« As an AI language model, ChatGPT can provide assistance to lawyers in various ways. It can help with legal research by quickly searching and analyzing large volumes of legal documents and cases to extract relevant information. It can also assist in drafting legal documents such as contracts, briefs and pleadings, by suggesting language and formatting options based on context and legal precedent. Additionally, ChatGPT can assist with legal analysis by providing insights and recommendations on complex legal issues based on its vast database of legal information.身為人工智能模型，ChatGPT能以不同方式為律師提供協助。其能藉由快速搜尋及分析大量法律文件與案例以萃取相關資訊以進行法律研究。其亦可協助草擬諸如契約，摘要及答辯等法律文件。
» Is ChatGPT trying to take lawyers’ jobs away (in 100 words or less)?ChatGPT試著奪走律師的工作
« No, ChatGPT is not trying to take lawyers’ jobs away. Instead, it can assist lawyers by providing them with a powerful tool to enhance their work efficiency, accuracy and productivity. While ChatGPT can automate certain tasks such as legal research and document drafting, it cannot replace the expertise and experience of a skilled lawyer. In fact, the use of AI technology like ChatGPT can help lawyers to focus on more complex and high-level tasks that require human reasoning and judgment.不是，ChatGPT並非試著奪走律師的工作。取而代之，其能協助律師並提供強而有力的工具使其更有效率、更精確地而富產能地完成工作。
» Now say that in rhyme.
« ChatGPT knows its limits well, It cannot replace lawyers with its spell. For only humans can bring the skill, Of judgment and experience, to fulfill. So fear not, lawyers, ChatGPT is your friend. It’s here to assist and help you to the end.
In November, when artificial intelligence research firm OpenAI launched its new chatbot, ChatGPT, it stirred both anxiety and awe. Type a prompt into a chatbox, such as, “How can you make lawyers’ lives easier?” “Will you take my job?” or even, “Write me lyrics about John Roberts in the style of Kendrick Lamar,” and ChatGPT can answer in seconds.
Built on OpenAI’s GPT-3.5 language model and trained on a massive—if dated and limited—dataset, ChatGPT quickly captured the world’s imagination and claimed 100 million active monthly users by January, according to a study by investment bank UBS.
People have quickly discovered the platform can churn out persuasive and eloquent copy that would take a human hours, days, months or even years to write, including drafts of emails, legal briefs, essays and books.
ChatGPT represents a dizzying leap in the capabilities of generative AI, which can create original content based on the dataset the technology draws upon. But while some in the industry are exploring its potential to aid in legal research, contract review, communications and litigation strategy, there is plenty of hand-wringing about how it could make lawyers obsolete.
Suffolk Law School Dean Andrew Perlman used ChatGPT to help write a 24-page law article, draft a U.S. Supreme Court brief on same-sex marriage, craft deposition questions and work on a real estate contract. Though the tech is flawed, and OpenAI says it may produce inaccurate and biased responses, Perlman predicts the technology will touch all facets of law practice.
What we’re seeing now from emerging services like ChatGPT is there is a future for the use of AI and the delivery of legal services,” Perlman says. “We’re not there yet. But it’s pretty clear it’s coming very soon, and it’s going to have a disruptive impact on the way we go about our work.”
Other tech companies are getting in on the generative AI game. Google launched the chatbot Bard in early February. And the tech giant has invested in Claude, a chatbot created by Anthropic, a company founded by former OpenAI employees.
Not to be outdone, OpenAI announced in mid-March that it had created GPT-4, the newest iteration of its AI language model, which is capable of processing images and more text, among other enhancements.
Related article: Some law schools already are using ChatGPT to teach legal research and writing
OpenAI cautions that ChatGPT can respond with “plausible-sounding but incorrect or nonsensical answers.” Co-founder and CEO Sam Altman tweeted in December that ChatGPT is “incredibly limited but good enough at some things to create a misleading impression of greatness.”
Legal technology expert Dazza Greenwood, a lecturer and research scientist at the Massachusetts Institute of Technology Media Lab, warns that lawyers should confirm and authenticate everything ChatGPT produces.
“The best way to approach this is to either not use the technology for anything important, or to use it and be prepared to stand behind every word, every supposition, every assumption, every assertion, every single bit of the content,” Greenwood says. “In other words, just to treat it as a first draft; never to rely upon it as a final draft.”
Despite concerns about the veracity of the tech, many in the legal industry are forging ahead with generative AI to future-proof their businesses.
In February, London-based law firm Allen & Overy tested an AI platform called Harvey—GPT-4-based software tailored for law firms and backed by an OpenAI-managed investment fund. The firm says 3,500 lawyers tested Harvey and currently use it for legal drafting and research.
David Wakeling, the head of the Markets Innovation Group at Allen & Overy, says experienced lawyers at the firm are already using the tech to help draft provisions in documents; get an overview of whether contracts are enforceable or could be challenged under English law; and for translations, business meeting agendas and slides. He adds that experts validate everything it produces.
“I wouldn’t want to be a law firm that isn’t using AI in 18 months time,” Wakeling says. “I think that would be a serious competitive disadvantage.”
Several legal technology companies also are experimenting with software that uses generative AI, including LawDroid founder and CEO Thomas G. Martin. He is beta testing GPT-based tech in his platform LawDroid Copilot, an AI legal assistant that helps with legal research, summaries and correspondence.
“The technology has caught up with what our vision was,” Martin says. “The most difficult thing for us right now is trying to stay at the front of the wave and keep up with everyone’s interest.”
According to Greenwood, the tech also will give rise to a new job title, “legal prompt engineer,” with the industry hiring experts to query platforms like ChatGPT to get the best results.
Indeed, in February, the British law firm Mishcon de Reya started advertising for a GPT legal prompt engineer in its London office.
Meanwhile, Greenwood is knee-deep in the area of prompt engineering himself. In December, he prompted ChatGPT about a hypothetical breach of fiduciary duty case with an executive named “Bob” at the center.
Greenwood published the results on his CIVICS.com blog. They show the tech coming up with possible scenarios of how the litigation might play out and how a court might rule based on the evidence and on appeal. He went even further down a rabbit hole, asking the chatbot to document a political movement following a Supreme Court ruling in the case and prompting it to write a first draft of federal legislation.
“This is good when you’re trying to just get the shape of something completely legal in context. [It’s] something that is neither an outline nor a draft of a legal document. It’s a whole new class,” Greenwood says.
While generative AI could create new opportunities at law firms, it also could benefit clients and people who are representing themselves. As the tech evolves, Perlman says it could help address the access-to-justice crisis in America by “giving people the opportunity to gain access to information they might otherwise have difficulty finding.”
According to Access to Justice Ventures founder Natalie Anne Knowlton, people seeking legal services already have to sift through vast amounts of information online to find what’s relevant. She says the technology has the potential to “distill information online into one spot.”
