Few technologies have generated as much hype across the legal industry as Generative AI. Open AI’s ChatGPT, Google’s Bard, and a host of new solutions being developed on Large Language Models (LLMs) are making waves across the world.

To put that into context, ChatGPT acquired 100 million users in a little over 2 months – making it the fastest growing consumer software application ever. To put ChatGPT’s success into perspective, it took TikTok nine months to achieve a similar number of users. Other apps took much longer to do so — Facebook took almost 5 years; Canva took 9 years; and Instagram needed 2.5 years to reach 100 million users.

Greg Dickason

Managing Director, LexisNexis Asia-Pacific

Lindsay O’Connor

Head of Core Product Pacific, LexisNexis

Generative AI refers to artificial intelligence that can generate new content, such as text or images. And as you might expect, this has far reaching implications for everything in the legal industry - from the future of work, to specific practice areas. And as tends to happen whenever a new AI technology gathers momentum, a slew of headlines questioning whether lawyers will now become obsolete have emerged. They won’t - but this new technology does raise a number of interesting questions and opportunities for our industry. Lindsay O’Connor, Head of Core Product, LexisNexis Pacific recently chatted with LexisNexis Managing Director, Greg Dickason, on how he views these changes.

“From my perspective, the hype around Generative AI comes from what I call a ‘step change’ in technology,” says Greg.

“Essentially, the technology has now hit a point where it can do something vastly more useful, and impressive, than it has been able to do in the past. Another big step change I can think of was switching from AltaVista to using Google’s search engine in the 90s. It instantly showed us the potential of a search engine to rank and present information, and as a result, people didn’t go back.”

Netflix, smartphones - also step-change technologies that almost instantly evolved the way people behaved because they showed a fundamental understanding (powered by AI) of their users and how they wanted to interact. And so it is with Generative AI. A tool like ChatGPT fundamentally rewrites (pun intended) the rules for human-AI interaction. Rather than searching a term and having relevant results returned to you by an algorithm, that algorithm now has a voice. You can ask questions and refine answers in exactly the same conversational way you would when talking to another person.

“ChatGPT has gone through a very fast hype circle,” says Greg. “But most people have played around with it by now, and they know it has limitations - it’s not always right. But I think that means people will begin to trust it quickly, because they have an idea of what it can and can’t do.”

Greg notes that the real distinction between the new models underpinning the cohort of Generative AI platforms and those that came before them, is the amount of training data available. He notes there is a very apt analogy between what we’re seeing now, and our own human development.

“These are neural networks, and humans are neural networks too. Something that parents know is that children develop in stages - for a long time they’re learning, learning, learning, then all of a sudden they can talk fluently; and to some extent that’s what we’re seeing here.”

LexisNexis has been working and experimenting with various AI models in products since 2018. These include models from Google, and earlier GPT models (such as GPT3) to bring the best of these technologies to bear in legal solutions. “Fundamentally, the law is centred on language,” says Greg. “And that’s why we can very quickly determine that this technology will impact the way we practice law.”

There are immediate benefits presented by the use of large LLMs in law. The first is having a competent AI reviewing, drafting and editing. LLMs also have the ability to change tone, and can even match tones to existing documents. They can rapidly shift the way lawyers conduct research, by not only returning the most relevant results and highlighting relevant parts within, but also providing synopses or summaries as to why a particular result is relevant. Even the basics of client communication - like sending letters or emails - can be made significantly more efficient with help from Generative AI. AI requires human oversight - and the benefit is in process efficiency, not in the farming out of work - something of great value.

“You’d be hard pressed to find someone in the legal industry who doesn’t want more time in their day,” says Greg.

But there are limitations to what the technology can do - the first and most commonly discussed is an accuracy score of about 75-80%, which is certainly not enough for a lawyer to place too much faith in the output without thorough review. The second, says Greg, is currency. “So if you think that the most recent GPT model was finished at the end of 2021, there’s 18 months of data that is not in there, so when you ask it current affairs questions it doesn’t know the answer. In law, where currency is extremely relevant.” He also notes that the models currently in the headlines are trained on public datasets, which means they don’t have access to a vast amount of subject-specific data which sits behind paywalls and firewalls.

Looking to the future, Greg is both upbeat and wary about the prospects for this exciting technology to be integrated into more legal solutions.

“I think we’ll see good technology come out quickly, but in the short term I think a lot of startups will use it to provide quick point solutions that will ultimately disappoint. You’ll also see companies like us who will share solutions that have been developed with customers. In our case, we’ve used both our technology and our content strength - which we’ve built for decades - to create solutions that are far more robust and actually meet customers’ needs.”

Greg notes that the real key for firms looking to put the power of AI to its most effective use is to get their content into formats and locations that can easily be used by the AI. The benefit of using built-for-purpose solutions, such as LexisNexis, is that the data hygiene is already done - all that remains is for users to familiarise themselves with the tools and get to work. Greg says that this will be the key edge in the future.

“These capabilities are coming fast, and it’s not that they’re going to disrupt law firms, it’s that the law firms that are using them more effectively are going to be much more competitive. And therein lies the challenge.”

This article is taken from Episode 1 of the AI Decoded, Legal Talk podcast series. Listen to the full episode here

At LexisNexis our purpose is to advance the rule of law, this purpose guides our actions and so when developing solutions we ensure that all of our developments are in line with RELX’s responsible AI principles: relx-responsible-ai-principles-0622.pdf. These principles ensure that our solutions develop in line with our values and maintain our stance as a thought leader in the market.

You need to be a member to post comments. Become a member for free today!