Legalweek 2024: AI and the American Arbitration Association

Episode Notes

Transcript

Laurence Colletti:

Hello, welcome to another edition of On the Road with Legal Talk Network. This is Laurence Colletti and I’m the host for today’s show just being recorded from Legal Week in New York City. And we have not been at Legal Week since 2019. So this is four years later. Wonderful show, wonderful people. And I have a tremendous panel of guests joining me today. We’re going to be talking about ai. We’re going to be talking about arbitration. So I have Bridget McCormack joining me across the table here. Welcome to the show.

Bridget McCormack:

Great to be here.

Laurence Colletti:

And I have Linda Beyea, welcome to the show.

Linda Beyea:

Thank you.

Laurence Colletti:

And then we have Diana Didia, welcome to the show.

Diana Didia:

Thank you. Glad to be here.

Laurence Colletti:

Well, before we tear notes, so you just, you are FFS, which is fresh from session that you presented at, and the title of it is How the American Arbitration Association embraced AI to Manage Complex Internal Case Files, forms, data, and External Workflow. But before we get into it, just want to learn about your biome background. Where do you work, what do you do? And Bridget’s a returning guest. We’re going to start off with you.

Bridget McCormack:

Yeah, it’s great to see you again. I’m the president and CEO of the aaa, and I have been in that seat for almost a year. It’ll be a year and two days. And up until last December, I was the Chief justice of the Michigan Supreme Court. That was a career I had for about 10 years. Before that, I was primarily in legal, academia and a litigator before that. But I had lots of fun conversations with you when I was leading the Michigan courts and it’s great to see you again.

Laurence Colletti:

Thank you so much for coming by and wonderful on social media. Still doing, you were a big Twitter person for a while and now X, but are you doing the Instagram now?

Bridget McCormack:

I am on Insta, I’m on TikTok. I mean, I don’t actually post on TikTok, but I’m on there following because that’s where stuff’s happening. And obviously we’re on LinkedIn quite a bit, but I’m still on Twitter just because stuff happens there.

Laurence Colletti:

Alright, Linda, let’s hear from you.

Linda Beyea:

So I’m the VP of innovation at AAA based in Atlanta, Georgia. And I have been with AAA for 23 years in a variety of roles and have been full-time, VP of Innovation for one year.

Laurence Colletti:

Alright. And Diana, let’s hear from you.

Diana Didia:

Yeah, hi. I’m the Chief information and innovation officer for the aaa. I’ve been with the AAA about 12 years. I’m based in New York City and I’m responsible for everything tacky at the aaa, including software development, networking, innovation as well, and also our data research and analytics and insights team.

Laurence Colletti:

Okay. So my opening question to you, just want to get the skinny here, know y’all have been working on some projects, I understand there’s three products that have come out of that. So if you could just give us the fly by on it just to kind of get everybody up to speed.

Bridget McCormack:

Well, lemme give you the top line and then Linda and Diana will be able to fill in lots of details. But we’ve been approaching generative AI as the disruptor that we think it’s going to be for the legal profession, business law, practice of law. And therefore for those of us who serve lawyers and their clients, we’ve kind of embraced it top down, bottom up. And as part of our learning curve, we’ve built three products that incorporate generative ai. And I’ll let Diana give you sort of the top line on what those are and then we can answer any questions you have about them.

Diana Didia:

Yeah, sure. So the three learning pilots that we’ve executed on, the first one is a clause builder ai. So it is a tool that builds arbitration and mediation clauses and ensures that they are effective and enforceable. And it’s using a chat ux window to allow users to interact with our library of perfected clauses. We also use generative AI to automate scheduling orders. And then we are also working on leveraging generative AI to create a, we’ve already created a filing assistant chatbot.

Laurence Colletti:

Tell me a little bit more about the clause builder. What’s the function for that? Why is that, I guess, how is the consumer going to use it?

Diana Didia:

So we find in our case administration that there are a lot of poorly written arbitration and mediation clauses, which can make enforceability an issue. And so we have expertise in drafting very effective and enforceable arbitration clauses. So using a chat UX that is powered by generative AI makes it easy for users to interact with the tool to ask the specific kind of clause that they want and then they can cut and paste that clause into their contracts.

Laurence Colletti:

Now these three products, as I understand it, are the beginning of some much bigger actions that are coming to be. So what’s the big picture here? What’s the big plan?

Bridget McCormack:

I don’t think anybody knows because I think anybody who thinks they know what’s happening next week with generative AI is selling something that’s probably not worth buying. But we definitely think there’s going to be some big new products and services that we want to be in a position to deliver to our users and also not only to the users we have now, but to potential new users who right now have nowhere to go to get their disputes resolved. You probably know that most small and medium businesses don’t have access to lawyers, they can’t afford lawyers or they literally go legally naked, but they have disputes. And so there are lots of users who could benefit from better, faster, cheaper services. The AAA sits on an enormously large data set that this team long before I got there, hoarded where they’ve never licensed it, never shared it. And so we have an awful lot of data to work with to build some bigger products and tools and services that we think at least some parts of the market will eventually want.

Linda Beyea:

And I just wanted to add, I think AI, unlike anything before, gives us an opportunity to maybe unlock the potential to help parties prevent disputes. So we’ve tried that for many years, we understand the value of prevention, but it’s been too expensive a proposition, but I think that AI creates the opportunity for unlocking that.

Laurence Colletti:

But walk me through that a little bit. I did see that in your slide deck and I thought that was really interesting. I’m a former business owner and in my past life as a lawyer I represented business clients. And so how does this tool or the tools that you’re envisioning help people Ward off future conflicts?

Linda Beyea:

Yeah, well I think with generative AI’s ability to read and understand natural language, that it can analyze a set of documents and identify patterns. And we believe it has the potential that it could identify where there might be mutual benefit for the parties or Flagg emerging issues. And so that these things can be caught early on. So we’ll see. I don’t think we haven’t started testing any tools with that use case yet, but we will be.

Laurence Colletti:

Bridget was kind of broached into my next question here, the learning set. So this data, the large language model that’s coming from, so where is the source of this information coming from that’s feeding this machine that’s generative ai, it’s kicking out a product now. So where’s that information, the data set? Where’s it coming from?

Bridget McCormack:

Well, let me turn that over to Diana. She can tell you that it’s a different data set for each of the products that we have built or test cases. So let me let Diane

Laurence Colletti:

Answer that. There’s no overlap between them or,

Diana Didia:

Well, the trick right now is to pick really small niche data sets to drive learning. So the three products that we’ve developed, the data set for clause builder is a library of perfected clauses. So you can see how that wouldn’t be very large. The dataset for our scheduling order is actually a zoom recording of an administrative call that is considered the dataset and the dataset for our filing chatbot is our rules and FAQs for filing. So right now, using those smaller data sets, we’re learning how to integrate the generative AI into products and solutions. And then we’re getting our data ready. This several decades worth of arbitration data that we sit on. We’re cleaning it and tagging it and maturing our capacity and ability to manipulate that data to be ready for training our own model. Then we would be using it to train our own large learning language model that could be then put to use on use cases such as predictive and prescriptive analytics or early case evaluation or even a digital decision maker. So that would be in the future. But the steps we’re taking now is allowing us to understand how to integrate these tools at all and how they behave and operate. And then it stages us for taking those bigger bets.

Laurence Colletti:

So it sounds like it’s a project that is continuing to develop internally. And I know we talked a little bit about this in the pregame, but the plan I think ultimately is also to have artificial intelligence take on some of the role of the arbitrator so it learns enough and now it understands and it can create answers to problems for people out there that are trying to deal with the conflict. So is that kind of the long-term vision is that we’re going to maybe turn some of the arbitration over to an artificial intelligence model to kind of free up the courts and free up the human resources. And I guess how long would that take do you think?

Bridget McCormack:

I think this is the question that we get most often at lots of meetings, conferences, and interacting with lawyers and arbitrators. We do think that we want to prepared to have a model that can make decisions, at least for users who want to test out the value of their dispute. You could imagine that use case being one the market would want very early, right? Let’s figure out what the value of this dispute is so we can resolve it as quickly as possible. There probably will be other disputes where some parts of the market will be interested in turning them over to an automated decision maker that’s trustworthy and trained on data that produces results that people can have confidence in. And you could imagine what those disputes might be. Document only disputes in caseloads where you see the same dispute over and over again, high volume caseloads. And so we want to be ready to have those services and products when some parts of the market want it. We are absolutely confident that there will be some disputes that people will always want a human decision maker in the middle of right bet. The company disputes and disputes where there’s a lot of people involved, but there’s just an enormous part of the dispute resolution market that right now has nowhere to go. And those automated tools we want to be prepared to deliver.

Laurence Colletti:

Wait, I was thinking about it. We have small claims courts out there for disputes that just probably aren’t worth hiring an attorney for and getting the court system involved. And so people come in and represent themselves and so it kind of sounds similar to that. Like hey, this is kind of a lower hanging fruit where we disagree, we definitely want to figure out what to do, but I don’t want to pay an attorney X amount of dollars. I don’t want to involve the courts system. I don’t want this to take forever. So an avenue, like you said, for the underserved market there

Bridget McCormack:

And how high the fruit is hanging probably depends on the user. You could imagine some businesses where claims that wouldn’t be eligible for small claims court would still be ones that they might be interested in a more efficient automated process. So absolutely that category of cases, but lots of other ones I think we want to be ready for.

Laurence Colletti:

Alright, last question for you. So just, I know there’s a lot going on here. We’ve already made so much progress built three separate products and we’ve got our vision on something bigger and larger coming down the pipeline. So if you could just quickly roadmap it for us, maybe some predictive mile markers and things like that just to leave it with our audience, that would be wonderful.

Diana Didia:

Yeah, I think that we are sort of tied to how these models are maturing themselves. So we’ve already hit against some of the limits of them. So for our clause builder, we’re struggling to get it to have deterministic results and that isn’t on our part. That’s what the model is capable of. And we’re also seeing that there are actually limits, they throttle your usage and there’s big questions about the compute power to run all these large use cases using these tools. So I think it’s a little hard to predict. I think our appetite for doing it is pretty expedited or accelerated. So I mentioned getting our datasets ready and we’ve actually hired a new data scientists and we’re investing in new tools for getting our data ready. And I think our learning is going to meet the moment, if you will, but predicting exactly when. I don’t know. I mean I would hope maybe it’s 12, 24 months, 36 months, but I think it really depends on things outside of our control.

Laurence Colletti:

Well, only time will tell. Thank you so much for joining us today. And so if our listeners out there want to learn more, maybe they want to get involved, maybe they can contribute in some way, how can they find you? They

Linda Beyea:

Find us on our website adr.org and check out our AI lab page on that website and there’s a place where they can enter their contact information.

Laurence Colletti:

Alright, well thank you all for joining us today.

Diana Didia:

Thank you for having us.

Laurence Colletti:

And thank you to our listeners for tuning in. If you like what you heard, please rate and review us in Apple Podcast, Spotify, Amazon music, or best yet your favorite podcasting app. Until next time, I’m Laurence Colletti and you’ve been listening to On the Road with Legal Talk Network.

Speaker 5:

If you’d like more information about what you’ve heard today, please visit legal talk network.com. Subscribe Via iTunes and RSS, find us on Twitter and Facebook or download our free Legal Talk Network app in Google Play at iTunes. The views expressed by the participants of this program are their own and do not represent the views of nor are they endorsed by Legal Talk Network, its officers, directors, employees, agents, representatives, shareholders, and subsidiaries. None of the content should be considered legal advice. As always, consult a lawyer.