In a couple of weeks, ChatGPT – the chatbot that brought generative artificial intelligence into the mainstream – will have fun its second birthday.
The event also provides a natural opportunity for reflection for the legal industry, whose business model has been identified by analysts because the industry with the best opportunities through the adoption of large-scale models underlying generative AI tools.
However, because the generative AI “hype cycle” enters its nadir of disillusionment, law firms are moving away from big announcements and as an alternative taking a detailed take a look at how and where AI tools might be most helpful. A report released this summer by management consultancy Deloitte found that only 6 percent of corporate clients believed they might profit from using generative AI of their firm. However, greater than 70 percent said they expected the technology to deliver cost savings on legal work and faster turnaround times.
Proponents of integrating AI into the legal industry consider that such efficiency gains shall be possible within the near future, in the event that they should not already present.
Here are some key topics that early adopters of the technology are currently specializing in:
Data protection
Law firms realized early on that a one-size-fits-all AI system wouldn’t work in a world industry where data protection and confidentiality are paramount.
For starters, regulations dictate that some data should be stored locally, and the underlying models equivalent to OpenAI are increasingly unavailable in certain countries equivalent to China. Tara Waters, a partner at Ashurst until last month, notes that the tools needed to find out what data might be utilized by a generative AI model and where, “don't exist on any of the third-party platforms” – not even market-leading offerings equivalent to Microsoft's Copilot. Because of this, she described the corporate as “not in a spot where it feels comfortable with Copilot.”
Ashurst continues to be in a holding pattern with some clients who need assurances about how data shall be isolated and where it is going to be stored for compliance reasons.
David Wakeling, a partner at A&O Shearman, says his firm has 1.5 billion documents that big tech corporations might need to feed into an algorithm en masse, but “we’d like to guard our clients' data,” so his firm is rolling out the complete stack of AI capabilities just for specific clients and specific tasks.
Once AI is used for cross-border remittances, he says, “it gets very complicated.” In that scenario, “we only must deal with processing in that jurisdiction to fulfill those requirements… that tech stack suddenly becomes difficult because now you would like cloud processing within the Middle East or Switzerland or Singapore.”
Hogan Lovells' bespoke AI tool, Craig, uses “auto-pseudonymisation”, auto-encryption and no data retention – meaning its algorithm quickly deletes information on which it has drawn its conclusions. Even so, partner Sebastian Lach says not all users can have access to a full range of products to comply with local regulations.
Reducing drudgery
While he admits his firm may not use the “sexiest” AI applications, Peter Werner, a partner at Cooley in San Francisco, says a few of the tools that should not “specific to providing legal services” can deliver immediate time savings. For example, the firm, which has offices in London and Brussels, uses the technology to summarize long email chains and permit employees to catch up more quickly on missed messages after coming back from vacation or before a gathering.
While Hogan Lovells' Lach, who co-leads its technology division Eltemate, acknowledges that features equivalent to creating an initial draft contract or an initial briefing document may not seem particularly revolutionary, he argues that they “replace the work that no one likes to do” and “unencumber your brain to do the things that actually matter… creating justice, good outcomes and good laws.”
Securing jobs
When Goldman Sachs analysts warned last yr that 300 million jobs may very well be lost to generative AI, they identified the legal industry as one of the crucial at-risk sectors. Yet there aren’t any signs of mass layoffs; only lawyers are reporting incremental efficiency gains. A&O Shearman says the firm is seeing productivity gains of 20 to 30 percent in some cases – or about seven hours per contract review – due to the firm's internal system, ContractMatrix, utilized by 2,000 of the firm's lawyers. Overworked employees are the heavy users, no less than for now, because they’re in additional demand than ever.
Waters predicts that it is going to take “an excellent three to 5 years before anyone within the industry actually has a transparent idea of how their business model might need to alter.”
However, along with the long-term fear of staff cuts on account of AI efficiency improvements, there may be one other concern.
Cooley's Werner worries that employees “turn off their brains to create a primary draft of a temporary or a contract” and lack the in-depth training they should think quickly and advise clients. Thanks to AI tools, Cooley employees can now start an organization in minutes. But that's “just pushing buttons,” Werner says.
“How can we get them to the purpose where in 15 years they’re a one that can teach a brand new generation how you can be an incredible lawyer?”
In-house training
No law firm—not even certainly one of the industry's biggest—can afford to develop its own large-scale language model or compete with well-funded generative AI startups. But somewhat than accept out-of-the-box solutions like OpenAI, many have chosen a middle ground: training algorithms using their very own data sets and mental property. The key, says Lach, is “giving the AI good books to read”—or literature tailored to the needs of their lawyers and clients.
At Hogan Lovells, the result known as Craig, which will help users with all the things from navigating regulatory updates to preparing a prospectus for an IPO. The system is used each internally and externally by most of the firm's larger clients.
A&O Shearman – which, like Allen & Overy, began rolling out the generative AI tool Harvey in late 2022 – found that lawyers were using it primarily to beat “author’s block,” Wakeling says, while they shied away from more sophisticated applications for fear of inaccurate results.
The company developed its ContractMatrix drafting tool partially to avoid what it calls hallucinations – the transmission of false or misleading information. The tool's output includes hyperlinks back to the unique, signed work referenced, allowing lawyers to simply confirm facts and conclusions on the source.
