尊敬的用户您好，这是来自FT中文网的温馨提示：如您对更多FT中文网的内容感兴趣，请在苹果应用商店或谷歌应用市场搜索“FT中文网”，下载FT中文网的官方应用。
They consume vast tracts of content, cost a packet to train and graft well past normal office hours. Junior lawyers have much in common with generative artificial intelligence. Galling, then, for the former to face pay stasis — Slaughter and May is freezing their salaries at £150,000 for now — while more spending is being thrown at AI.
初级律师需要大量阅读各类内容，培训成本高昂，还常常加班到深夜。他们与生成式人工智能有许多相似之处。令人郁闷的是，初级律师却面临薪资停滞——司力达律师事务所(Slaughter and May)目前将他们的薪资冻结在15万英镑——而与此同时，更多资金却被投入到人工智能上。
Expect the machines to continue shouldering more of the workload. Fusty image notwithstanding, lawyers have been deploying tech for nearly a century: dictaphones in the 1950s and two decades later the clunky red UBIQ that enabled case law search without recourse to libraries.
预计机器将继续承担更多的工作。尽管律师行业形象古板，但实际上，律师们近一个世纪以来一直在应用技术：20世纪50年代使用口述录音机，二十年后又有了笨重的红色UBIQ，使得无需去图书馆也能检索判例法。
Today tech is corralled to zip through documents, conduct due diligence, summarise cases and even draft simple ones. It can handle matters like conveyancing or litigation; one of England’s newest law firms uses AI to prepare “polite” debt chasing letters for just £2.
如今，技术被用来快速浏览文件、进行尽职调查、总结案例，甚至起草简单的案件。它可以处理如产权转让或诉讼等事务；英格兰一家新成立的律师事务所利用人工智能，仅需2英镑就能准备出“礼貌”的催债信。
Nor is it all just grunt work. LexisNexis’s Lex Machina — no relation to this column — helps predict the outcome of litigation cases based on past behaviour of courts, counsel and judges. A&O Shearman’s antitrust AI tool works out which jurisdictions require regulatory filings to be lodged and what information they will need before drafting the necessary requests for any missing data.
而且，这些工作也不仅仅是简单的体力劳动。LexisNexis旗下的Lex专栏Machina（与本专栏无关）能够根据法院、律师和法官以往的行为，预测诉讼案件的结果。A&O Shearman的反垄断人工智能工具则可以判断哪些司法管辖区需要提交监管备案，以及在起草补充所需信息的请求前，需要哪些具体资料。
A few years down the line all this may look as laughably quaint as the Dictaphone. AI boosters see it plugging gaps in the constitution, highlighting potential legal action — think well-informed ambulance chasers alerting you to a breach of copyright, say — or even acting as judge. Parties input their grievances, the model spits out a resolution.
几年后，这一切或许会像听写机一样显得可笑而古怪。人工智能的支持者认为，它可以填补宪法的空白，指出潜在的法律诉讼——比如像消息灵通的“救护车追逐者”提醒你可能存在的版权侵权——甚至还能充当法官。各方输入各自的诉求，模型便会给出裁决。
For now, the case for junior lawyers remains. Finances stack up. Hourly billing rates vary hugely, but assume £600-£700 at a magic circle firm. Applying the lower end to 1,500 billable hours leaves several times their salary to be tipped into the partners’ pot.
目前，初级律师依然有存在的理由。这有着充分的财务方面的理由。小时计费率差异很大，但在“魔术圈”律所一般为每小时600到700英镑。即便按较低的600英镑计算，1,500个可计费小时所产生的收入，仍然是初级律师薪水的好几倍，这些都将流入合伙人的口袋。
Today’s juniors are also tomorrow’s seniors: succession planning relies on an intake of young blood. Algo-generated reports still need human oversight; that usually entails at least some degree of amending too. The Panglossian view on AI applies in law too: if it is easier to launch cases, more people will do so, thus expanding the pie.
今天的初级员工就是明天的高级员工：继任计划离不开新鲜血液的补充。算法生成的报告仍需人工监督，通常还要进行一定程度的修改。对人工智能的过度乐观在法律领域同样适用：如果提起案件变得更容易，就会有更多人这样做，从而扩大整个市场。
But there’s a more fundamental role for humans. AI, with tentacles in every sphere of business and society, requires its own rule book. That is a massive undertaking, spanning ethics, intellectual property, privacy and much else besides. Budding legal bigwigs still have a case.
但人类还有一个更为根本的角色。人工智能的触角已经延伸到商业和社会的各个领域，因此需要为其制定专门的规则。这是一项庞大的工程，涵盖伦理、知识产权、隐私等诸多方面。对于有志于法律事业的人来说，依然大有可为。