Bringing the best of generative AI to legal work

In a recent episode of the Legal Talk podcast, Jeff Reihl- Chief Technology Officer and Jamie Buckley – Chief Product Officer from LexisNexis^® Global joined Greg Dickason, Managing Director of Asia & Pacific to delve into the transformative impact of generative AI on legal work. This blog unpacks some of the key points of the discussion. To listen to the full episode, click here.

While the hype around generative AI continues, it’s fair to say that people across industries are starting to become more aware of what it can do, and where its limitations lie - and while the technology is undoubtedly transformative, expectations are being tempered as reality sets in. The most common realisation is that while broad-spectrum public generative AI solutions like Chat GPT or Bard are very useful, in order to get really valuable results for any specific use case, you need a custom tool - especially for professions where accurate advice is critical to client success.

Enter Lexis+ AI^™, our generative-AI-powered legal research solution, which takes the raw power of foundational large language models (LLMs), and fine-tunes them for the specific use cases that are most relevant to lawyers. Backed by LexisNexis authoritative content, Lexis+ AI provides a more relevant, more reliable generative AI experience for lawyers - opening the door for transformed ways of working.

“There are four primary use cases for Lexis+ AI,” says Jamie Buckley, Chief Product Officer of LexisNexis Global. “The first one is the ability to ask a complex legal research question. The second one is drafting a legal document- from just a simple prompt, you can now have a complete draft of a legal document. The third scenario is about summarising long documents - so, this is around like case law documents, for example. Then the fourth one is that it provides our customers the ability to upload a document and then ask a question about it or summarise it.” In every market across the world, Lexis+ AI is built around these four use cases - augmenting some of the most common tasks for lawyers with the most prominent capabilities of generative AI.

While generative AI has been a part of the common lexicon for the last 12-18 months, the roots of this technology stretch back much further - and Lexis+ AI has been built off the back of a wealth of knowledge. “The reality is that we've been doing AI for well over a decade now,” says Jeff Reihl, Chief Technology Officer at LexisNexis Global. “We really started with basic Natural Language Processing type capabilities and some machine learning. But I would say that everything really changed when chat GPT came out, because that's when the generative capabilities really hit critical mass. And that fundamentally shifted our focus in this area, which we backed with over $1 billion USD of investment into research and development of this and other products.”

Having just been launched in Australia, Lexis+ AI has been available in the US for a while already - with customers discovering the advantages of having a specifically tuned and purpose-built LLM for legal use cases. One of the things that our customers found out very early on is that chat GPT just by itself out of the box is just not good enough for complex legal scenarios, which has driven many positive reviews from clients who have already trialled the solution.

Dealing with the limitations of LLMs is something the product and technology teams at LexisNexis are very familiar with, and it goes without saying that having your AI give you an imagined answer can be a real problem in law. Similarly, explainability is a key issue - as a lawyer, you have to know where you got the information from. And third, currency. A lot of publicly-available foundation models are trained on data that’s over a year old.

By grounding all the data in LexisNexis content, and utilising the strong digital infrastructure that has been built up over the years through previous solutions, those limitations can be addressed. Through the unique LexisNexis search engine, citators can be leveraged to make sure that whatever citation you bring forward is still good law, and our content ensures that the answers are being drawn from the most relevant data sets. As might be expected in the development and rollout of a project of this scale, customer input and feedback have been an essential part of the journey. Customers have been informed of many aspects of the product’s creation through a commercial preview programme, in which a small set of customers were given early access to the solution.

“During that time, we found out that uploading documents and the ability to ask questions on those documents was really important. That was always on our product roadmap, but it increased dramatically in priority based on feedback from customers. We also saw the importance of training, because not everyone is well-versed in the prompting of LLMs. We found that specific training was helpful in allowing customers to get the best results. And we've also built in some additional functionality - like specifying jurisdiction, for example - to really help ensure that you're getting a good response.”

Following the US release, feedback has been positive overall. Many customers have found significant time savings through Lexis+ AI, especially around complex legal research and drafting documents. Similarly, the greater relevance has improved research, as has the explainability of having citations included with any answers - along with links to original documents so that users can continue their own investigations. But one of the strongest areas of feedback has been around the solution’s performance - its ability to provide a relevant answer in a reasonable amount of time. “We’ve gotten really positive feedback from customers that our answers tend to be about twice as fast as a major competitor,” says Jeff. “Which is great - nobody wants to wait for an answer that's inferior!”

But despite a global rollout that is proving incredibly successful, there are no signs of slowing down when it comes to further innovations on the product. Planned updates include increased performance across the four key legal use cases of asking complicated legal research questions, drafting a document summarising, and uploading. These will be supplemented by expanded use cases, and expanded content sets - along with language translation as the product continues to roll out across non-English speaking countries. And in response to customer requests, a feature allowing customers to apply the Lexis+^® LLM to their own content is also in development.

So, for lawyers or firms currently preparing to dip their toes into the world of gen AI, what are the best ways to get started? “It's really important for lawyers to just try it out, right?” says Jamie. “You can get a free trial anytime, so I highly recommend people try it out. And also, it’s important people get educated about generative AI because the more they understand, the more use it will be to them.” For Jamie, some recent customer trials have informed his advice. “Start with experimentation, and start with a small group of people before expanding from there.”

“As the rollout of Lexis+ AI continues, it’s safe to say that the product represents a very exciting - and possibly fundamental - shift in the way that law is practised,” says Greg Dickason, Managing Director LexisNexis Asia Pacific. “We’re really excited to be working in close collaboration with a number of customers across our entire ecosystem here in Australia with really strong positive feedback and excitement not seen before. By bridging the capabilities of powerful generative AI technologies and applying them to the specific use cases that are most helpful to lawyers, Lexis+ AI can produce relevant, accurate, and explainable results in a fraction of the time that was previously possible. The future of legal practice is here, and Lexis+ AI is paving the way.