GenAI in the legal sector: yes, it will affect you, the only question is when

  By Manuel Fernandez Condearena

When we live in a very intense media moment, it is respectful to add more fuel to the fire. As consultants, we try to escape the shine of the latest novelty, and provide solutions that have potential, even if they are less dazzling in appearance. But it seems that this time we have to surrender to the wave that has already arrived, and accept that generative AI is going to affect all professionals in the legal sector, that it is only a question of when and to what extent.

To paraphrase Furlong, a favorite quote about transformation of the legal sector, we all have resistance to change, what happens is that we lawyers are especially good at it. If we look back, the technological changes that have really affected the legal sector have been changes that were imposed on society in a general way. At the beginning of digitalization, knowledge databases arrived on DVD, replacing paper (some of my professors were scandalized because “anyone” was going to find the relevant jurisprudence, and thus something about the profession was lost). Then it came to all of us, including lawyers, the use of the Internet (with that loading bar for websites that moved slowly) and email.

And now? A technology that has been evolving “under the radar” for some time now comes to light, capable of managing large amounts of information, that can work on unstructured text, and that on the basis of all that information can prepare summaries, make categorizations, generate high-quality text, with the only requirement that you ask things very well... Gosh, it sounds just like a lawyer, right?

What makes this change have a high impact for the legal sector is that, being a technology that is not specific to the legal sector (which means that it is developed outside of investments in the legal sector), Legaltech), It has a very high potential in its application in the sector. It is true that language models that are more trained in legal matters, or specific to the legal sector, will arrive, and their contribution may be even greater, but what we have today already provides us with value. If you follow Dan Katz, you already know that GPT4 passed the UBE exam, an exam created by the National Conference of Bar Examiners (NCBE) in the United States that is designed to assess the knowledge and skills that every lawyer should have before obtain a license to practice law.

Much has been written about the possible use cases for this new technology, some already indicated above, others less obvious, such as legal risk management or monitoring regulatory or jurisprudence changes in multiple jurisdictions. It seems clear that, apart from specific AI solutions that are created to address specific cases, one of the keys to success will be to start working on the use cases with the greatest potential.

Thinking about this topic, the other day I was rereading a report from our Deloitte AI Institute which proposed a simple and practical methodology to test use cases of generative AI. It is about assessing, on the one hand, the effort required to carry out the task without AI, and on the other hand, the effort required to validate what the AI ​​has generated. The first variable analyzes whether the use case is worth the effort it avoids. In the second, it is about covering the need to verify that what technology has generated makes sense and adds value. And it offers a series of examples that make the relevance of the model very clear. If we use AI to generate an image, which is something that requires a lot of effort, we can check at a glance if the image meets our expectations, and thus we arrive at an interesting use case due to the combination of high effort saved and little effort. effort required to validate. If we use the solution to generate a contract, which is something that also requires a lot of effort, a distinction arises depending on the user: if he is a user with legal knowledge, he can validate the result with relatively little effort; On the other hand, if you do not have that knowledge, the validation effort is very high, and the use case loses interest.

With this scenario of almost certain change ahead, it is worth asking what the best way to face it may be. First of all, address the new scenario with a change of mentality: AI is here to stay, also in the legal sector, and the sooner we, including lawyers, begin to understand it, the better prepared we will be. Secondly, start looking for those use cases that, in our case, in that of our legal advice, in that of our office, will add value and are worth exploring. Thirdly, aligning our vision on AI with that of the rest of our organization, it is necessary that we align both in some way. Fourthly, be positive: all changes carry risk, but they all have the potential to help us improve our profession and our services, and that logic will undoubtedly pay off.

And in that positive vision of change, two final notes and a warning. The first note is that this may be a moment for those who have not yet been able to focus on the transformation of their legal advice to do so, new possibilities open up. The second is that business projects that include the use of AI will require legal advice to be very close advising on its many legal implications, and it is a new opportunity for it to grow in value and in its contribution to the business. And the warning is that the organization's transversal projects will be more complex, with the involvement of business, technology, and AI teams, the involvement of legal counsel and external consultants and advisors, and due to this complexity, project management will be more relevant than ever.

Cheer up for the first use case!