Leveraging AI in legal research: Shovels versus spoons

^{Photo: Scott Rollwagen, partner at Lenczner Slaght LLP}

This article was produced in partnership with Alexi

The decision to implement artificial intelligence (AI) within your firm — or not — calls for some fundamental introspection, says Scott Rollwagen, partner at Lenczner Slaght LLP and head of the firm’s research team. 

“This technology pushes people to think, what are we in the business of doing?” explains Rollwagen.

For example, when firms express hesitancy, it typically revolves around wondering what juniors will do if AI is handling what has historically been their tasks — but “we need to consider carefully the implications of that question,” he warns.

When doing talks on AI, Rollwagen often shares the story of American economist Milton Friedman who in the early 1970s was advising on the organization of another country’s large development projects. Observing the construction of a canal by thousands of people digging with shovels, he suggested investing capital and bringing in machines to get the job done faster. A government official asked, “But then what will our people do?” Friedman answered, “I thought you were trying to build a dam. If you’re trying to keep people busy, take away their shovels and give them spoons.”

“The ‘but what will our juniors do’ argument is another way of saying, take away their shovels and give them spoons and my advice is to step back and ask, is that how we want to do business? And does that kind of model have a future?” Rollwagen says. “It’s not just armies of people charging for their time. The law has become so complex that solving client legal problems involves an incredible amount of digging — but if you have the means to take that away, you create space for your lawyers to be more of what they would have believed they're trained to be.”

An ‘eye-opening new tool’

A major advantage of AI is that, when properly trained large language models (LLMs) are correctly deployed, it has the potential to accelerate the development of students-at-law and junior lawyers. Helping them reach more maturity of analysis sooner, they come to see their role as integrating the authorities with the facts of their case as opposed to trawling through reems of documents. Not just for new lawyers, it alleviates the deep digging part of the job across the board.

Calling Alexi “an eye-opening new tool,” Rollwagen says his group is using it much more extensively as their comfort level increases. Identifying the Research Memos function as its most valuable feature, overall he says Alexi allows for a much more responsive and immediate initial canvas of issues that subsequently enables lawyers to respond in real time to client issues.

Rollwagen points to an instance where he had a tight timeline on an injunction that needed to be worked up quickly. A few queries on AI saved the team a lot of time by almost immediately showing that some of the angles they were contemplating pursuing didn’t have authority to support them. In another case, the tool was able to ground a statutory issue “that was quite a brilliant idea but wouldn’t automatically have come to us off the top of our heads.”

“There have also been times I’m on the phone with somebody and have the tool running in the background, searching for information or to call to mind cases I may have forgotten, particularly more obscure ones,” he says, adding that it ultimately allows for much more risk-free exploration of alternative ways of looking at a case that lawyers wouldn’t have been able to previously due to pipeline problems.

Instead of someone taking a day or two just to hit a dead end, AI “lets us chase down strings to a reasonable level of comfort in a way that doesn’t waste time, and the end result is solving client problems more efficiently by spending less time trolling around for authorities and more time reading cases and deepening our understanding of them.”

Tying back to development, the experience of chasing a potential angle to its end is worthwhile, but does the marginal educational return justify the cost of ten wasted hours? By delegating that exercise to AI, a junior’s efficiency is increased in a more valuable way because they have the time to deep-dive into the higher value work — the human side of the equation where AI’s capabilities fall short.

“Every case is different, as are the particular ways in which a client’s problem presents itself, and there’s almost always room for play in how you position a case and take it to the last mile,” Rollwagen notes. “AI doesn’t solve the problem; it helps you assess faster whether a strategy is optimal. The human side is critical but there isn’t as much occasion for it if we’re wasting time doing the digging ourselves.”

AI in the copilot seat

The law will never be a system where lawyers are relegated to software engineers outputting language with no connection to the human element: understanding individual clients’ priorities, objectives, and agendas and integrating them with the raw material of the law is critical. The utility of AI is that it serves up questions or angles for that higher-level analysis, with the tool serving as a powerful copilot.

Rollwagen points to a book by internationally recognized scholar of contract law and legal history, the University of Toronto’s Stephen Waddams, Dimensions of Private Law Categories and Concepts in Anglo-American Legal Reasoning, as illustrative of the importance of human lawyers. The study of a 19th century case revolving around an opera singer who breached her contract with one concert hall by signing another contract with its rival, it was found that the singer couldn’t be forced to sing at the first venue, but by granting an injunction she could be prevented from singing for the second one. 

Waddam’s fundamental point is that if you’d taken all the case law up until that decision and put it into an infinitely intelligent supercomputer, it wouldn’t have spat out that result. Maybe AI can give you an idea, but it will always require human judgment to decide, is this going to play? Is this an appropriate strategic decision to get the desired remedy as opposed to another one?

“The book is almost prescient in terms of its explanation of how we think of the law as fixed and certain in all cases and if we can just find that secret algorithm to generate the right answer we wouldn’t need lawyers,” Rollwagen explains, but there was “a very human reconciliation of different lines of cases and turning them into something more.”

It underscores the partnership between the human mind and artificial intelligence, with the latter supporting the former by providing unprecedented powers that nonetheless must be wielded by lawyers with skill. 

“All the law can do is look backwards; especially in cutting-edge cases, we need the human aspect. The ability to look forward in a way that’s rooted in an understanding of how values, business, and life itself is evolving, is uniquely human. LLMs accelerate the assessment of where we are but can’t tell us where we’re going. AI tools are valuable because they provide a very good sense of what the solution space is of a problem, but the actual solution must come from a human being.”