The computer scientist Alan Turing, best known for his code-breaking during World War II, developed an experiment in 1950 to test whether a machine was artificially intelligent. A participant is sat down in front of a computer and begins a text-based conversation with another human and a machine. The participant cannot see either the machine or the human and must decide which is which through dialogue alone. If the participant is unable to tell, Turing argued that the machine involved in the conversation could be called artificially intelligent.
Under this definition, true AI does not exist yet. However, basic AI is now a reality and you’ve probably used it this week. Siri, the voice-activated personal assistant on an iPhone or Apple computer, uses machine learning to better answer questions over time. Amazon is another example. The online retailer uses a mass of algorithms based on your previous pages clicks and purchases to predict what else you might want to buy. Neither are perfect, but what makes them artificially intelligent is that they are getting better and better at what they do every year. This is because the codes and formulae that underpin AI software allow it to learn from trial and error as well as pattern prediction, much like a human or other intelligent animals.
Companies have also developed basic AI for law firms. IBM’s ROSS Intelligence was the software most frequently mentioned by law firms taking part in our survey. With its origins in software that beat two national champions on the US quiz show Jeopardy, ROSS allows lawyers to ask it questions in natural language. The platform then scans legal codes and databases to provide an answer. Unlike a simple search tool, it gets better the more it’s used, artificially learning from patterns and predicative behaviours.
Our survey of Latin Lawyer Elite law firms found one in three are trialling or using AI like ROSS. These early adopters fall into three groups. One group is comprised of independent, local firms that are not part of international tie-ups. Brazil’s TozziniFreire Advogados is a member of this contingent. It is testing software that automates contracts and speeds up due diligence.
The second includes firms working with international outfits, but are independent from them. For example, Galicia Abogados in Mexico is collaborating with UK Magic Circle firm Slaughter and May to test the due diligence AI platform Luminance for the Mexican market. “We’ve asked them to do some tests so we could get involved and get an understanding of how it works,” explains managing partner Manuel Galicia. “Things are evolving so fast and it’s important to be close to the process and see if we can tropicalise some of these platforms.”
The final group of Latin American firms testing AI are local players that are fully integrated into larger international outfits, such as Hogan Lovells (Mexico). Managing partner Juan Francisco Torres Landa says being part of an international firm is a major advantage. “Because we are part of a global outfit, we have more human resources and capital than many local firms,” he says.
Human resources and capital are also important. Software capable of conducting contract reviews can cost US$30,000, whereas a sophisticated system for hundreds of users might require investment of as much as US$250,000. AI software isn’t plug and play either. Someone must run the platform and teach others how to use it, which also entails significant costs.
AI is far from commonplace in the legal industry and the software is still evolving. Even if law firms have the cash, few want to risk purchasing expensive software that brings marginal gains.
John Aguilar Quesada, managing partner of Central American firm Aguilar Castillo Love, tells a cautionary tale about a visit to IBM’s Costa Rica headquarters in the early 1990s. “I was there for notarial work, but a salesman approached me and told me I needed to buy this new machine called a server and told me two of my competitors had already bought one,” he recalls. However, Aguilar was shocked when he heard the US$250,000 price tag. “I sat down in my car and was devastated – how was I going to afford that!” In the end, he didn’t buy it. “Not only was I able to stay competitive, but that equipment is out of date now.” The lesson he says is that you don’t need technology for technology’s sake and when you do make a purchase, make sure you need it.
The perceived risk of being an early adopter is reflected in our results. Around two thirds of Latin Lawyer Elite firms are yet to trial or use AI. For now, most are just investigating its potential. Given the survey focused on firms with deep pockets and sophisticated management, one can assume the wider Latin American legal market has also not engaged fully.
While that may be the case, no one interviewed by Latin Lawyer was in denial about how quickly AI may emerge as a viable tool for the legal industry in Latin America.
“Technological change can creep up on you slowly and suddenly you’re operating with a different structure,” says Pablo Berckholtz of Peru’s Estudio Echecopar – Member firm of Baker McKenzie International, who warns against complacency. “I remember when I got my first Blackberry in 2001 – it was text only, had a small screen, and 15 years later and you have the internet, remote accessibility and pretty much everything else.”
Torres Landa is even more strident. “AI is not a trend, it’s not the flavour of the month and anyone who falls asleep will be run over by the competition,” he says. “We are on the brink of dramatic change.”
Siri on steroids
There are three areas where AI has the most potential for legal work: predicting litigation results; drafting contracts; and document review.
In the case of litigation, a majority (55%) of respondents think AI will be superior to human judgement alone within a decade. Typically, AI litigation software works by a lawyer asking the computer programme to assess how likely their client is to win a given case. After cross-checking appropriate court rulings, the AI responds by giving the questioner a probability. This may extend to recommendations on how the lawyer should adjust a legal document to increase their client’s likelihood of success. Two of the 34 managing partners surveyed thought this sort of software may even drive litigation strategy soon.
Galicia Abogados’ Galicia thinks the role AI can play in litigation is clear-cut. “A lot of times, we’re asked by clients to provide them with a list of potential litigation risks in their contract and the possibility of winning a case under those circumstances,” he says. “If you have tools to better predict the future, you should use them.”
His opinion is shared. TozziniFreire’s Rodrigo de Campos Vieira gives an example. “Suppose we have a client and we find they are always losing in court because of section eight in their contract; we can use AI to recommend an adjustment to the clause which is more likely to succeed.”
However, a sizeable minority of respondents (38%) think a litigator’s judgement will always be essential. While a proponent of AI, de Campos Vieira still falls into this category, noting AI’s current struggle with nuance. He provides another example to clarify his point. “If a rock concert was cancelled, all the consumers are in the same situation; this sort of situation can be handled well artificially,” he says. “However, if you have a labour dispute involving two employees who have been dismissed for the same thing, but are claiming compensation for different reasons, that would need to be adjusted for by a human.”
Beside litigation, AI’s developers tout its ability to automate large swathes of work, including contract drafting. Here, AI programmes can even kick into effect at the due diligence stage, cross-checking companies’ relationships with third parties and raising red flags for compliance concerns or conflicts of interest. Moreover, AI developers are keen on automating everything from corporate incorporation documents to trademark renewals and securities filings. Latin Lawyer Elite firm managing partners expressed a high opinion of AI capabilities in this respect. Of those surveyed, 30 believe AI can make legal drafting more accurate and efficient, while a fifth believe it will completely automate the process. None of the respondents believe AI will be of little use in this area.
Document review, which can be especially costly for law firms given the manpower required to go over huge volumes of paperwork, is another service AI developers promise to address. Again, Latin American law firm managing partners believe AI is well placed to address this process. Nearly 80% of respondents think AI platforms will make document review more efficient and accurate and none think AI will be of little use.
One in three managing partners surveyed believe AI is already transforming the legal profession, but the remainder think it will take another decade before it makes a real impact.
Technical imperfections aside, language is one reason why some don’t believe AI will be a transformative force until the 2020s. This is because most of the AI software currently under development is for English-language jurisdictions. “We are following the developments very closely, but we still haven’t come across a Spanish-based AI equivalent to ROSS or iManage, and that limits the possibility of implementing anything in our region soon,” says M & M Bomchil Abogados managing partner Javier Petrantonio in Argentina. “The Anglo-American legal space is a huge market and developers have to balance whether it’s worth making such a big investment for Latin American law firms with limited resources,” he adds.
Furthermore, Latin American court systems do not always have the up-to-date digital databases necessary for AI platforms to work. “The UK and US court systems have databased their precedents, but countries like Peru are a bit behind and don’t have these databases for AI to use,” notes Berckholtz. He says his firm is testing AI at a global level, which could be applied to local deals and cases once these databases are available in Peru. This is a problem throughout the region, restricting AI’s current application to a limited number of cross-border deals and cases.
Still, few would argue that AI software will remain confined to English-speaking jurisdictions. Given the commercial incentive to adapt AI for non-English speaking legal markets, it’s only a matter of time before it comes to the region. While Latin American law firms can’t boast the same revenues as their international counterparts, the industry is far from struggling. “If there is a business opportunity to help law firms, someone will be on it and we won’t need to run our own computer labs for it to happen,” says Berckholtz.
Simone Musa, who co-manages Trench, Rossi e Watanabe Advogados in Brazil, strikes a similar tone. “We think in the longer-run, there will be partnerships between law firms and start-ups,” she says. “This is something we should invest in and stimulate.”
Latin Lawyer’s survey also sought to understand what impact AI may have on law firm jobs and organisation. AI does not have a particularly good reputation in this respect. In fiction, humans often give AI important responsibilities only to find the machines decide they know best. For example, in Stanley Kubrick’s 1968 science-fiction film 2001: A Space Odyssey, a computer called HAL 9000 is given the job of piloting a spaceship to Jupiter. After a malfunction, HAL (shift the letters forward and you get IBM) concludes the humans on the ship are jeopardising the mission and tries to kill them. Fiction often reflects reality, and some prominent figures, including the astrophysicist Stephen Hawking and business tycoon Elon Musk, have called AI one of the greatest threats to humanity. Even if a Terminator-style nuclear apocalypse isn’t imminent, predictions of job losses caused by AI can make for alarming reading. Deutsche Bank CEO John Cryan made headlines in September when he predicted a “big number” of his employees will soon be replaced by robots. Consultancy firm PwC is also predicting major layoffs. It expects more than 10 million jobs, or 30 per cent of the UK workforce, will be replaced by a mix of robots and AI in the next 15 years. While few managing partners expect such significant job losses at their law firms, just six of the 34 polled believe AI will have no impact on who they employ and how they structure their teams.
Junior associates, paralegals and law clerks are widely considered to be at greatest risk of losing their jobs to AI. This is because they are often involved in the research-intensive work computer software is best positioned to replace. M & M Bomchil’s Petrantonio believes jobs losses are likely, but will be very limited. “AI probably will result in fewer associates, but we are not the mega international firms and it’s something that could be managed and would be less significant,” he says. After all, few managing partners want to destroy their legacy by pulling their associate-base from under them and leaving no one to replace them.
Instead, what a junior lawyer does may change. Trench Rossi’s Musa offers one potential outcome: “Instead of carrying out research and proofreading documents, I think junior lawyers will work much more closely with technology and start becoming experts, customising the software to each client’s demands,” she says.
AI’s erosion of steady day-to-day work may also change their contractual relationship with law firms. “We foresee less fixed contracts and more work based on individual projects,” says Musa. She admits this could result in young lawyers working for more than one firm, depending on what work is available.
Project-based work for associates would result in law firms taking a step toward the “gig economy” spearheaded by companies like Uber, which contract rather than employ most of their workers. Invariably, the consequence is less job security and more a precarious livelihood, but flexible working hours and conditions along with greater day-to-day variety. This less rigid approach to what work is and where it is conducted is already a growing trend, even in the somewhat conservative legal sector, notes Musa.
Whatever junior lawyers’ contractual relationship may be, our research revealed most managing partners expect associates will need to be tech-savvy data analysts and not just lawyers if they want to keep their jobs. “I want junior lawyers who can talk easily about technology and understand the trends and notions driving software development,” says TozziniFreire’s De Campos Vieira, summarising the attitude of others.
Perhaps if young lawyers can also become proficient at using and even programming AI software, they will meet the expectations of most managing partners (73%), who expect AI will result in legal teams composed of senior lawyers and professionals with skills different to law, such as data analysis.
But it won’t be enough for young lawyers to be just good at law and computing. As computers become ever more invasive and the need to work in an office is eliminated, clients will place a premium on dealing with humans. Strong interpersonal skills, which are already highly valued in the legal industry, will become even more important. “Computers are helping us a do a better job, but some skills remain relevant and one of the most important is being able to develop strong personal relationships,” says Galicia. “Clients need to feel confident, share their concerns and get a personal approach, and that will never be replaced.”
Junior associates and even law clerks and paralegals can therefore breathe a collective sigh of relief. While six of the 34 respondents surveyed do expect a significant chunk of their work to be done by AI, the overwhelming majority (82%) believe lawyers will still be needed to check the results and fulfil other roles. Janet Stanton, a partner at Adam Smith, Esq, who has conducted extensive research on the subject, thinks this balanced approach is best. “There seems to be a clear-eyed appreciation of how best to implement AI at firms; the optimal approach is for AI to contribute to and enhance the judgement of lawyers and paralegals,” she says.There was also a lot optimism about the positive impact AI may have on the profession. About eight in 10 believe machine-learning will free lawyers up from more mundane tasks so they can focus on more intellectually-rich work. Galicia gives an example: “I don’t think doing due diligence where you need to review 100 indirect exchange of control agreements is the most challenging work for young people to do,” he says.
The degree to which managing partners think AI will reshape the legal profession above the junior level is more mixed. A small majority (41%) believe AI will completely overhaul low-level tasks, largely involving junior lawyers. Nearly as many (38%) expect the changes to significantly change the profession, perhaps affecting senior lawyers as well. The remainder (20%) expect AI to change the legal industry, but not in an especially dramatic manner – improving productivity to a limited degree.
Still, even those who expect AI to have a transformational impact are not very concerned about its impact on senior roles. “I might be wrong, but I don’t see AI putting lawyers out of a job within a generation,” says Musa. “A large part of the business is negotiation, persuasion and relationships, and I don’t think that will be totally replaced.”
Moreover, there was a consensus that the more subjective aspects of the profession, especially where a lawyer plays the role of both legal advisor and businesses consultant, were future-proofed. “You will always need experienced people that challenge the law on a substantive basis, not on a formulaic one,” says Aguilar of Aguilar Castillo. Laws reflect societal values and are not an exact science like medicine, he notes.
The bottom line
In the 1998 surrealist thriller Pi, the protagonist, a mathematician called Max Cohen, programmes a computer to predict the stock market. It doesn’t end well. By the end of the film, Cohen’s powerful invention has driven him mad. The film ends with him at peace on a New York park bench, but only after destroying his invention and trepanning his own skull with a power drill. Most managing partners are more optimistic about AI’s ability to make their firms more money: a large majority (70%) believe AI software will be good for their bottom line.
Interviews revealed that AI’s profit-boosting potential is most likely to come from firms being able to take on more work rather than reductions in the number of people they employ or contract. “I’m reluctant to say AI will replace lawyers, but it will save a lot of money spent on research and act as a tool to save us doing a lot of work,” says José Luis Ambrosy of Chile’s Claro & Cía.
Torres Landa agrees. “AI will be a cost-savings aid allowing firms to focus on specialisation and really knowing our clients’ business so they know how to provide them with solutions and not just a service,” he says.
However, not everyone thinks AI’s positive impact on profits is a certainty. “We can’t reasonably believe that only law firms will reap the benefits of AI,” says Berckholtz. “There’s a lot of competition and a lot of client-driven forces to keep costs down.”
Client-driven cost cutting may impact law firms in diverse ways. One possibility is that AI allows businesses to take entire practices in-house. Some managing partners already see this happening. “Only last week, we had a meeting with one of the big multinational Mexican companies and their general counsel was telling us how he has already used an AI programme to do an M&A transaction,” says Galicia. He adds that Mexico’s financial technology sector, which is the largest in Latin America, will expect law firms to adopt AI. This phenomenon will only increase as the number of transactions in the technology sector increases, resulting in clients, who – rightly or wrongly – believe computers can do everything.
Still, Mexico is not necessarily representative of the rest of Latin America. Responses were mixed concerning clients’ demand for AI. While a small majority (42%) think clients will demand AI to make services more efficient and cut costs, a slightly smaller group (39%) believe clients may benefit, but won’t necessarily demand AI. Some (18%) went so far as to say clients don’t care about AI so long as the job gets done.
Lawyers who do not think their clients care about AI should be careful. The Latin American Corporate Council Association (LACCA), which is Latin Lawyer’s sister association, found in its most recent technology survey that general counsel were increasingly turning to digital solutions, including AI, amid budget cuts and shrinking headcounts. Adam Smith, Esq reinforces this point here. “As a matter of routine, sophisticated clients now expect law firms to demonstrate their technological capabilities and expertise when pitching for work as part and parcel of their overall qualifications for the engagement,” writes Bruce MacEwen, Adam Smith’s president. “Anyone who doubts that AI is here to stay in Law Land is living in some past decade.”
A brave new world
Not all cinematic depictions of AI are bad. In the 2015 action film Chappie, South African director Neill Blomkamp makes an artificially intelligent robot the film’s protagonist. In the movie, the robot, Chappie, is neither good or bad. Instead, it has a childlike innocence and learns and mimics the values of its owners. While the robot is abused and used for illegal purposes, it ultimately proves to be a force for good.
Today, AI is becoming a reality, whether it’s beating a human on quiz shows or suggesting how to change a client’s M&A contract. Lawyers have an especially important role to play in how it is used. “Now is the best time to be a lawyer,” say TozziniFreire’s De Campos Viera. “AI, driverless cars, crypto currency; this whole new world needs to be regulated and regulated well.” Be too conservative and entire countries, not to mention law firms, become uncompetitive. But embrace untested technology too eagerly and the consequences can be just as devastating, whether its on jobs or deal-ruining mistakes. Much like Blomkamp’s Chappie, it’s up to humans to get the balance right and create something that makes work more intellectually-rich and productive and not a robot that recklessly terminates jobs.