← Texas Family Law Insiders
Texas Family Law Insiders
Carolyn Elefant | Leveraging Artificial Intelligence in Family Law: Tools, Risks, and Policy
00:47:24
Loading summary
Transcript70 lines
- [00:00]
A
And I think it frees lawyers up to really focus on the creative process and exercising their judgment, which are things that I really doesn't do. I mean, whether it will do it in 20 years from now, maybe, you know, that's too far away to understand. But right now I think it frees lawyers up to focus on the kind of work that we enjoy doing and that we went to law school to do.
- [00:22]
B
You're listening to the Texas Family Law Insiders Podcast, your source for the latest news and trends in family law in the state of Texas. Now here's your host attorney Holly Draper.
- [00:34]
C
Today, I am so excited to welcome Carolyn Elefant to the Texas Family Law Insiders Podcast. Carolyn is the founder of the law Offices of Carolyn Elefant, a national emergency and environmental justice firm. She launched my shingle.com, the longest running blog for solo and small firm lawyers, in 2002 and is a recognized voice in legal innovation and tech. She's the author of Solo by Choice and Social Media for Lawyers, and she frequently speaks on Modern law firm Ownership, Generative AI and Advancing Women in the Law. She founded the Lawyer Mom Owner Summit and has received numerous honors including the 2023American legal technology Lifetime Achievement Award. Thank you so much for joining me today.
- [01:16]
A
Thank you for having me. This is great.
- [01:19]
C
So why don't you start and just tell us a little bit about yourself.
- [01:23]
A
So I have owned my law practice for, I think it's been over 30 years. I think once you get into the third decade, you stop counting. And I started my career at government and then worked for a larger firm and I started my own firm because I was told I wasn't partnership material. But I always like to say I wasn't associate material because I didn't want to do the work on the projects that I was being told to work on. And that kind of is not a good thing when you're an associate. So I started my firm thinking it would just be a gap holder on my resume. And I discovered that I liked it, that it was flexible, that I could take the kinds of case as I wanted to. And so I continued and I guess sometime 10 or 12 years into it, blogging came on the scene. And that's when I started the blog my shingle.com to share my experiences about law firm ownership and to see if there are ways to encourage other people who are unhappy with the law to consider opening their own firm and seeing if that would be a better fit for them.
- [02:20]
C
So tell us about my shingle.com and what attorneys can find now through that site.
- [02:28]
A
So right now for the next week or two, the site is under renovation. I'm in a partnership with LexBlog which is on a mission to help preserve soul and small firm, really to preserve all lawyer blogging. But so it's a little bit rough right now, but it basically has blog posts about the details of starting a law practice. But also really more importantly to me is kind of the role that solos and smalls play play in the legal ecosystem. We hear so much about big law, everything. It seems like, you know, when you talk about legal, you're talking about the problems that are experienced at big law. But 70% of attorneys practice in solo and small firm settings. And I wanted to talk about the issues that affect solos and smalls, like whether I guess when I started the blog I did some posts about whether ethics regulators were disproportionately targeting solos. More recently I've done posts about people starting firms in this current political climate to be able to fight for justice or separate themselves from a more toxic environment or environment where there were people who they didn't agree with the prevailing views. And so I talked about again a lot of the how and I guess more recently I've been talking about AI, how solos and smalls can use AI, how to get started with it very easily because it can be so overwhelming. And yet it's something that I think is so vital for the survival of solo and small firms to be able to understand AI so that they can stay abreast of their big firm colleagues.
- [04:03]
C
So, and we're recording this right before Thanksgiving, but it will actually not go out for probably another month or so. So it sounds like by the time people are hearing this, you're website reno should be complete.
- [04:15]
A
Okay, great.
- [04:17]
C
And so speaking about AI, you and I had actually met at the Lawyer on the beach conference earlier this year in May. And you spoke there about AI and it was just the best presentation I've heard about AI ever. I'd been to many presentations or listened to many blog posts about it or not blog post, but podcasts. And it just always seemed so over my head and I'm like, I am too old to be learning this. This is not, you know, I hope the transition is slow because I'm never going to be able to do this. I don't want to learn it all the things. So after your presentation, it was just such practical good advice for somebody who wasn't on the cutting edge of AI and really didn't know how to use it or anything like that. And I'm proud to say that now I feel like I'm using AI quite a. I know I've just recently created policies for my firm with the help of the materials that you provided. And, you know, it's. I'm excited to see what we can do and where it can go. So that's why I wanted to have you on the podcast today, so you could kind of share some of that wisdom with the family lawyers in Texas and probably any other lawyers that happen to pop onto this to listen. But let's just start with why AI should matter for your typical lawyer.
- [05:43]
A
I think there's a bunch of reasons, but I think mostly it should matter because it enables lawyers to amplify their reach and do more with the resources that they have. And I think that as we progress into an aih, attorneys are going to see that there's a lot of other firms who are using AI that are producing more content or finishing documents or filings more quickly. And so lawyers have to understand how to. How to use that. And I think it really also helps. It helps lawyers address a lot of the things that they don't like doing things that take a lot of time or things that are kind of grunty or create sort of a mental load. And I think it frees lawyers up to really focus on the creative process and exercising their judgment, which are things that AI really doesn't do. I mean, whether it will do it in 20 years from now, maybe, you know, that's too far away to understand. But right now I think it frees lawyers up to focus on the kind of work that we enjoy doing and that we were went to law school to do.
- [06:51]
C
I was listening to a podcast recently about law firm management and AI, and they predicted that within the next five years, jobs in the legal field would decrease significantly. I want to say they said 60 or 65% reduction in the number of jobs out there because AI was going to be taking on such a big role. Do you agree with that? Where do you see it going?
- [07:18]
A
I think there probably will be a shift in jobs. I know that there are projects that I once outsourced to freelancers, small motions to compel, and more routine matters that I just don't do that anymore because AI can do that. And if I. And my firm is set up with more of a distributed model. So I work. I work with contractors as opposed to staff. But if I had an. A junior attorney who was just doing those types of motions to compel in more routine litigation or transactional matters. I probably would find them, they probably would be phased out. But I think that a lot of these people who do get phased out for these junior positions may move on. They may decide to start their own firms and go into practice on their own. And so there's still the demand for attorneys and they may be able to offer services at lower cost and so they may be able to find employment in that capacity. So I think there'll be a shift. I mean, I think overall there probably will be a reduction in attorneys, but I don't think it's as drastic as people are predicting. I wouldn't say 60 or 70%. And I think it will be. It's not going to happen overnight. It's not going to be like, you know, like in 2008 when 14,000 people were laid off essentially overnight. I think it'll be more gradual. People will come to be phased out. People move on to different types of areas of practice.
- [08:40]
C
Many lawyers still feel intimidated by AI or aren't sure where to start. What is the most practical way for an attorney to get started in learning about and using AI?
- [08:54]
A
So I think the first thing is to focus on using IT for simple tasks. If you go into any of these online Facebook groups that talk about technology, you're going to see people doing these unbelievable automations, like 80 step automations or things like that, that is just over the top. That will cause people that's very intimidating. So I think just using it for basic things that you've been wanting to get to and you never have, like for example, you said you were using it to help draft policies. If you don't have a law firm policy or if you don't, if you want to put a system in place like a checklist for what you do before you make a filing at the court of appeals, just talk to the AI and have it do that. It's a very simple way to do it. It's like having a conversation with an associate or a paralegal where you're telling them to make a checklist. So I think for things like checklists, creating checklists and systems, I think for writer's block, if you have to respond to, if you've gotten a very nasty email and you don't know if you can control yourself, you ask AI to assist you with providing a more polite response. And then also for coming up with ideas, brainstorming ideas for marketing, things like that. And so they're kind of low hanging fruit, they're low risk, you don't have to worry about the hallucination problems because you want to, of course, check your work, but it's a lower standard if you make a mistake in a blog post. It's not. Doesn't carry with it the same consequences if you cite a fake case in a court brief. So I think those are some ways that you can start getting accustomed to. And once you start using it, you start thinking of other ways that it could be a value.
- [10:31]
C
What do you think are the top three skills lawyers should learn to become effective users of AI?
- [10:38]
A
That's a good question. I think one of the skills that they should learn is to be able to communicate clearly what they want. The more information that you can give to AI in terms of the background context and the steps that you take that you might take in carrying out a task, the better the output is going to be. So it doesn't necessarily. I know people talk about prompt engineering as if it's some fancy technical thing, but prompt engineering is really just communication. Think of how you would communicate an assignment to a law clerk, and that's the way you would want to communicate it to AI to be able to get a good response. So I think focusing on clear communication of your ideas, I think the second skill you would want to have is it's helpful to have substantive knowledge of your practice area. Even if you're asking AI to do research, just to be able to have that ability to look at the output and say, this doesn't sound right. So maybe it's more critical thinking skills that are important, because sometimes you're going to get a response that just doesn't seem right and you can't say why. And then when you check it out, you find that the AI has hallucinated or done something incorrect. So I think maintaining your critical thinking skills helps you to better assess and verify the results of the AI. And I think the last skill is just curiosity, just being curious and having an open mind about the things that AI can do, because it's really. I'm always amazed when I speak at different events, hearing the myriad of ways that lawyers are using these tools. And every time there's always going to be like at least three or four people who are using it in ways that never occurred to me. So I think being continually curious is also really important.
- [12:25]
C
So I love that you hear about all these different ways lawyers are using it that never even occurred to you. Can you give us some examples of ways that lawyers are using AI that people might be able to adopt quickly and build into their lives and Practices.
- [12:44]
A
So I attended an AI conference last week, and there was one attorney who has a master prompt for his entire firm. So essentially what he's done is he's put together a prompt that has different sections. It has sections on what kind of voice and tone his firm has, what the firm's mission is. It has substantive information about different marketing tools and about his different practice areas. And he just uses that master prompt to provide context for every subsequent prompt he does. And so as a result, he doesn't have to keep repeating the same things over and over again. I think his master prompt is like 100 pages or. I mean, he's built it up over time and I just thought that was incredible. His name is an attorney named Jay Ruane. And I just thought that was an incredible use case to be able to have the AI serve as the overarching backbone for your law practice. I think there are other people who are using AI to evaluate the effectiveness of their intake. They're recording intake calls and then creating. Asking AI to create rubrics for assessing the effectiveness of intake and using it to help improve the performance of people who are doing intake or doing sales calls. One way that I used it that I've demonstrated that's usually popular is that I've set it up to be able to simulate an oral argument. And so I upload the briefs from the case and I can have the AI act as the panel of judges. And I actually used that to help a colleague of mine do a moot court back in January. He was very self conscious about responding to questions when I asked them. So I set the AI up for him and he used it and he did a really good job.
- [14:26]
C
So I love that never occurred to me. Now my. I do appellate law as well, and oral arguments aren't that common. So I haven't had one since I started implementing AI. But I have a couple that cases that could get set. And now I'm like, okay, I'm absolutely going to do exactly what you said. Are you doing that with ChatGPT or what are you using to do that?
- [14:50]
A
So that one I've used ChatGPT for just because it was one of the first platforms to have the best voice interaction. I'm not sure which other platforms, I guess Gemini will talk to you now too. So since I started using it for using ChatGPT for that, I just continued using it because it had the voice interaction capability.
- [15:15]
C
So it will listen to you making your arguments and then it will verbally speak the next question.
- [15:23]
A
Right. So I Instructed it at the beginning. I also uploaded a transcript from another oral argument. And so I said, chatgpt, I want you to play a panel of judges. I've uploaded a transcript to give you an example of how a panel would ask questions in this type of case. I want you to, you know, to simulate how the argument goes. So do the opening with the OIA to put me in the right frame of mind and then ask me questions from the brief and evaluate my performance. And it worked pretty well. The one area I did test it, I tried being really obnoxious, like interrupting it and saying, oh Judge, you're a stupid idiot. And it, it didn't scold it. A real judge would have kicked me out of the courtroom. It just says something like, you have to be a little calmer or this not, you know, that's not an appropriate response. I mean, it was not as harsh as it could have been. But I'll, I can send a link to, to just to one of the demos that I did of it. And yeah, and it goes back and forth. And then, then I would stop and could you evaluate my response? Or I'd say something like, can you give me a model response? So I think it's helpful. Again, it doesn't necessarily obviate the need for having a human moot court, but sometimes if you're in a small practice setting or if you're on a very limited budget, you don't always have the opportunity to do a live moot with colleagues or with other people in your practice.
- [16:46]
C
What about in the litigation context? Because a lot of our listeners are family lawyers that are doing traditional going to court hearings, trials, that sort of a thing. What are some easy ways that people can use AI to help with those?
- [17:04]
A
So it definitely can help to organize discovery responses. If you've sent out discovery and you get responses back, you can ask AI to create a spreadsheet and put the responses into a spreadsheet. You can ask it to go through and help with, with objections. You can either train it on similar cases where you've done similar objections, or else what you can do is you can kind of look at the responses you've gotten. Just go through and say, I want you to object to questions 1, 5 and 7 and give an explanation. It can do that. It can draft motions to compel, it can identify trends or, you know, help put together reports. I guess if you've gotten like a lot of financial documents, you can ask it to put together financial picture of somebody's assets. So it can do A whole bunch of things and is particularly good at that, at organizing that kind of factual information.
- [18:03]
C
So you mentioned training it. How do you train it?
- [18:08]
A
So there are a couple of ways that you can do that. One thing that you can do or one thing I've done, I do some administrative litigation in my energy cases. We don't have traditional discovery requests, but we have data requests with the same concept. And so what I did was I set up a, I set it up in Claude, so it's called Project, but you can also set this up in ChatGPT as a, as a custom GPT. And I uploaded sample data requests, I uploaded the administrative order which identified, you know, how the, what the request should be like and what the formatting should be. And then I, I think I may have given it one or two other examples and then I would ask it to respond based on the knowledge in its knowledge base. And so that's really what the training was. And for a custom GPT, and then I would give it special specific instructions. I can't remember exactly what the instructions were. You give it some specific instructions and that's really what the training is. And the more you use that same tool over and over again with those documents and I mean, it really does get better. And I don't know exactly how that happens, but it just continues to reinforce itself and it just gets, I found it to get much better over time. And then you also want to correct it. If it gives you output that's not satisfactory or not consistent with what you wanted, you tell it, you know, this isn't the way that you were supposed to do it. You should have looked at this particular regulation or this particular rule and follow that. So that's kind of how the training process works on sort of the smaller scale level. I know there are professional or tools that are developed for legal which supposedly have that baked in already. They were supposedly trained on those types of things. I don't think there's anything specific for family law, but I know that in the personal injury space there are a couple of different platforms that have been developed to help personal injury attorneys with the discovery process.
- [20:15]
C
So you mentioned a couple times about custom GPTs. Can you explain what that is, how you create one, and why that's better than just asking every question in the main GPT of chatgpt.
- [20:31]
A
So what you can do is when you open up ChatGPT on the left bar, there'll be something that says explore gate. And it's kind of hidden. Like some of these things are counterintuitive but there's an item that will say explore GPTs. And if you click on that, you'll get a menu item that says Create GPT. So you click on Create and then you'll get an interface where you can upload documents, you can explain what the purpose of this GPT is and you know how you want responses formatted and just whatever details you want to provide. And I think the reason that it's better than using the general purpose for two reasons. I think first of all, you can be very specific about your instructions. So if you want the responses formatted, you know, single space, you wouldn't want to put that as a general instruction for everything you do because you might have things that you want double spaced or something. And the next thing is, and the most important thing is that you can upload those documents and you can create a knowledge base which it can always refer to. And I think that's really the reason why you want to use either the Custom in GPT or it's called Gems in Gemini, or it's called Projects in Claude, because you can upload documents as an example that create this kind of mini knowledge base. And so you're getting very narrow information that's specific to the tasks that you're performing.
- [21:57]
B
This episode of the Texas Family Law Insiders podcast is sponsored by the Draper Law Firm. Providing family law litigation in Collin, Denton and Dallas counties and appeals across Texas. The Draper firm has represented parents in cases before multiple courts of appeals and prevailed in the Texas Supreme Court in one of the biggest parental rights cases in Texas history. For more information, visit DraperFirm.com or call 469-715-6801.
- [22:28]
C
What are the risks associated with doing that, uploading documents and putting that information out there?
- [22:36]
A
So I just there. As you know, whenever you go to different bar sessions, everybody says don't upload information. And I finally did a deep dive into the terms of use for ChatGPT. So if you're using a free service, you're putting a lot of stuff at risk. So you don't want to be using a free service for a law firm office because you just don't know what's going to happen. I guess one of the risks is that ChatGPT will train on your materials and if you've uploaded confidential information, it could theoretically find itself out on the, on the Internet. But if you have a business account or an enterprise account, it really does provide substantial levels of confidentiality. They're contractually protected. So there's a services agreement in place that states that ChatGPT won't train on your materials, that bank grade encryption level is provided. And that's really even more, I mean, the ABA still allows us to use unencrypted email. So having encryption for all of these files is something that is really beyond what we're ethically required to do. And then you can also, with the business accounts and the enterprise accounts, you can delete things after you've done them. So if you're uploading information that you're just, just even in spite of all these protections, you just feel uncomfortable about having it there. You can run a chat and then you can delete it and it will be eliminated from the ChatGPT servers within 30 days. And you can, even if you get an enterprise account, you can even do it faster than that. You can have it deleted within a matter of a few days if you have real concerns. So I feel like with all and all of that, all of that applies to all information except hipaa. There's a little bit of an exception, I think, for if you're a personal injury attorney, you're dealing with HIPAA data, you would need to have an Enterprise account under ChatGPT, because ChatGPT's terms of service, say, or OpenAI's terms of service, say that unless you have an enterprise account and a health care addendum, they won't process HIPAA protected information. So that would be the one type of exception where you'd need to bump up to enterprise. But I think for everything else, the business accounts, which start at like $25 a month, really provide. I have to say I was kind of surprised myself. I thought the privacy protections were relatively robust.
- [25:01]
C
Are there any areas where lawyers should avoid using AI?
- [25:06]
A
I don't know that I would necessarily say avoid using AI, but avoid relying heavily on it. I think AI has been shown to be much less reliable if you're researching issues of first impression or where there's not a lot of reported case law or a lot of information about it. And in fact, I just saw an article recently a couple days ago that found the same thing was true in the sciences. If there was an area that wasn't well researched, ChatGPT's or AI's results were a little bit less relevant. And I think the other place that AI where you maybe wouldn't want to use AI or rely on it is for issues of, of judgment. So if there's two things that are, you know, two potential arguments or two potential approaches that are right, but one of them Makes less sense. But you know, as an attorney that one of those outcomes makes less sense because of your experience. You probably wouldn't want AI to make the final call on something like that.
- [26:15]
C
Earlier you mentioned that there are people out there using AI to automate a lot of tasks. What are the easiest parts of a law practice to automate with AI without needing coding or technical skills.
- [26:31]
A
So I think some of the things that you can automate very quickly are, for example, if you follow a lot of news stories, you could set AI up to pull those news stories and provide you with five social media posts about those news stories or five blog posts in your email every day or in a spreadsheet. And that's something that's relatively simple. Depending on the tools and depending on how your court system works, you could use AI to download and rename files that come through from the court. But there's some courts that provide that information in a way that's very easy to automate. There's some that do it in a way that might be a little bit more difficult. But I think the easiest automations are ones where you're just asking it to grab items from a newsfeed or go through your email, and every week, you know, send you an email about different, you know, news items or something that you received.
- [27:31]
C
So when we're talking about creating automations, is each thing going to require a different type of software or a different program to create those automations, or is that something that can be done just with ChatGPT?
- [27:45]
A
So there's a lot of different ways to provide to do automation. ChatGPT did just adopt something called automations, where you can set up. You can set up automations in ChatGPT, but you can use other tools like Zapier, Make.com. i use one called String.com, which is very easy because you can. Instead of even having to, like, set things up or link things together, you can just tell it what you want automated, and it walks you through the entire process. So there's a bunch of different platforms you could do it with. I haven't used the chat GPT1 yet, and it's still relatively new, but I think probably by the time this podcast comes out, it'll be in full bloom.
- [28:28]
C
I'm gonna need to try string.com because I tried to use Zapier and it was beyond my skill set from a.
- [28:33]
A
Tech standpoint, then you should definitely try the string.com because I feel the same way and I get frustrated very easily. And so I. I don't you know, stick it out to play around with these things. But string.com, you will be able to definitely, you know, automate something relatively simple like maybe pulling court cases and summer summarizing them or something like that. You can just. And it walks you through everything you have to do. So, so that one is a good gateway, I think, to automation and I think for really, you know, I think as we get more into the automation age where you're automating things like, you know, pulling court documents and filing them, I mean, I think in those situations you probably want to work with a professional or really want to know what you're doing because they're mission critical. I mean, if a court order comes in and your automation doesn't work, that's not going to excuse you when you tell the court you didn't see the order because, you know, zapier, you set zapier up the wrong way. So those are definitely situations where working with an expert can be helpful.
- [29:38]
C
Have you come across ways that we can use AI to reduce email volume and help us manage our inboxes? Because I personally know that I'm looking at my email right now and there's probably a thousand unread emails sitting there.
- [29:55]
A
So I know some people use. And I've, I actually have meant to look at it earlier this week a service called Fixer, which helps manage, it's an AI tool to help manage an inbox. I don't know yet. I, I mean you may be able to again set up some automations with some of these, you know, the more general purpose platforms. But I think this tool, Fixer, is AI powered and can help. It's gotten very good reviews from people who are flooded with emails.
- [30:27]
C
What are some ways we can use AI to help improve client communications?
- [30:33]
A
So I think one of the best ways to use AI is, or at least with client relations, if not communications, is recording conversations with clients, of course, with their consent, not secret recordings. I know that when I have an initial intake with clients, I have a number of clients who have opportunities to have solar built on their property and they have to sign a lease and there's always certain things they're concerned about, you know, like they might have sheep that they don't want to be impacted by the solar or they have a driveway they don't want blocked. And those are the kinds of details that when I just talk to the client or I take notes, I either forget or I can't read my notes and then I have to go back to them. Or I give them something back and they say, I forgot about my sheep in this agreement that you marked up. So I find that recording those conversations and having a transcript and being able to input it directly into agreements has really been helpful. And the clients really appreciate having that kind of written record of their communication. So I feel like that's definitely one way that it can improve client relations. I think another way that it can is, you know, again, you may be able to set up some type of automation where if there's something that happens in a client file, to be able to automatically send an email that's already been drafted explaining to the client what happened in the file, or at least to ping you so that you can see what's changing it. With some of these automations too, I'm not really trusting of them being automatic just yet until they're more proven. But just to be able to help clients stay on top of their cases. And then I think you can also use it to prepare summaries and things that are more in layperson terms to help clients better understand their matter without having it take a lot of time to, you know, to put together.
- [32:32]
C
What do we need to disclose to clients about our use of AI?
- [32:37]
A
So that's a really good question. And there's kind of different authorities. I know that the ABA Formal Opinion 512 Governing Ethical Use of AI says that if you're going to input confidential client data into an AI system, that it has to be disclosed. Now, that opinion came out several months ago where there were already more serious concerns about using AI tools and compromising confidentiality. But that's what formal opinion says. You'd have to consult with your individual state bar rules to see if they have anything on disclosure. I think there's a couple of bars that take that approach. So they say if you're going to upload sensitive client information, that would be something you'd want to disclose to them. Otherwise, I think the disclosure is really just a judgment call. I know I like to tell my clients that I use AI because I like to portray myself as being a very tech, tech forward law firm. And so I tell them that I use and of course, you know what safeguards I have in place. And then of course, if you're using AI to be more efficient and you know something that once took you 10 hours, takes you one hour, and you bill the client for 10 hours, well, that's not something you're supposed to be doing. You would want to disclose that as well.
- [34:00]
C
Speaking of, you know, things that used to take 10 hours that now take significantly less time if you're using AI. So in as family lawyers generally, at least in Texas, we are all billable hour people. One of the things that I see AI doing is erasing the billable hour and turning us into more of a flat fee type situation. I have a, you know, we use co counsel with Westlaw, which is their AI tool. And you know, I had a case where I plugged in the two transcripts, the courts ruling and over a thousand pages of exhibits and asked it to give me certain information for an appeal. And in within 10 minutes it had spit out information that would have taken me 20 hours to come up with on my own, with links to each exhibit where it was talking about that or to the cases that it was citing to, so you can verify everything. And so that just really got my wheel spinning about how fast past this could change the landscape of family law. I'm curious what your thoughts are about that.
- [35:05]
A
I definitely agree with that for family law because one of the objections that you always hear from family law attorneys with regard to flat fees is, oh, I don't know how long this is going to take because my opposing counsel might file 75 motions or I don't know how much discovery there's going to be. I don't know how much follow up I'm going to have to do. So I can't really limit myself to a flat fee because I might wind up on the short end of it. And I think a lot of that work can be done much more quickly. And I think that a lot of the fees in family law cases do wind up being generated because you had all of that work to do. I mean, if you were to get a document dump, you have to go through it. You can't, you know, not do it because you have a flat fee in place. So I do think family law attorneys in particular should want to be thinking about ways that they can at least incorporate flat fees for certain parts of their practice and figuring out how to, how to structure that. Because clients are going to start, you know, clients are going to start asking about that too. And if they get a bill for, you know, 20 hours of document review, they're going to wonder why it took so long when an AI tool could have, you know, done it in five hours. And there's always, you're always going to have to build in time for attorney oversight. I'm not suggesting that, you know, I'm sure after you had co counsel crunch everything, you still had to go in and check all the links and things like that. But that still doesn't mean that there's not a time savings.
- [36:38]
C
What should law firms implement as far as internal policies related to AI?
- [36:45]
A
So I think first of all it's got to be out in the open, people. I think it's important to encourage attorneys to use AI and to disclose that they're using it to take the stigma away. I think the worst thing that can happen and we see this happen, it's documented in these hallucination cases. I would say in at least a dozen of the hallucination cases, the attorney delegated a project to somebody in their office, a paralegal, law clerk, whatever. They used AI. Everything was filed and then turned out the cases were fake. And the attorney never knew that, that their people were using AI. So I think that the policies have to be very clear that, you know, encourage people to use it, but also to disclose it. I think you want people who are using it to try to keep records of some of the prompts that they're using and the results that they get. Just like when you have people do legal research, you might want them to pull the cases for you and give you a memo not just with their conclusions, but with what the cases are. And I think that, I mean, I think that those are some of the most. Oh, and of course also the last policy is that everything has to be verified. I was actually thinking about that today because I'm onboarding a new person and I'm usually pretty chill when I work with people. But if I find out, if I get a document that has a case that's not verified, I mean, to me that's a termination related offense. I mean, I don't like to be so heavy handed, but there's just too much at risk. So I think you really can't overemphasize the importance of verification.
- [38:21]
C
So I was recently working on our own internal AI policy and using some of the materials from Layer on the beach that you had provided to do that. And I know one of the things included in your suggested policy was about training of the teams of our teams for using AI. Do you have any recommendations for how to train? If there's specific programs out there that would provide a good basis or is this just kind of up to me as the law firm owner? I'm going to show you how to use it and that's it.
- [38:55]
A
I think some programs are being developed. I know I've been developing some programs and I do some trainings for law firms. So There are people who can come and train your team. I know I'm doing a training next month for a mid sized firm in the D.C. area. So there are people who can do trainings and I think it's important when you're having somebody do a training as somebody who's used AI and maybe as an attorney. And then there's, there's a couple of cles. I don't know how useful some of those are for training, but I don't know. I mean, I think that's one of the challenges that we're starting to, to see in this space is it's very different from substantive law where you can go to a CLE and you're kind of, you know, you're done for at least a year. This is really an ongoing thing and I think that over time we'll probably see the emergence of more ongoing programs. I know some firms are also hiring, what do they call them, like AI? Not AI, you know, like Chief happiness, Officer, Chief AI, like somebody who's on their team or maybe acts as a fractional AI person to like review different vendor agreements and provide advice. But I think this is something that's going to be ongoing and I think we're starting to see some of those things emerge now.
- [40:13]
C
I just saw someone post in one of the layer Facebook groups looking for a fractional AI. I don't remember the exact term they used, but officer or something like that. And I thought I had never heard of that. I don't know what it is, but it sounds like something that I need.
- [40:30]
A
Yeah, because it is, it's, it's really hard to stay on top of all of these tools and you're just getting, you're getting emails from vendors all the time and then you know, you've got to demo the tools and look at the tools and you know, you want to make sure you don't miss anything. But at the same time, you know, you still have a law firm to run. You can't turn into a technology officer.
- [40:52]
C
So for those of us that own our own law firms.
- [40:58]
A
Or are responsible.
- [40:58]
C
For rain making, what are some ways that we can use AI to help improve our marketing?
- [41:05]
A
I think AI is a really great tool for that. I think the deep research feature either on ChatGPT or Gemini, I think Claude has a deep research too, can really help you get a lot of resources. So like one approach I did in one of the trainings I did was I was teaching a class on how to start a firm and one woman said she wanted to market her practice by doing speaking engagements. And so we had AI identify different groups in her geographic area that would be appropriate for her to speak at and then provide the contact information for people she could get in touch with and also draft an introductory email. And so it was able to do all of that. And in 20 minutes she had a list of, you know, like 15 places that she could reach out to and an email to send to them. So I think it's really good at helping to get contact information and get you in touch with people. I think I mentioned somewhere I'd written a law review article on AI and energy law practice, and I wanted to send the article to everybody who I cited in it. So I asked ChatGPT to get me all the email contacts, go through the article and find everyone I cited and get them. And it did. And then I sent the emails out and like three, three people thanked me. They said, oh, this is so nice of you to, you know, tell me about this article and send it out. So it helps you really find people. You could do that with court briefs and outreach. You can also just help it have it brainstorm content or help generate content, especially for videos. You can put scripts together and you don't have to read through the script, you know, like a robot. But it's helpful to see a script and see how ChatGPT or how AI is suggesting you communicate the ideas and then kind of putting your own spin on it. But that really speeds up the process because I know that doing short, short videos for me is challenging just because I'll just keep talking and rambling.
- [43:02]
C
Yeah, I've recently started using it. I was convinced by marketing people that I should do lives on the various social media platforms every week. And I was thinking, I just don't want to come up with content. How, where am I going to get these ideas? And I plugged into ChatGPT and I was like, I need weeks of content for lives. And because I use chat GPT a lot, it already knew my law firm, it knew what I do, and even though I don't have custom GPTs, it was able to find that through the various things I've asked it over the course of the last however many months. And within a matter of seconds it spit out, here's 12 weeks of this. And then I could say, oh, give me three more points about this. And it would change it. And it's like, do you want me to put this into a calendar? Do you want me to put, you know, it'll give you ideas of where to go from there. And I think it's just such a huge time saver and I love it.
- [43:57]
A
Yeah, no, it's also great for like, for funnels or follow up email series. I know that when I promote programs it will come up with like a series of seven emails to follow each other up because as you know, you need to follow up with people in order to get them to, to sign up or to be responsive. And so it's very good at doing that too.
- [44:18]
C
So besides ChatGPT, what are your top favorite AI tools that you personally use on a regular basis?
- [44:26]
A
So I like I said I use ChatGPT, I use Claude a lot for professional drafting. So if I'm writing an article or I do some work as a hearing officer where I sometimes need assistance writing my decisions, I use Claude for that, for writing sections of briefs. I really like Claude for that. I use Perplexity for doing things where I need more of a web search. So if I'm doing a presentation on AI and I want to link to different companies terms of service, I use Perplexity. It can like go out on the web and pull that kind of stuff. And then I use some of the AI recording recording systems I use use. If I go to conferences, I use Plod, which is like a little pin on clip that does recordings and then it puts the scripts on my phone and then I can, you know, transfer them to my, the transcripts and then I use Fathom for or the Zoom recording. There's just so many online AI recordings that you can use. I use those also. And like I said, I started playing around with string.com too for basic automation. So that's, those are some of the kind of the suite of tools that I'm using right now.
- [45:44]
C
So if you could give lawyers one actionable challenge that they should complete within the next week to move forward with AI, what would that be?
- [45:53]
A
So the challenge would be first of all, if you haven't used one of these tools to actually go out and use it, and if they're already using it just for fun things, I would just encourage them to find something that bothers them in their law firm. Maybe it's a website that needs an update, maybe it's some marketing materials that need to be sent out or something that needs to be organized and just taking one task that's been bugging them for a while and seeing how AI can help address it.
- [46:26]
C
So we're just about out of time. But where can our listeners go if they want to learn more about you?
- [46:31]
A
Yeah. So they can either come to my law firm website is just CarolineLefont.com or they can go to my Shingle.com where I have the AI resources. I have a whole AI resource page. And then I'm probably these days Most active on LinkedIn and TikTok, both under my name, Carolyn Elephant.
- [46:50]
C
Perfect. Well, thank you so much for joining me today. For our listeners, if you enjoyed this podcast, please take a second to leave us a review and subscribe. Subscribe so you can enjoy future episodes.
- [47:02]
B
The Texas Family Law Insiders Podcast is sponsored by the Draper Law Firm. We help people navigate divorce and child custody cases and handle family law appellate matters. For more information, visit our website at www.draperfirm.
- [47:23]
A
Com.