Legal Leaders Exchange - Podcast episode 20

Demystifying ChatGPT: The Wolters Kluwer Approach to Generative AI

On this episode of Legal Leaders Exchange, ELM Solutions Associate Director of Legal Operations and Industry Insights, Jennifer McIver, is joined by Ken Crutchfield, legal industry thought leader and Vice President & General Manager of Legal Markets for Wolters Kluwer Legal & Regulatory. Jennifer and Ken discuss the profound effect that generative AI is having on legal industry professionals, with a particular discussion of the headline-making GenAI pioneer ChatGPT.

Listen to hear our experts discuss:

• The difference between traditional and generative AI
• Reasons why GenAI seems to be making more significant inroads among legal professionals than most new technologies do
• The definition of large language models (LLMs) and how they work to create new content
• The risk associated with using some GenAI solutions and how legal departments can safely use GenAI without compromising privacy or security
• Why and how Wolters Kluwer’s approach to AI broadly, and to GenAI in particular, emphasizes security and usefulness to clients  

Be sure to follow Legal Leaders Exchange on:

Apple Podcast | Spotify | Google Podcast | Audible | iHeart Radio

Transcript

Greg Corombos

Hi, I’m Greg Corombos. Welcome to “Legal Leaders Exchange”. This podcast series was created to deliver insights on optimizing legal operations for corporate legal and insurance claims professionals. On today’s episode, we’ll be covering a topic that’s been a global buzzword since 2023. We are talking about generative AI, or GenAI, the latest evolution of artificial intelligence that is radically changing the way corporate legal departments review legal invoices, manage claims, and complete other time-consuming tasks more efficiently than ever. But there are several caveats to the proper and ethical application of GenAI in general and ChatGPT in particular —that’s the reason why we brought in Ken Crutchfield, Vice President & General Manager of Legal Markets for Wolters Kluwer Legal & Regulatory, and Jennifer McIver, Associate Director of Legal Operations & Industry Insights for Wolters Kluwer ELM Solutions. Jen and Ken, please introduce yourselves.

Jennifer McIver

Thank you, Greg. And hi, everyone. This is Jennifer McIver and I'm one of the newest additions to the Wolters Kluwer ELM Solutions family. For those listening who don't know me, I'm an attorney, who on a whim, left the billable hour private practice behind and discovered my passion for connecting in-house legal teams with technology. I spent the last 12 years working with corporate legal departments to improve operational efficiency, reduce costs and collaborate with outside counsel. But honestly, that's enough about me today. I'm excited and thrilled to share the mic with Ken Crutchfield, Vice President and General Manager of legal markets here at Walters Kluwer to talk about AI and, specifically at my request, to demystify all the buzz around ChatGPT. So Ken, thank you for joining today.

Ken Crutchfield

Thank you, Jennifer. It's great to be here.

Jennifer McIver

And you know, I know you've been with Wolters Kluwer for a while would you mind sharing a little bit about your background about joining the family here and how you kind of made that journey?

Ken Crutchfield

Sure I joined Wolters Kluwer about four and a half years ago, I oversee the legal markets business, which is the legal research and content business. Vital law is the core platform that we market to large law firms in particular, but we serve the broader market in general. I also have responsibility for legislation, which is an all-in-one legal software product, targeting the mid-market of law departments and corporations. That product is managed out of Europe. And I'm basically the reseller of that product, if you will in the United States. I started my career with LexisNexis. And I've worked at Lexis twice. Bloomberg, Dun & Bradstreet, a startup in the dot-com era, helped sell a company to Wolters Kluwer Tax & Accounting. And I've also worked for Thomson Reuters. So I've been around the block, I have an engineering degree and an MBA and love this industry.

Jennifer McIver

All right, so we've got an engineer with an MBA with an attorney who hated practicing, to be honest with you. The billable hour just – it just did not work for me. And we both crossed paths at Thomson Reuters, albeit by, I think, literally a millisecond. So I'm really excited today. And the work that you've done, I think puts us in a great position to have this conversation.

I'll be honest, I'm kind of techie, I found my passion, right, I can go in I can help with SaaS technology, I can help attorneys understand. I've done a lot of implementation. But when all the hype about ChatGPT came out back in November of 2022, I'll be honest, I was just blown away. I really didn't understand why there was so much buzz. And we recently had a town hall here at Walters Kluwer, our Chief Strategy and Innovation Officer spoke and talked about the history of AI, just really taking it back away from ChatGPT, and the initial development of it in the 50s. And that really got me thinking that we've been doing this for a while. So if we've been doing this for a while, what is all the buzz around ChatGPT? Why now? Why ChatGPT?

Ken Crutchfield

Right, I think November of 2022 was a real watershed timeframe with ChatGPT 3.5 Turbo, and when that came out, it really set the bar in terms of being able to answer a question conversationally and provide really good answers. That's, I think, the thing that was pretty amazing. You know, I remember reading something where it explained a concept and did it – basically the Einstein theory of relativity – but did it it's a poem, and I thought it got incredibly powerful. So those sorts of things, I think, got the attention of a lot of organizations, Meta, or Facebook as it's also popularly known, had done a similar product, like three weeks earlier, called Galactica, instead of the OpenAI ChatGPT. And it basically spewed out a bunch of craziness. And so they actually pulled it after three days. So the fact that OpenAI was able to deliver ChatGPT – that was a big deal, because it really actually worked in a way that was convincing and scary.

Jennifer McIver

So let's take a step back though. I mean, how does ChatGPT – or, I guess it's generative AI – how does it differ from traditional AI, the AI that Wolters Kluwer has been working with for a good 15 years?

Ken Crutchfield

I think the biggest thing is that it creates content. That's my biggest point on it. But I'd also point out that the backdrop of AI has been around us for such a long time, we don't really think about it. Basically, when you go on to Amazon, it will suggest other products that are similar to the product that you're purchasing and make suggestions for additional products to buy that are frequently bought together. We typically use Google or Bing or some other search engine. And those search engines use natural language processing to answer questions to provide quick summaries and to be able to direct you to information. And when you're working on your favorite social media account, and it suggests that you tag a friend like Facebook, that's facial recognition. These are all AI capabilities. I could go on with Tesla and neural networks and Siri and Amazon Echoes that do speech recognition. So it's around us all over the place.

Jennifer McIver

It’s kind of funny. The other day – so I'm a bowler, kind of random here – but I was talking with some friends and said I really, really want to get a new bowling ball. And left the conversation later in the evening. Got up the next morning and everything on my phone was about bowling balls. So, needless to say, I haven't purchased one yet, but I've got all the information I need thanks to this technology. So we're talking about AI and really, again, focusing back onto ChatGPT generative AI creating content: why do you think it's so important? What do you think is creating all the excitement around it now?

Ken Crutchfield

I look at it and I see parallels to the Gutenberg printing press. That's my big point is that the world changed when the printing press was invented, and content became more freely available and accessible to many people. It had geopolitical implications. It also had socioeconomic implications. And from the legal perspective, it created new industries like publishing, an industry that's very familiar to Wolters Kluwer’s roots. And it also created new areas of law, like copyright, because you didn't need copyright law before you could make copies of something. And so, I think there's a lot of parallels that I look at with ChatGPT and generative AI, because it's creating content. And it's going to have geopolitical considerations. It's going to have socioeconomic considerations, as we look at studies from Goldman Sachs and others saying, you know, 44% of all jobs in the legal industry can be automated. There's just a lot going on there.

I should point out that if you look at things from a geopolitical perspective, there's the CHIPS Act that was passed recently, that's basically saying all computer technology, microchips, need to be developed onshore or not in Taiwan or not in China. There are definitely geopolitical aspects of this going on. And this is something that you can't stop now because the technology is out there and it's going to be used by nation states. It's going to be used by people who want to come up with ideas. And it's not just something that you can say, oh, let's do a pause. That's why I think this is so big.

Jennifer McIver

Okay, so help me out here, Ken. I understand the content generation, so I have that. I mean, I'll be honest, I even did a poem for a conference. It was about spend management, it was quite humorous. I started off a conversation or a chat that way. But I hear a lot about LLMs, or large language models. Is that correct? And so, I'm just kind of curious, can you talk about how those play in especially when we're looking into the practice of law and the use here in the legal tech industry?

Ken Crutchfield

Sure. I think a model is just what it sounds like. It describes, it models, it creates an image of what something is. So, a large language model is just what it sounds like. It is diagramming, it is coming up with different relationships in language. So, every single word that is out there on the internet, in most every language, is basically cataloged. And so, if you have a word like “context,” or “conflict,” or “construct,” CON is what amounts to, what's called the token, in a large language model. Each little bit of a word is categorized. And there are things like the sources that we think of for synonyms. The language model just basically creates all these different relationships and catalogs words that are nouns or verbs or other things. It provides context around these.

So, take the word “green,” the word green is a color. But it can also have a environmental reference, like green energy, or it can mean an experience, like somebody who's green at a job. And so, by identifying those relationships and those definitions, you can take a sentence of input into a large language model and it's basically going to match the words and the similarity, based upon all of this cataloguing that's been done around the large language model, and about language itself, to be able to create an output. It basically does things by using what are called vectors, which is just basically math or angles and indices. And it just becomes a very simple math equation to walk through and play numbers together. And then those numbers, you know, map to these tokens and create output.

Jennifer McIver

Okay, I'll be honest, you hit math and the attorney side of me just shut down. So that's the engineering side of you on that, Ken! No, that's really, really helpful. And I think for me, that really takes us to the excitement about including generative AI in our daily practices. It's almost confusing to me, the aspect that lawyers and the legal industry… I mean, lawyers are known – and whether it's true or not – to be a little risk averse, a little, you know, resistant to change. Yet, in our Future Ready Lawyer survey report here at Wolters Kluwer, 43% saw this as an opportunity. And even more so, 73% of legal practitioners responded that they were expecting to actually integrate generative AI into their legal work in the next 12 months. And that's really hard for me to grapple with, because I'll be honest, I've worked with a lot of in house counsel teams, where I have a hard time getting attorneys to actually want to review their invoices on the screen and not hit the print button. So why do you think this is going to make its mark? And what's different with this technology, maybe, than some of the other technologies that have become more prevalent in the last 15 years?

Ken Crutchfield

Yeah, let me hit a little bit more on the large language model piece as a quick bridge into that question, and then I'll answer that question if it's okay. So one of the simple ways to think about generative AI is your smartphone, and sending a text message. If you say “Hey, I am walking…” your phone, if you have a dog, is probably going to suggest “the” and then “dog” or the name of your dog or something else, because it's just working off of some simple basic probability. So going back to the math, but expressing it in a way that's easier to understand: that's basically generative AI. It's very simple. Large language models just do that at massive, massive, massive scale based upon all the information that has been scraped off of the internet by OpenAI or one of the other providers. So that's maybe just a simple way to kind of understand this a little bit more.

But back to your question. I think generative AI is the first thing that I hear law firms talking about where partners actually are interested in the technology. Partners at law firms tend to dismiss technology, I think, a little bit more than they should. But this is one that’s got them sitting up straight in their chairs. And they're asking questions, and clients are asking questions. Some clients are saying “Don't use GenAI on anything that you do for us.” And others are like ”Hey, if you can save some money on my bill, I'm kind of interested in you maybe using GenAI and what you can do as a firm.” So I think that's the biggest thing is that there's a recognition that this technology is going to be more transformative, I'll use that word to describe the changes that are going to happen in in the legal space. Disruption is often thrown around there. Yeah, this is going to be kind of disruptive at one level, but I think most of the change is going to be something that can be embraced by law firms and in house counsel alike

Jennifer McIver

This might be elite, but do you with some of the clients suggesting that their law firms not use ChatGPT or generative AI, could that just be because of all of the news about hallucinations? I mean, I was reading something recently, it was earlier this month in Florida, there was a district court case where an attorney was suspended because they created pleadings and they were frivolous arguments based on fabricated cases. Do you think that that really has clients worried about, when they're asking their firms, whether they're using generative AI?

Ken Crutchfield

I think that's an interesting case all by itself. It's probably one of the inputs. I think the bigger issues are privacy, and security of information, as opposed to hallucination. But the idea that attorneys have done this in multiple jurisdictions and have submitted cases that don't exist, because they used ChatGPT or some other generative AI tool to be able to create a brief or a pleading, and it gave a very convincing answer. But it shows and demonstrates that they didn't do their due diligence and actually read the case because it didn't exist. That's, I think, the bigger issue. It's exposing that there are the attorneys out there that don't read everything that they're supposed to read as part of their job. So I think that's kind of my response to the Florida announcement or the New York lawyer that got caught, you know, submitting fictitious cases. So those are one dimension of the issues.

I think that the bigger and broader thing is that it's just hard to get confidence from a law firm or a law firm suppliers that information that is confidential isn't finding its way into the fine tuning or the actual foundation model of a language model. So if you go to the consumer ChatGPT and start uploading documents into it, those are now part of OpenAI's model. So, if you put a confidential contract and upload it, or if you put confidential, privileged information in, that's there now to be incorporated into other answers that are output for other individuals. That's, I think, the bigger issue that is really kind of pervasive, and there aren't easy answers for it until law firms and vendors have really strong processes that are transparent, that can be understood.

Jennifer McIver

See, I was all for it earlier. Now I'm a little scared. I think that's a good segue, Ken into talking a little bit more about Wolters Kluwer’s history with AI and our approach to AI, especially when we're talking about privacy and the worry. So maybe if we can kind of jump a little bit from the techie talk and talk a little bit more about how long we've been working with AI here at Wolters Kluwer. I think you started in 2019 and we were already there, weren't we?

Ken Crutchfield

Yeah, we've been working with AI for a long time, I think in our analysts presentations in speaking to investors last summer, we mentioned that our digital revenue, 50% of its touched by AI, there are different types of AI that are not generative that had been involved in products. I think LegalVIEW BillAnalyzer has an AI capability. The Legisway product is actually using AI to extract information from a contract to be able to tag and provide attributes so that you know what's in a contract, but also, what's in all of your contracts. So there are different things that have been done with AI that are very good applications of using that technology to do things that are not terribly unlike what you get comfortable with in in the pervasiveness of AI, you know, suggesting, “Hey, do you want to tag your friend because this face looks familiar and we're pretty sure this is Ron.”

Jennifer McIver

That’s amazing. And as we're looking forward, and especially if we're looking to incorporate generative AI, there's a lot of talk out there. I was reading, for example, an ALM, legal tech news article recently about AI washing, and about how a lot of folks are just putting a wrapper on ChatGPT. Really not doing anything too innovative, more repackaging what's already been there. And I think with us at Wolters Kluwer, we approach it a little bit differently. So can you talk about how – as much as we can, obviously – the strategy and our focus on moving forward and really working with the technology?

Ken Crutchfield

Sure. So Wolters Kluwer, our tagline is “When you have to be right” and we serve professionals in healthcare, financial services, tax and accounting, in addition to legal, and these are important areas and professions where making a mistake isn't really tolerated. It just can't be. I mean, think of a doctor, it's just, you know, you don't want to be diagnosing something incorrectly, or prescribing the wrong medication. These are important areas. So, we're not going to be at the, if you will, the bleeding edge of some of these things. But we're going to be responsible and think through, and how we learn from some of the very early adopters in this, and some of the innovators to deliver solutions that work for our customers.

So, one of the things that we're seeing is if you look at different solutions, taking information and asking a large language model to summarize what's in a document, create an executive summary, tell me what the key points are, give me a bullet list, those sorts of things; AI does pretty good and generative AI does a pretty good job. And the likelihood of making something up is very low. So there are things like that, that I think will be applied within the VitalLaw world or others later in this year, that I think are going to be very compelling and that are known. There's actually a term for it called retrieval augmented generation or rag.

That is the, probably the more popular way that organizations and law firms are likely to use generative AI to be able to search their own data that's controlled, that they know doesn't have bad information in it. And then working through the concept of summarizing using generative AI to summarize or extract what's in that type of document creates a controlled environment. So you don't worry about the privacy issues or the security issues of data getting out or being part of a large language model. And using the summarization capabilities in a controlled environment where, say for example, you have your own ChatGPT instance like Wolters Kluwer or other organizations do using Microsoft as the vendor. You control what gets input into those language model foundations and you delete it all each time because it's your version of it.

Jennifer McIver

So that's a great way that we're protecting our clients. And the aspect that you mentioned earlier, everything you upload, especially if you're using OpenAI for example, it's just out there for everybody. But we're definitely taking the steps to protect that information. And moving a little bit about, you referenced LegalVIEW BillAnalyzer. And what I love about that myself is, being a little bit of that risk-averse attorney, is it's actually leveraging more that traditional AI, but also combining experts as well. And I think that's something that is really important to note about Wolters Kluwer, that we have a lot of experts. We have data scientists, and we do definitely have our center of excellence that's focused on AI. So I'm really excited to see what comes in the future. And we work with clients, don't we, to help actually generate the new features?

Ken Crutchfield

That's a great point, Jennifer, I think that one of the things that is important about the way Wolters Kluwer approaches things is that we view AI as a technology, and humans have to interact with it. So we have editors on our staff, that we combine and leverage that kind of human-and-machine hybrid to be able to create better products and leverage our deep expertise. So that's one of the things I think we do very well. I'm spending three, four or five meetings a week talking with customers and sharing some of the things that we're doing around generative AI as we look to incorporate capabilities into VitalLaw, we are engaging with our customers regularly to get their feedback to understand context to be able to test features and say “Would you like something this way or that way and why?”. And we're really focused on making sure that what we create and deliver for customers is going to have fundamental value, and meet their needs, because we speak with them on a regular basis as part of the journey of product development.

Jennifer McIver

I think that's amazing. And that almost sounded like an open invitation to meet with you if anybody wants to see something in our product. But I don't think that's what you meant. But I do love the idea, Ken, that we're working with our clients. And I think that's the most important piece of the journey here is making sure that we're not checking a box here at Wolters Kluwer, that we're actually developing and innovating in a way that's going to make sense for the individuals using our products.

Ken Crutchfield

One hundred percent. And if there are customers out there, or organizations that are interested in understanding what we're doing, reach out. We're happy to talk and listen and understand what the needs are. And to the extent that we can share, we will absolutely share what we're doing.

Jennifer McIver

I'm excited for it. Being new to the Wolters Kluwer team, I have been absolutely moved by the involvement of the product team and everybody in the organization to really, really help our clients move forward. And, Ken, thank you so very much. You really helped me today personally, thinking about generative AI, wondering exactly how it plays in the market and some of that behind-the-scenes techie tech stuff that I didn't understand. So, thank you so very much for your thoughts, and I really appreciate your time today.

Ken Crutchfield

Thank you, my pleasure. It's been wonderful having this conversation and I hope others benefit from it too.

Greg Corombos

Those were the words of Ken Crutchfield, Vice President & General Manager of Legal Markets for Wolters Kluwer Legal & Regulatory, and Jennifer McIver, Associate Director of Legal Operations & Industry Insights for Wolters Kluwer ELM Solutions. This podcast is hosted by Wolters Kluwer ELM Solutions, the market-leading provider of enterprise legal spend and matter management, contract lifecycle management, and legal analytics solutions. For more information and additional guidance, please visit www.WoltersKluwer.com or call 1-713-572-3282. Please join us for future podcasts on optimizing legal operations and achieving your legal and business goals.