尊敬的用戶您好，這是來自FT中文網的溫馨提示：如您對更多FT中文網的內容感興趣，請在蘋果應用商店或谷歌應用市場搜尋「FT中文網」，下載FT中文網的官方應用。
Last year, as ChatGPT and other generative artificial intelligence chatbots were taking the world by storm, ASML — the Netherlands-based maker of chipmaking equipment — saw scope for hiring a “prompt engineer” to deploy the technology in its in-house legal department.
去年，隨著ChatGPT和其他生成式人工智慧聊天機器人風靡全球，荷蘭的晶片製造設備製造商阿斯麥(ASML)看到了在其內部法律部門中聘用一名「提示工程師」來部署該技術的機會。
In a post on LinkedIn, ASML’s deputy general counsel, Douwe Groenevelt, said he envisioned “a new potential role that could bridge the gap between AI and our legal team”. The position called for a candidate who could write generative AI prompts — the queries that must be fed into an AI tool to generate the desired output — and could train up colleagues.
阿斯麥的副總法律顧問道格•格倫費爾特(Douwe Groenevelt)在領英(LinkedIn)上的一篇貼文中表示，他設想了「一個可以彌合人工智慧和我們的法律團隊之間差距的新潛在角色」。該職位要求候選人能夠撰寫生成式人工智慧提示——必須輸入到人工智慧工具中以生成所需輸出的查詢，並能夠培訓同事。
The only catch: the job didn’t exist.
唯一的問題是：這個工作並不存在。
As Groenevelt has since explained, the intention behind that post, which attracted more than 200 likes and around 30 comments, was to contribute to the discussion about the future of legal roles in the age of generative AI.
正如格倫費爾特後來解釋的那樣，該貼文的目的是爲了對生成式人工智慧時代的法律角色的未來進行討論，該貼文獲得了200多個贊和大約30條評論。
Fast forward to this month, though, and the job has become a reality. ASML has announced a genuine vacancy in its in-house legal department for a first-of-its-kind legal prompt engineer. It has concluded that, with the rapid development of AI tools, a dedicated role is warranted.
然而，快進到本月，這項工作已經成爲現實。阿斯麥宣佈其內部法律部門有一個首個合法提示工程師的真實職位空缺。它得出結論，隨著人工智慧工具的快速發展，有必要設立一個專門的角色。
And the creation of this role is an indication of how the in-house legal profession is transforming itself to take advantage of generative AI, as companies around the world explore the risks and opportunities the technology presents.
而這一角色的設立表明，內部法律職業正在轉型，以利用生成式人工智慧的優勢，因爲全球各地的公司都在探索這項技術所帶來的風險和機遇。
Generative AI is a “game-changer” for the legal industry at large, says ASML’s chief legal counsel Sandrine Auffret.
阿斯麥首席法律顧問桑德琳•奧弗裏特(Sandrine Auffret)表示，生成式人工智慧對整個法律行業來說是一個「改變遊戲規則的」技術。
Lawyers have been using other types of AI for years, in the form of tools for contract management and e-discovery. But generative AI is different, Auffret says, as it “seems to be genuinely capable of transforming our legal service delivery model and having a much more substantial impact”.
多年來，律師一直在使用其他類型的人工智慧，例如合同管理和電子發現工具。但是，奧弗裏特表示，生成式人工智慧與衆不同，因爲它「似乎真正有能力改變我們的法律服務交付模式，併產生更大的影響力」。
The large language models used by generative AI are “extremely good at automating and enhancing any text-based activities,” she says, “and it so happens that many legal activities are to a large extent text-based”.
她說，生成式人工智慧所使用的大型語言模型「非常擅長自動化和增強任何基於文字的活動而許多法律活動在很大程度上都是基於文字的」。
ASML expects the technology to deliver substantial time savings for its in-house legal team — more than any other tool it has introduced, Auffret says. But it is also seeing an uptick in effectiveness and quality of service.
阿斯麥預計這項技術將爲其內部法律團隊帶來大量的時間節省，比引入的任何其他工具都要多，奧弗裏特說。但它也看到了效率和服務質量的提高。
Its team was one of the first to try the ContractMatrix tool launched by law firm A&O Shearman late last year, Groenevelt says. Now, about 15 of the group’s roughly 100 in-house lawyers are using the generative AI-powered tool to draft and review contracts, he notes.
格倫費爾特表示，該團隊是最早嘗試A&O Shearman律師事務所去年底推出的ContractMatrix工具之一。他指出，現在，該團隊約100名內部律師中的大約15人正在使用這款生成式人工智慧工具來起草和審查合同。
“Early feedback of the team shows that the tool does not only save us time with drafting, but . . . also increases the quality of drafting,” Groenevelt says. The tool can be asked to suggest alternative clauses, which also “boosts your legal creativity”, he adds.
「團隊的早期反饋顯示，這個工具不僅可以節省我們起草的時間，而且還提高了起草的質量，」格倫費爾特說。這個工具可以提供替代條款的建議，這也「增強了你的法律創造力」，他補充道。
Groenevelt says senior lawyers are getting more out of the tool than juniors, who may be more easily impressed with the initial results and less inclined to pressure-test them. He stresses that “human supervision remains essential”.
格倫費爾特表示，高級律師從這個工具中獲得的收益比初級律師更多，初級律師可能更容易對初步結果印象深刻，並且不太願意對其進行壓力測試。他強調，「人類監督仍然是必不可少的」。
Other company legal teams have focused on experimenting with generative AI internally, rather than hiring specialists.
其他公司的法律團隊更多地專注於在內部進行生成式人工智慧的實驗，而不是僱傭專家。
For example, Conduent — a business services company spun off from Xerox in 2016 — explored the possibility of hiring a dedicated AI position in its in-house legal department: a subject matter expert to review AI use within the department and across the business. But it quickly realised that this was the wrong approach for the group, says general counsel Michael Krawitz, because generative AI touched on many sensitive areas of the business.
例如，2016年從施樂(Xerox)分拆出來的商業服務公司康迪恩特(Conduent)曾探索在其內部法律部門中聘請一名專門的人工智慧職位的可能性：一位專家來審查部門內以及整個業務中的人工智慧使用情況。但是，總法律顧問邁克爾•克拉維茨(Michael Krawitz)表示，該集團很快意識到這種方法是錯誤的，因爲生成式人工智慧涉及到業務的許多敏感領域。
So, rather than hiring one individual, the company set up a working group of internal professionals with different specialities — such as intellectual property, regulatory compliance, privacy and risk — to explore how to apply the technology and appropriate safeguards.
因此，公司不是僱傭一個人，而是成立了一個由內部專業人員組成的工作小組，他們擁有不同的專長，如智慧財產、法規合規、私隱和風險等，來探索如何應用這項技術並採取適當的保障措施。
Like ASML, Conduent, which has about 50 in-house lawyers, has been experimenting with how generative AI can be used in its contract work.
與阿斯麥一樣，康迪恩特擁有約50名內部律師，一直在嘗試將生成式人工智慧應用於其合約員工作中。
“We went into [a contract tool] pilot thinking it would help us identify clauses that didn’t meet our standards, but the technology has not met our standards,” Krawitz reports. The pilot was marking contract clauses as being “compliant” when they were not, which was concerning, he explains.
克拉維茨報告說：「我們參與[一個合約員工具]試點時，認爲它會幫助我們識別不符合我們標準的條款，但是這項技術並未達到我們的標準。」他解釋說，試點項目將合同條款標記爲「合規」，但實際上並不合規，這令人擔憂。
Still, Krawitz says: “I remain hopeful that we’ll see that feature start to work well.” For now, though, it is “distracting and expensive”.
然而，克拉維茨表示：「我仍然希望我們能看到這個功能開始運行良好。」然而，目前來說，它是「分散注意力和昂貴的」。
However, Conduent’s experience of using a translation program powered by generative AI has been more positive, accelerating the review and approval of contracts when working with clients around the world.
然而，康迪恩特使用由生成式人工智慧驅動的翻譯程式的經驗更爲積極，在與全球客戶合作時加快了合同的審查和批准。
Krawitz can remember, back in 2022 when ChatGPT was rising to prevalence, some legal departments contemplating a ban on its use.
克拉維茨還記得，2022年ChatGPT開始流行時，一些法律部門考慮禁止其使用。
“I think that’s the wrong tone,” he argues. A better approach is “to show that we are working with generative AI, and we are doing it in a way that protects our ideas, our data and our people”, he says.
他爭辯道：「我認爲那種基調是錯誤的。」更好的方法是「展示我們正在使用生成式人工智慧，並且以一種保護我們的想法、數據和人員的方式進行。」
According to a report from PwC, there is also a financial incentive for legal professionals to embrace AI.
普華永道(PwC)的一份報告顯示，法律專業人士擁抱人工智慧也有經濟上的動機。
The professional services firm’s AI Jobs Barometer, for May 2024, found a wage premium of 49 per cent in the US and 27 per cent in the UK for lawyers equipped with AI skills.
根據專業服務公司的AI就業指數（2024年5月），在美國，掌握人工智慧技能的律師的工資溢價爲49%，在英國爲27%。
“This is indicative of the growing importance of AI proficiency within the legal sector,” says Sandeep Agrawal, a partner in legal business solutions at PwC UK. “The future of in-house legal teams will probably be characterised by a diverse blend of roles, with lawyers who have AI training working alongside engineers and computer scientists.”
「這表明在法律領域，AI技能的重要性日益成長，」普華永道英國(PwC UK)法律業務解決方案合夥人桑迪普•阿格拉瓦爾(Sandeep Agrawal)表示。「未來的內部法律團隊可能會由各種角色組成，擁有AI培訓的律師與工程師和電腦科學家共同工作。」
The future of in-house legal teams will probably be characterised by a diverse blend of roles, with lawyers who have AI training working alongside engineers and computer scientists
未來內部法律團隊的特點可能是多樣化的角色組合，接受過人工智慧培訓的律師將與工程師和電腦科學家共同工作。
But, right now, many in-house legal departments are just beginning their journey in exploring this technology, says Amy Yeung, the former deputy counsel of US student loan company Sallie Mae, who has held several senior and executive in-house legal positions.
但美國學生貸款公司Sallie Mae前副法律顧問、曾在公司內部擔任多個高級和行政法律職位的艾米•楊(Amy Yeung)表示，目前，許多公司內部法律部門纔剛剛開始探索這項技術。
Because generative AI relies on large amounts of data, it comes with risks, she explains. “For example, implicit biases present in data sets, or within the process, amplify through generative AI.” Yeung says a comprehensive understanding of data governance is essential in mitigating these risks.
她解釋說，由於生成式人工智慧依賴大量數據，因此存在風險。例如，數據集中存在的隱含偏見或者在生成式人工智慧過程中的偏見會被放大。楊表示，全面瞭解數據治理對於減輕這些風險至關重要。
However, as more in-house legal teams begin to embrace collaboration between lawyers and tech professionals, the benefits should become more widely recognised within companies.
然而，隨著越來越多的內部法律團隊開始在律師和技術專業人員之間加強合作，這些好處應該在公司內部得到更廣泛的認可。
“The integration of AI into in-house teams is not just reshaping the composition of these teams but also redefining the value they bring to the table,” says Agrawal.
「將人工智慧融入內部團隊不僅改變了這些團隊的組成，還重新定義了它們所帶來的價值，」阿格拉瓦爾說。