？

And so I'm curious to hear your perspective about how Generve AI might change the lawyer client relationship, where maybe the client will come in and have written the brief or can do some of their own research with these generative engines and present similar cases or ideas to lawyers and whether it becomes more of a consultative type of approach when you go to a lawyer versus what typically happens now where the lawyer says, all right, dollars 15 a minute, and why don't you sit down and listen?

是的。

Yeah.

我认为我们会看到类似WebMD的情况。

I do think we're going to see the WebMD thing play out.

我们, 我们已经在经历这种变化了。

We already are.

这更多发生在民事法律层面，比如不太可能有财富500强公司拿着自己的研究成果去顶尖律所，但确实听说有人带着ChatGPT生成的意见来处理个人伤害或劳动纠纷案件。

This is happening more at the, I guess, civil law layer, where you're not going to probably have Fortune 500 companies coming into an AmLaw 10 firm saying, like, I did my homework on this and this is what I think we But should I hear stories that, you know, people bringing personal injury suits, for example, or employment suits are coming in with their own check GPT approved ideals and how they think this should go.

这太神奇了。

It's amazing.

你知道，我认为

You know, And I think

然后你会找真正的律师验证，对吧？

then you like vet it with a real lawyer, right?

然后他们可以给出肯定或否定的答复，而不是直接扔进法律系统。

And then they can sort of say yes or no versus just like throw it into the legal system.

是的。

Yeah.

而且我认为存在一个问题：这是否反而给律师增加了工作，需要解开当事人对案件走向的固有观念？

And I think there's a question like, okay, did this just create more work for the lawyer having to like untangle that person's preconceived notions about how things should go?

我认为更有趣的转变将是更多自助式法律服务的出现。

I think what we're gonna what's gonna be a more interesting shift is the arrival of more self-service legal services.

第一波法律科技公司开发软件卖给律所和法务部门，帮助他们更快更经济地提供法律服务。

So the first wave of legal tech companies were developing software to sell to law firms and general counsels to help them deliver legal services faster and more affordably.

新一波浪潮是公司直接从事法律工作本身。

The newer wave is companies doing the legal work itself.

他们要么是从大律所独立出来开设个人事务所的律师。

They are either attorneys spinning off from big law to open their own solo shops.

也可能是硅谷企业家们表示：'我认为我有比专攻主服务协议或保密协议的法律事务所更好的替代方案'。

They might be Silicon Valley entrepreneurs saying like, I think I have a better alternative to a legal shop that specializes in master service agreements or NDAs.

所以我们看到这些公司如雨后春笋般直接为消费者和企业提供法律服务，而且完全不按小时计费。

So we're seeing this surge of companies provide legal service directly to consumers and companies, and they're not doing the billable hour at all.

我的意思是，那样做就违背了他们的理念。

I mean, that would be kind of against their ethos.

他们更像是，你知道的，一个消费者友好型品牌。

They're like almost like we're a, you know, a consumer friendly brand.

我认为随着人们说'我不必去律师事务所'，这种模式将开始蚕食传统法律行业。

I think we will see that model start to like eat away at traditional law as people say, I don't have to go to a law office.

我可以，你知道的，通过Zoom联系我那位有人工智能辅助的律师，更快地完成工作。

I can, you know, Zoom my lawyer who's AI assisted and get it done faster.

我能就这个话题稍微跑个题吗？

Can I share a quick tangent on this?

有

Have

你自己做过这种事吗？

you done this yourself?

这就是你要告诉我们的吗？

Is that what you're going tell

我们？

us?

不是。

No.

但有个很有趣的现象正在婚前协议领域发生。

But a really funny thing that's happening is this is happening in prenups.

我刚写了一篇关于婚前协议的故事。

Just wrote a story about this in prenuptial agreements.

今年早些时候，我曾想报道一个关于几家科技公司的故事——抱歉，是有几家提供在线婚前协议服务的科技公司，你只需访问网站填写问卷即可。

Earlier this year, I'd wanted to do a story on there's a couple of tech companies that sell Sorry, there's a couple of tech companies that offer online prenups, and you go to a website and you fill out a questionnaire.

这和使用TurboTax报税没什么不同。

And it's not unlike using a TurboTax to file your taxes.

就像填写资产问卷一样，然后，砰的一下，几小时或几天后就会生成一份婚前协议，如果你付费升级，可能还会经过律师审阅。

You're, like, filling out the questionnaire with all your assets and then, you know, poof bada bing, it returns in a number of hours or days a prenup that's maybe been lawyer reviewed if you pay for the add on.

婚前协议在美国呈上升趋势，每年生成的婚前协议中约10%来自Hello Prenup、First或Neptune这类科技公司，它们为夫妻提供直接面向消费者的婚前协议服务。

And prenups are rising in this country and something like 10% of the prenups generated annually are now coming through these tech companies like Hello Prenup or First or Neptune that serve couples kind of like a d to c prenup.

这只是一个小众例子，但我感觉这已经在改变客户使用法律服务的方式了。

So that's just like one niche example, but I feel like that's already changing how clients use legal services.

如果按个播放键就能生成法律文件，写个提示就能生成诉讼或婚前协议这么简单的话——

So if it's that easy to just like press play and generate a legal document, write a prompt and generate a lawsuit or prenup.

这就引发了一个问题：这对律师就业会产生什么影响？

There's been this question about what this is going to do for employment of lawyers.

我很认同这个观点：它只会催生更多需求，因为现在发起法律行动太容易了，完全没有门槛。是的。

And I really like the argument that it's just going to necessitate more because it's so easy now to there's no barrier to initiate legal action Yeah.

你将需要更多受过培训的人来处理这些事情。

That you're just going to need more people that are trained to handle this stuff.

我喜欢这个论点。

I like the argument.

虽然不喜欢会有更多诉讼的事实，但就预测未来走向而言，这是个相当有力的论点。

Don't like the fact that we're to have more lawsuits, but I think that's a pretty good argument in terms of anticipating what's going to happen from here.

这就像个棘手的话题，因为

It's like a sticky conversation because

那就讨论吧，我还有些法律相关的事情想说，不过先听听这个。

Well, let's have it because I have some other stuff that I want to talk about on legal, but let's hear this.

我不想惹麻烦，但感觉现在很多公司都在推销'我们正在使法律服务民主化'的理念。

I don't want to get in trouble, but I feel like a lot of companies right now are selling a pitch around we're democratizing access to legal services.

是的，我认为对此的批评是这会导致更多无谓案件堵塞我们的司法系统。

And yeah, I think the criticism of it is that it's going to result in many more frivolous cases jamming up our justice system.

反正现在司法系统运转得这么好。

Which works so well as it is.