Latest / AI Tools for Practicing Lawyers / Episode 020: Pain Points, Grok And The Homeless Summer Associate

Transcript

Episode 020: Pain Points, Grok And The Homeless Summer Associate · AI Tools for Practicing Lawyers

• 0:03
  SPEAKER_00: Welcome to AI Tools for Practicing Lawyers.
• 0:07
  SPEAKER_00: Practical, no-nonsense guidance to help attorneys put AI power to work in their practice right now.
• 0:14
  SPEAKER_03: Most lawyers spend their careers learning the law.
• 0:18
  SPEAKER_03: Today's guest spent hers learning technology, business, aging, and then the law.
• 0:25
  SPEAKER_03: Now she's training AI systems, building legal software, and asking, maybe counterintuitively, whether solo lawyers may be better positioned for the AI revolution than anyone else.
• 0:39
  SPEAKER_03: I'm Ron Dresher.
• 0:40
  SPEAKER_03: I'm Heather Gardner.
• 0:41
  SPEAKER_03: This is AI Tools for Practicing Lawyers.
• 0:44
  SPEAKER_03: And our guest today is the well-esteemed Dr.
• 0:48
  SPEAKER_03: Sarah Kubik.
• 0:50
  SPEAKER_03: Sarah, how are you doing today?
• 0:52
  SPEAKER_02: Good.
• 0:52
  SPEAKER_02: Thank you.
• 0:53
  SPEAKER_02: That was a great opening.
• 0:55
  SPEAKER_02: Thank you.
• 0:55
  SPEAKER_02: I'm humbled.
• 0:56
  SPEAKER_03: Well, how did that happen?
• 0:58
  SPEAKER_03: How did you go down this yellow brick road of education?
• 1:02
  SPEAKER_02: My family is a bunch of academics.
• 1:04
  SPEAKER_02: And so I thought I'd remain in academia with my PhD.
• 1:09
  SPEAKER_02: I've always been interested in older adults.
• 1:11
  SPEAKER_02: And while I was an assistant professor, I had a baby and ended up in a divorce court situation.
• 1:18
  SPEAKER_02: And like many attorneys that come to law later in life, I thought, you know what?
• 1:22
  SPEAKER_02: I'm going to go to law school because I believed I could represent myself better.
• 1:28
  SPEAKER_02: Really, you know, had the opportunity to go part-time.
• 1:32
  SPEAKER_02: That's uh I went to Cooley.
• 1:34
  SPEAKER_02: I wanted to learn from practitioners and really was amazed by, I mean, no one in my family has any legal background.
• 1:42
  SPEAKER_02: So was amazed by the material being presented and just tried to marry, no pun intended, with family law, but tried to marry technology and legal, and that was a while ago, and there just really wasn't a connection.
• 1:58
  SPEAKER_02: And so basically I thought, oh, I'll be a consultant with a PhD and a JD.
• 2:04
  SPEAKER_02: And in my opinion, I think lawyers trust other lawyers.
• 2:08
  SPEAKER_02: So just having the JD wasn't enough.
• 2:10
  SPEAKER_02: So I became licensed, moved to the Chicagoland area because there's a lot more people around here, and I'm a Midwesterner at heart.
• 2:17
  SPEAKER_02: And because I knew family law from my own experiences, that's where I started was in family law.
• 2:22
  SPEAKER_02: And did that for a while.
• 2:24
  SPEAKER_02: Moved into estate planning uh a little while ago because, like I said, my interest is really in the older adult population.
• 2:33
  SPEAKER_02: I just like this area better.
• 2:35
  SPEAKER_02: And then along came AI, right?
• 2:38
  SPEAKER_02: And the gig economy.
• 2:39
  SPEAKER_02: And I was inspired by many people around me that are doing this professional gig type work.
• 2:45
  SPEAKER_02: And I saw these AI trainer roles on LinkedIn, and I thought, well, I'll just applaud, not knowing what they were.
• 2:52
  SPEAKER_02: And I've been doing that for over a year and a half now, too.
• 2:55
  SPEAKER_03: So what's your practice actually look like?
• 2:59
  SPEAKER_03: Do you have do you have a do you have a brick and mortar office?
• 3:02
  SPEAKER_03: Are you virtual?
• 3:03
  SPEAKER_03: How do you how do you manage the physical aspects of your practice?
• 3:06
  SPEAKER_02: Right.
• 3:07
  SPEAKER_02: I do not have a brick and mortar full office.
• 3:10
  SPEAKER_02: I'm in a co-space.
• 3:11
  SPEAKER_02: And so I do have a physical address, but I prefer to work from home.
• 3:15
  SPEAKER_02: And it's, you know, it's two miles from where I live.
• 3:18
  SPEAKER_02: If I want to make clients there, I can, but a lot of my clients are older and I will come to them.
• 3:23
  SPEAKER_02: I will drive to them.
• 3:24
  SPEAKER_02: And they're you make house you Yeah.
• 3:27
  SPEAKER_02: Why not?
• 3:28
  SPEAKER_02: I mean, yeah.
• 3:29
  SPEAKER_02: I don't I don't really love the traditional practice of law, how it is, where you know, you're this attorney standing behind a law library, crossing your arms, looking all serious.
• 3:40
  SPEAKER_02: So I practice differently, and one of them is house calls.
• 3:43
  SPEAKER_03: Let's talk a little bit more about how you're using AI in your practice.
• 3:48
  SPEAKER_03: You know what?
• 3:48
  SPEAKER_03: I've studied your LinkedIn, I looked at the articles you quoted, I read through your website, and I I'm getting the feeling that you're almost more into engineering the law practice than practicing law.
• 4:03
  SPEAKER_03: Right.
• 4:03
  SPEAKER_02: Is that a fair comment?
• 4:05
  SPEAKER_02: It is, but it's always buffered by I want to see how it works in practice.
• 4:10
  SPEAKER_02: You know, it's sort of like you get a degree and then you learn the reality.
• 4:14
  SPEAKER_02: So you may engineer something, but until you put it into production and use it, and I'm a solo, so I'm using it, you see where it breaks.
• 4:21
  SPEAKER_02: But yeah, I really am fascinated by AI and specifically how solo attorneys and small firms are using it.
• 4:28
  SPEAKER_03: So, what are you seeing regarding adoption by small firm lawyers?
• 4:33
  SPEAKER_02: Well, like I said, I started in family law and then I worked in estate planning, and there was very little use of technology in general.
• 4:41
  SPEAKER_02: Excel was considered remarkable, you know, and I think there's such a division between what we see online and what the solos and the smalls are doing, I'd say offline, who aren't on LinkedIn, of course, barely in the State Bar Association listservs, maybe doing email, but I don't see an adoption really strongly.
• 5:05
  SPEAKER_02: Well, I think a lot of people have Clio, that's a big name out there, but I don't see they they they have heard of ChatGPT.
• 5:13
  SPEAKER_02: I think that's for sure.
• 5:14
  SPEAKER_02: But I think it's more of a personal thing.
• 5:16
  SPEAKER_02: You know, like their kids have taught them, or maybe they're looking up their own medical information.
• 5:21
  SPEAKER_02: I don't think they're integrating it into the practice of law yet.
• 5:24
  SPEAKER_03: Do you think there's uh there's a need to educate lawyers, like to actually proactively do that, or let them figure it out for themselves?
• 5:34
  SPEAKER_02: I think there's definitely a need to proactively tell them because there's so much hype and there's so much fear, and there's such a huge fragmentation and split between big law and solos and smalls.
• 5:45
  SPEAKER_02: And so, you know, you see all these tools.
• 5:48
  SPEAKER_02: And I'll be honest, before I started working in AI, I thought, what is this AI?
• 5:53
  SPEAKER_02: You know?
• 5:54
  SPEAKER_02: And then when I got to see a little bit more about it and understand the back end more, you know, it's it's something that for solos and smalls and the unique workflows that we have, the use of AI is going to be much different than the big law firms and the big tools and the Harveys and the Lagoras.
• 6:10
  SPEAKER_02: And, you know, if you go online, that's what you're reading about.
• 6:13
  SPEAKER_02: I see you're pretty active on LinkedIn.
• 6:15
  SPEAKER_02: What have you gotten from from that activity?
• 6:18
  SPEAKER_02: I really have I worked in big some big law firms on the data analytics side.
• 6:22
  SPEAKER_02: And so I networked with some um big law firms, but not as an associate attorney.
• 6:28
  SPEAKER_02: So I'd made connections through LinkedIn that way.
• 6:31
  SPEAKER_02: But then I started posting, you know, how I'm using uh AI in my practice and how solos are using it.
• 6:38
  SPEAKER_02: And I'm I connect with a few, but more importantly, I'm connecting with the ones, the attorneys that are online that are interested in becoming a legal AI trainer.
• 6:47
  SPEAKER_02: And so that's the growing momentum of who's starting to follow me now.
• 6:51
  SPEAKER_02: So, do you do legal AI training yourself?
• 6:53
  SPEAKER_02: I work.
• 6:54
  SPEAKER_02: So the way it works, and it's a multi-billion dollar industry, which is crazy.
• 6:59
  SPEAKER_02: It's it's just blows my mind what is erupted, is the big research labs like the Open AIs and the Anthropics and the Microsofts and the Googles and the Metas, they will hire companies who hire people to become AI trainers.
• 7:15
  SPEAKER_02: So that middle layer is Mercore and Micro One and Handshake and I'm saying the name wrong, Pareto.
• 7:23
  SPEAKER_02: That's the middle layer.
• 7:24
  SPEAKER_02: So they will hire experts to work in the AI training roles as a contractor.
• 7:30
  SPEAKER_02: That's where I am.
• 7:31
  SPEAKER_03: You know, I'm gonna move a little bit to something that's in really interesting to me relative to your background, which is the gerontology piece of it.
• 7:41
  SPEAKER_03: And I I've doom scroll through social media like the best of us.
• 7:48
  SPEAKER_03: And I'm seeing a lot of hostility towards AI, primarily from young people, for all sorts of different reasons.
• 7:57
  SPEAKER_03: Are you, in your experience with older people, are they comfortable with AI?
• 8:02
  SPEAKER_03: Do they not even know that it's out there?
• 8:05
  SPEAKER_03: Are they as adverse to AI, you think, as younger people?
• 8:09
  SPEAKER_02: How do you think that plays out?
• 8:10
  SPEAKER_02: I always use my mom as an example, you know, that she's in her 80s and she'd had no clue.
• 8:16
  SPEAKER_02: And I think that because they had no clue what it is, and they've gone through such an amazing transformation of technology throughout their lives, if you show them, and again, I think medical is a very interesting area for older adults to sort of move into using ChatGPT, because we all have medical issues as we get older.
• 8:34
  SPEAKER_02: And so if you don't understand something, let's say, you ask ChatGPT, and I will I will teach my mom how to iterate and how to push back and how to learn more.
• 8:44
  SPEAKER_02: And then I always, always, always, always, always say, you have to consult with the doctor.
• 8:49
  SPEAKER_02: You know, so I think that the older adults that I've run into, either they completely don't use AI at all, or all of a sudden my mom's saying, Well, can ChatGPT look at my computer?
• 9:00
  SPEAKER_02: Can it find my phone?
• 9:01
  SPEAKER_02: And I'm like, no, no, no, my, yeah.
• 9:04
  SPEAKER_02: But she loves it.
• 9:05
  SPEAKER_02: She's like, I use it three times a day, and I tell everybody else to do it.
• 9:08
  SPEAKER_02: So I think they're very receptive because I think they're getting answers that they normally wouldn't have had otherwise.
• 9:15
  SPEAKER_03: Are you seeing different levels of adoption based upon the age of the user?
• 9:22
  SPEAKER_03: What are you seeing?
• 9:23
  SPEAKER_01: Yes, yes, very much.
• 9:25
  SPEAKER_01: And as she was mentioning with the older generation, I would say more hesitant and a little more frightened, but they are willing and able and become just as competent once they really just, you know, it's like everything else.
• 9:42
  SPEAKER_01: It's practicality, it's starting to do it, getting in there and doing it.
• 9:45
  SPEAKER_01: It's like when you go to college to get your degree, but then you really learn how to do that job in the office when you're on the job.
• 9:51
  SPEAKER_02: And then I'm sorry.
• 9:52
  SPEAKER_02: Um go on, Sarah.
• 9:53
  SPEAKER_02: No, no, no, go on, Sarah.
• 9:54
  SPEAKER_02: They become the biggest advocates then to each other, like to their their same age group.
• 9:59
  SPEAKER_01: And that's nice.
• 10:00
  SPEAKER_02: So I I have my mom's whole church calling me.
• 10:05
  SPEAKER_02: That's great.
• 10:06
  SPEAKER_02: And so there again, I did my dissertation and my theory was the technology acceptance model.
• 10:12
  SPEAKER_02: And often theory is proving common sense.
• 10:15
  SPEAKER_02: Don't tell any other PhDs thing.
• 10:16
  SPEAKER_02: But that's my input.
• 10:18
  SPEAKER_02: So basically, how do we adopt technology?
• 10:20
  SPEAKER_02: Is it easy to use and is it useful?
• 10:23
  SPEAKER_02: And I think ChatGPT or Claude or, you know, I wouldn't say grok because that's a little different, but Gemini, they're easy to use.
• 10:30
  SPEAKER_02: And are they useful?
• 10:32
  SPEAKER_02: Once they find out that they're useful, that adoption rate just soars.
• 10:36
  SPEAKER_02: What's different about Grok?
• 10:38
  SPEAKER_02: Well, just because I've heard it's a little bit more political, it's a little rougher around the edges.
• 10:43
  SPEAKER_02: I haven't used it as much just because I mean I'm on Twitter as well, but I know it's a lot of people are.
• 10:49
  SPEAKER_02: I personally just have not used it because of those reasons.
• 10:52
  SPEAKER_03: I'll tell you, if there's an AI model, I want to try it out.
• 10:55
  SPEAKER_03: And I've used Grok extensively, and I'm going to tell a quick little story.
• 11:00
  SPEAKER_03: My wife is a novelist, and she's wrapping up an amazing three-volume epic novel about the 17th century BC in Gnosis and Egypt and Santorini.
• 11:14
  SPEAKER_03: And we're talking about how to get it published.
• 11:17
  SPEAKER_03: And I suggested, I said, you know what would be really cool is if you used AI tools to create imagery and videos that you can put online and maybe give potential readers a taste of what you're doing, and maybe that could drum up some interest.
• 11:34
  SPEAKER_03: Her main characters are a 16-year-old girl and a 12-year-old sister.
• 11:40
  SPEAKER_03: And when she went into Chat GPT to try to get imagery, they shut her down.
• 11:47
  SPEAKER_03: No, they said they said, nope, if you use the word 16 next to the word girl, we're not doing it.
• 11:54
  SPEAKER_03: So she got really frustrated.
• 11:56
  SPEAKER_03: I said, try Grok.
• 11:58
  SPEAKER_03: And you know what?
• 11:58
  SPEAKER_03: Grok rendered those images beautifully and moved everything over to Da Vinci.
• 12:06
  SPEAKER_03: And now she's got some great clips that she's going to put online and open up a Patreon for that.
• 12:11
  SPEAKER_03: Yeah.
• 12:12
  SPEAKER_03: And so and she and she's 68.
• 12:14
  SPEAKER_01: They haven't been sued as much yet as uh OpenAI, so they're still allowing it.
• 12:18
  SPEAKER_01: But why would she not?
• 12:20
  SPEAKER_01: I would think Midjourney would be the most ideal place to create those graphics and and videos, like far superior to any of these LLMs.
• 12:28
  SPEAKER_03: Agreed.
• 12:29
  SPEAKER_03: We were working through Midjourney.
• 12:31
  SPEAKER_03: Shut her down.
• 12:32
  SPEAKER_03: Wow.
• 12:32
  SPEAKER_03: Okay.
• 12:33
  SPEAKER_03: I went to I went to Gemini to ask for tips on how to help her work around this pretty heavy-handed, though understandable, AI reluctance to generate these images.
• 12:49
  SPEAKER_03: And Gemini gave me a whole bunch of interesting tips.
• 12:52
  SPEAKER_03: I mean, we've gotten really far afield, but I think I think this conversation is useful in understanding that these tools, contrary to something I said in our second episode, these tools are really starting to develop their own personalities and their own uses.
• 13:09
  SPEAKER_03: Heather is smiling.
• 13:10
  SPEAKER_03: Why are you smiling, Heather?
• 13:11
  SPEAKER_01: Because you know I've thought that all along.
• 13:14
  SPEAKER_01: We've talked about it ad nauseum.
• 13:16
  SPEAKER_01: But yeah, I mean, I I named mine a long time ago because it already had a personality.
• 13:20
  SPEAKER_01: We were interacting so much, and told you I had friends that had named theirs as well for similar reasons.
• 13:27
  SPEAKER_01: Right.
• 13:27
  SPEAKER_03: Well, I just think it is interesting.
• 13:30
  SPEAKER_03: And maybe six months ago, there was less differentiation between the frontier models.
• 13:37
  SPEAKER_03: But now everyone's so wait, Heather, you just talked about how ChatGPT is coming out with all these tools trying to imitate Claude.
• 13:45
  SPEAKER_03: Yes.
• 13:46
  SPEAKER_03: Sarah, have you seen that?
• 13:48
  SPEAKER_01: Exactly.
• 13:48
  SPEAKER_01: So please share.
• 13:49
  SPEAKER_01: So it just came out with uh codex, ChatGPT work, and then three specific models similar to, you know, that are that are varying.
• 13:59
  SPEAKER_01: It's always had different models, but but it's always had like its old model and then its older model, but now it has three new models all at once.
• 14:07
  SPEAKER_01: And so they do different things similar to what Claude has.
• 14:10
  SPEAKER_01: So it is, I read an article this morning and it was, you know, Chat GPT is shamelessly copying Claude.
• 14:16
  SPEAKER_01: And it it was something, and Ron made a very good point.
• 14:18
  SPEAKER_01: They they didn't just come out with this just because Claude did.
• 14:21
  SPEAKER_01: They've been working on these tools for a long time.
• 14:23
  SPEAKER_01: And I said, I'm sure everybody in this industry knows, just like in the legal industry, we all know what's going on behind the scenes before it's happening.
• 14:29
  SPEAKER_01: So but I have a very interesting uh like thing to pose to you, Sarah, because your background is fascinating and extensive.
• 14:37
  SPEAKER_01: But with a a PhD in technology from Purdue, does that make you look at AI and the law?
• 14:43
  SPEAKER_01: I would guess it makes you evaluate it far differently.
• 14:45
  SPEAKER_01: And if you were in, if you were just fresh out of school and young Dr.
• 14:50
  SPEAKER_01: Sarah Kubrick had just gotten that PhD and they had shown you guys LLMs and AI, would it have shocked you back then?
• 14:57
  SPEAKER_01: Would you think, oh, this is exactly where I thought, you know, we were going, or would you have been like, wow, I would have never dreamed this is what was gonna happen down the roof.
• 15:05
  SPEAKER_01: I never would have dreamed.
• 15:06
  SPEAKER_02: I got my PhD decades ago.
• 15:08
  SPEAKER_02: So I'm old.
• 15:09
  SPEAKER_02: So of course what's going on is just mind-numbing.
• 15:12
  SPEAKER_02: And I really focused a lot on statistics.
• 15:15
  SPEAKER_02: So I was looking at structured data and unstructured data.
• 15:19
  SPEAKER_02: I was looking at um, you know, my research was qualitative and quantitative.
• 15:23
  SPEAKER_02: So it was really data focused, but then I had the design background of usability and user interface.
• 15:29
  SPEAKER_02: And so seeing now just how, you know, the prompt, the simple prompt window, it's glorious.
• 15:36
  SPEAKER_02: And yet all the underpinning, I underpin underpinnings underneath, it's mind-boggling.
• 15:41
  SPEAKER_02: Are there limitations?
• 15:42
  SPEAKER_02: Sure.
• 15:42
  SPEAKER_02: And I think I have always been somewhat skeptical of tech uh because I think it can be oversold.
• 15:48
  SPEAKER_02: And again, if it's not useful, going back to that Tam serie, then I'm not really in love with it.
• 15:54
  SPEAKER_02: And so I I would have looked at AI probably with the same view of, well, is it useful?
• 16:00
  SPEAKER_02: It's easy to use for the most part, but am I sold on it?
• 16:04
  SPEAKER_02: I don't know.
• 16:05
  SPEAKER_02: But I'm just amazed at how quickly it's it's changed.
• 16:08
  SPEAKER_02: That's the thing, too.
• 16:09
  SPEAKER_02: It's just changing so much.
• 16:11
  SPEAKER_03: So at what at what point did you realize that you can actually start building things yourself with AI tools?
• 16:18
  SPEAKER_03: Yeah.
• 16:19
  SPEAKER_02: I think I loved the term vibe coding when it came out.
• 16:23
  SPEAKER_02: I had I had worked in a law firm and I saw someone had vibe coded some um Power BI DAX, some code.
• 16:29
  SPEAKER_02: And I thought, what are you doing?
• 16:30
  SPEAKER_02: Because she didn't understand how to even build what she was building by herself.
• 16:34
  SPEAKER_02: She's like, I don't understand it, but it works.
• 16:36
  SPEAKER_02: And I'm like, well, that's scary.
• 16:37
  SPEAKER_02: So I still believe in learning the fundamentals and uh now we're talking about like sort of coding and building things, so that you could identify when something breaks.
• 16:46
  SPEAKER_02: Again, moving from beta to production.
• 16:48
  SPEAKER_02: If it breaks and you're like, oh, I don't know.
• 16:50
  SPEAKER_02: I mean, how useful is that?
• 16:52
  SPEAKER_02: So vibe coding is great, but it doesn't have to be vibe coding something that necessarily uses AI.
• 16:58
  SPEAKER_02: On my website, I threw up a uh Medicare, Medicaid calculation calculator, you know, Medicaid penalty calculator.
• 17:05
  SPEAKER_02: And it was HTML and JavaScript.
• 17:07
  SPEAKER_02: Did I build it from scratch?
• 17:09
  SPEAKER_02: Heck no.
• 17:09
  SPEAKER_02: I asked one of the models, you know, hey, this is what I want.
• 17:12
  SPEAKER_02: And I had to go back and forth because some of the things were wrong, and I knew how to put it into my WordPress site, but that's because I'm understanding sort of the structure behind it.
• 17:22
  SPEAKER_02: I think that's where you're more important than just blindly accepting what the AI is giving you.
• 17:26
  SPEAKER_03: I noticed on your website you've got the Medicaid penalty calculator, you've got a community spouse calculator, you've got a child support chatbot, you've got legal dashboards, you're teasing that you have an AI agent currently in development.
• 17:43
  SPEAKER_03: Did you use AI to help you develop all this?
• 17:46
  SPEAKER_03: No, no.
• 17:47
  SPEAKER_02: I mean, you know, I've been building dashboards for a while.
• 17:51
  SPEAKER_02: And like I said, I was a big Power AI user.
• 17:53
  SPEAKER_02: So I I know how to pull data and I I know SQL and data analytics.
• 17:59
  SPEAKER_02: That was another sort of pivot.
• 18:00
  SPEAKER_02: But no, I didn't use AI.
• 18:02
  SPEAKER_02: I where the newer things that have come in with these calculators, yeah, I used AI.
• 18:06
  SPEAKER_02: I saw a need when I was talking to clients, you know, especially with Medicaid.
• 18:11
  SPEAKER_02: And that's where I started in estate planning, which is a little bit more common.
• 18:16
  SPEAKER_02: More people are in it.
• 18:17
  SPEAKER_02: And then I wanted to move into elder law.
• 18:19
  SPEAKER_02: And that is much more complicated.
• 18:21
  SPEAKER_02: And there's figures and calculations.
• 18:23
  SPEAKER_02: So I knew for myself, when I'm talking to clients, I want a calculator.
• 18:28
  SPEAKER_02: I want to know how long the penalty is.
• 18:30
  SPEAKER_02: I want to know what happens if there's a spouse.
• 18:32
  SPEAKER_02: And that's when I vibe coded it.
• 18:34
  SPEAKER_02: And it could reference, AI can reference the laws.
• 18:37
  SPEAKER_02: It could pull in, you know, the HTML.
• 18:40
  SPEAKER_02: And it was just, I think, really helpful.
• 18:42
  SPEAKER_03: I'm seeing now you you see this on LinkedIn.
• 18:45
  SPEAKER_03: Obviously, Heather, you're seeing all this kind of stuff.
• 18:48
  SPEAKER_03: People advertising these clawed skill workshops.
• 18:52
  SPEAKER_03: It's, you know, learn ChatGPT codecs in four hours, et cetera, et cetera.
• 18:58
  SPEAKER_03: Are lawyers really gonna do that?
• 19:00
  SPEAKER_03: Are they gonna really sit down in front of their computers and vibe code?
• 19:05
  SPEAKER_03: Or are they really, are they waiting for Godot?
• 19:08
  SPEAKER_03: Are they waiting for these tools to come and be made more useful for them so that they could just incorporate them and do lawyering like they always have?
• 19:19
  SPEAKER_02: I think as any good academic, we're gonna answer both.
• 19:21
  SPEAKER_02: What I mean by that is when I talk to solos, I just talked to some at the um Indiana State Bar Association, one of their conferences, and I said, start with your pain point.
• 19:31
  SPEAKER_02: What do you want to do that's right now painful for you?
• 19:35
  SPEAKER_02: And I think when you can identify in your firm where is your pain point, then you go to the tools that are correct.
• 19:42
  SPEAKER_02: Or you go to how much can I handle?
• 19:44
  SPEAKER_02: You know, how much do I know?
• 19:46
  SPEAKER_02: So I don't think it's necessarily that we're waiting for the tool because some people are a little bit more tech savvy.
• 19:52
  SPEAKER_02: Um, but I don't think it's then jump to the tool to teach first.
• 19:56
  SPEAKER_02: It's start with the pain point and then ask that to ChatGPT and say, what do I need to know?
• 20:02
  SPEAKER_02: What else is out there?
• 20:03
  SPEAKER_03: I mean, either way, the the lawyer is going to be sitting in front of a computer wrestling with technology.
• 20:10
  SPEAKER_03: And I th I think there's always been lawyers that are very comfortable with that.
• 20:15
  SPEAKER_03: And there's always been lawyers, and there always will be lawyers that just aren't.
• 20:20
  SPEAKER_03: Just they just aren't, you know.
• 20:22
  SPEAKER_03: So that leads me to my next question.
• 20:25
  SPEAKER_03: What's in your stack?
• 20:26
  SPEAKER_03: What do I use?
• 20:27
  SPEAKER_03: That should that should have been a 1970s commercial.
• 20:30
  SPEAKER_03: What's in your stack?
• 20:32
  SPEAKER_02: What's your AI stack look like these days?
• 20:34
  SPEAKER_02: Well it depends on it depends on what uh model I'm testing with my AI training jobs, that's for sure.
• 20:40
  SPEAKER_02: But I love Claude for just the default writing skills.
• 20:44
  SPEAKER_02: Um, I will usually take something and when I get a response, I will throw it in from Claude to GPT.
• 20:50
  SPEAKER_02: Um I will go into Gemini as well.
• 20:52
  SPEAKER_02: And these are just all the default settings.
• 20:55
  SPEAKER_02: And so, like I said, Claude, Gemini, Chat GPT, I will tell you with data, you know, it's Excel.
• 21:03
  SPEAKER_02: If I were working for a company, I know SQL, I know Power BI to pull in bigger data sets, but it's pretty basic and it's like Word and PowerPoint.
• 21:13
  SPEAKER_02: So there's so much that's cloud-based right now that, you know, again, it's it's what do I want to do?
• 21:19
  SPEAKER_02: And can I find it online?
• 21:21
  SPEAKER_03: So you what do lawyers you think misunderstand about AI?
• 21:26
  SPEAKER_02: I think lawyers think that AI models are like Google searches, that there is this immediate query of the entire web.
• 21:34
  SPEAKER_02: And, you know, when I talk to lawyers, I say there is an end date, and you know the law is always changing.
• 21:40
  SPEAKER_02: So you may not know when that end end date is.
• 21:43
  SPEAKER_02: I think, you know, everybody talks about hallucinations and they talk about, well, I don't want to get into AI because of the hallucinations.
• 21:50
  SPEAKER_02: And I think that that's the wrong approach because you certainly need to be mindful of hallucinations, and judges are sanctioning attorneys and pro se litigants for including.
• 22:00
  SPEAKER_02: Them and their filings.
• 22:01
  SPEAKER_02: But what did they get wrong about AI?
• 22:04
  SPEAKER_02: That it doesn't help you learn to be a better lawyer.
• 22:07
  SPEAKER_02: So again, I did estate planning and I, for the life of me, tried to find a mentor to help me with elder law.
• 22:13
  SPEAKER_02: And I was sending snail mail letters to people locally, and there was just no one returning.
• 22:18
  SPEAKER_02: So I asked ChatGPT when I have questions.
• 22:21
  SPEAKER_02: I'll ask Gemini, I'll ask Claude.
• 22:23
  SPEAKER_02: It may be wrong, but it's going to point me down something where I'm going to go down a rabbit trail, rabbit hole, find out what the codes are, what the case law is.
• 22:32
  SPEAKER_02: Yeah.
• 22:32
  SPEAKER_02: And of course I have what is it, decisis through my uh bar association, my state bar association.
• 22:38
  SPEAKER_02: So I'll do legal research that way.
• 22:40
  SPEAKER_02: So I think that's where lawyers are getting it wrong.
• 22:42
  SPEAKER_02: They think it's like something that's trained every single or something that's queried every single time on the entire internet.
• 22:48
  SPEAKER_02: They think that it knows everything, which it doesn't.
• 22:51
  SPEAKER_02: Um, and they think that it can't help because it gives wrong answers or partially wrong answers.
• 22:56
  SPEAKER_03: Yeah.
• 22:56
  SPEAKER_03: My wife and I discovered that ChatGPT does not know everything when we asked it to give us an interpretation from the TV show Pluribus.
• 23:08
  SPEAKER_03: And it didn't really know anything about the TV show Pluribus, even though it was really hot and it had just wrapped at the time.
• 23:16
  SPEAKER_00: Yeah.
• 23:17
  SPEAKER_03: So I studied your LinkedIn page and I saw that you quoted from a couple of articles that talked about some surprising statistics about the practice of law.
• 23:29
  SPEAKER_03: One of them was a statistic that 40% of all law firms are solo practices, and about 75% have six lawyers or fewer.
• 23:39
  SPEAKER_03: How can smaller law firms best optimize their use of AI and their and their adoption of AI into their firms?
• 23:48
  SPEAKER_02: Well, I think again, what do they what do they want to use it for?
• 23:51
  SPEAKER_02: What do what's the problem?
• 23:53
  SPEAKER_02: I mean, right now, is it client acquisition?
• 23:55
  SPEAKER_02: Is it general legal knowledge?
• 23:57
  SPEAKER_02: Is it finding referrals?
• 23:59
  SPEAKER_02: So I think that first they they need to say, like, what am I getting problems with in my practice?
• 24:03
  SPEAKER_02: And then I say, go to ChatGPT and the free model and just start prompting, right?
• 24:08
  SPEAKER_02: Just start learning how to interact with the technology and not to be sort of afraid of it and and listening to the scare of, you know, it's going to hallucinate everything or um, you know, it's gonna solve solve cancer.
• 24:26
  SPEAKER_03: You're you're an estate planning attorney, which is a very document-intensive practice area.
• 24:32
  SPEAKER_03: Wow.
• 24:33
  SPEAKER_03: All right.
• 24:34
  SPEAKER_03: How have what have you done to incorporate your playbook into AI?
• 24:40
  SPEAKER_03: Are you doing that?
• 24:42
  SPEAKER_03: What do you mean by my playbook?
• 24:43
  SPEAKER_03: Well, like your your history of wills and trusts and advanced directives and powers of attorney and those kind of documents.
• 24:53
  SPEAKER_01: Yeah, are you using it to help automate your own firm and these are sure.
• 24:56
  SPEAKER_02: So we all have documents and way too many documents, and we do a file save as with a new client, right?
• 25:03
  SPEAKER_02: I have definitely uploaded my own, you know, stripped of personal identifying information, PII, I upload my document and I'll say, take this, um, let's say it's a trust or a will and modify it so that instead of one person, it's two people.
• 25:20
  SPEAKER_02: Maybe they have a special needs child, maybe they have minors that are going to be the beneficiaries.
• 25:25
  SPEAKER_02: So anytime I need to amend my documents, I'll upload them for sure.
• 25:30
  SPEAKER_03: Have you created like persistent memory skills in Claude where you, all right, I've got the Jones will enter a couple of items into the prompt and then it churns out all the documents you need.
• 25:44
  SPEAKER_03: Have you set it up in that way?
• 25:45
  SPEAKER_02: I should, right?
• 25:46
  SPEAKER_02: I should, because I want to learn more about skills.
• 25:48
  SPEAKER_02: And I think agentic AI is where the solos are going to have the most benefit.
• 25:53
  SPEAKER_02: And I think that especially running with um local models on their own machines, we have so many files that we can use to have our own models referenced, right?
• 26:07
  SPEAKER_02: Instead of feeding it up to the cloud.
• 26:09
  SPEAKER_02: Uh, so that's my and and help us with our workflows rather than just a prompt.
• 26:15
  SPEAKER_02: I think it starts with a prompt, but I don't think that's the most powerful.
• 26:18
  SPEAKER_01: You're talking about building the skills to use, correct?
• 26:21
  SPEAKER_01: Like in the AI.
• 26:22
  SPEAKER_01: Yeah, like what we talked about last week.
• 26:24
  SPEAKER_01: I have a quick question.
• 26:25
  SPEAKER_01: I mean, it's probably not quick, but with your background, because it is so fascinating because you have a doctorate in technology, you practice law, you train AI models, you've done data.
• 26:36
  SPEAKER_01: It looks like this is such a rare combination.
• 26:38
  SPEAKER_01: Does this make you have a vastly different perspective?
• 26:40
  SPEAKER_01: I would think it does on what attorneys should really be paying attention to with AI.
• 26:44
  SPEAKER_01: You've talked about the pain points and not being as afraid of it.
• 26:48
  SPEAKER_01: But bigger picture over the next, you know, few years, because AI is changing so rapidly, it's evolving so rapidly, no one, including the courts, can keep up with it.
• 26:57
  SPEAKER_01: What what does having someone having, I feel like you kind of have everything, you know, that those all of those different degrees and perspectives.
• 27:06
  SPEAKER_01: What would you say attorneys should be really focusing on in AI world?
• 27:11
  SPEAKER_01: You've said something many times that I think you agree with me.
• 27:14
  SPEAKER_01: I think that AI is giving small and solo attorneys a huge advantage that only big law has had for all these many years.
• 27:21
  SPEAKER_01: Yeah.
• 27:22
  SPEAKER_01: Yeah.
• 27:22
  SPEAKER_01: I well, I think so.
• 27:24
  SPEAKER_01: Thank you for complimenting me.
• 27:25
  SPEAKER_01: You guys are so sweet.
• 27:27
  SPEAKER_01: It's a rare combination of achievements.
• 27:30
  SPEAKER_03: All all of this expertise in in one person, you can't do it.
• 27:33
  SPEAKER_03: Tell us more.
• 27:36
  SPEAKER_03: That's well, and I'm not sure.
• 27:37
  SPEAKER_01: We're trying to get everything we can out of this, Sarah.
• 27:39
  SPEAKER_02: Right, right.
• 27:40
  SPEAKER_02: But again, thank you.
• 27:41
  SPEAKER_02: I I think one of the things is we are looking at such a push with AI in our society, in our world, you know, that it's not ending, right?
• 27:51
  SPEAKER_02: So it's it's happening.
• 27:52
  SPEAKER_02: And so as lawyers, and you mentioned this, Heather, it's not just lawyers, it's the judges, it's the courts, it's the legal aid societies, it's our whole world is going to be impacted by this.
• 28:04
  SPEAKER_03: Starting in the fall, it won't be for one L's at Chicago.
• 28:10
  SPEAKER_00: Oh my gosh.
• 28:11
  SPEAKER_03: University of Chicago.
• 28:13
  SPEAKER_03: What do you think about the fact that the University of Chicago Law School is banning electronics in first-year law classes?
• 28:21
  SPEAKER_03: I told him what do you think about that?
• 28:24
  SPEAKER_02: Well, and I'm like 40 minutes away from there.
• 28:26
  SPEAKER_02: So I I want to just like pass out meta glasses and be like, just use.
• 28:35
  SPEAKER_02: I think it, you know, I understand everybody says when you go to law school, you learn how to think differently.
• 28:41
  SPEAKER_02: But this is a tool that makes us better.
• 28:44
  SPEAKER_02: It really is.
• 28:45
  SPEAKER_02: And so crippling students, I said, get your pencils out.
• 28:48
  SPEAKER_02: That was my my post on LinkedIn.
• 28:50
  SPEAKER_02: Get your pencils out.
• 28:51
  SPEAKER_03: Yep.
• 28:52
  SPEAKER_02: Um, you know, there's always gonna be people that are are not learning the correct way, but I just think it's it's just the wrong move to say that they shouldn't have it in their one-l classes.
• 29:01
  SPEAKER_01: It's just I give it, I give it one semester.
• 29:04
  SPEAKER_01: One semester.
• 29:06
  SPEAKER_01: And even a full year.
• 29:07
  SPEAKER_01: Maybe a year, but I guarantee they're gonna have to, they're gonna walk that back.
• 29:10
  SPEAKER_01: I I just can't imagine the because lawyers, you need to learn to think with this now, right?
• 29:16
  SPEAKER_01: There are this, they shouldn't be like she even used the word crippling, Sarah.
• 29:20
  SPEAKER_01: That's a perfect little description because you're basically setting them back a year on on learning to use AI properly, responsibly, ethically, blah, blah, blah.
• 29:30
  SPEAKER_01: Right.
• 29:30
  SPEAKER_01: Right.
• 29:30
  SPEAKER_01: Yes, I disagree.
• 29:32
  SPEAKER_03: To me, well, you know, my ongoing controversial hot take is that lawyers should record everything because it's all data.
• 29:40
  SPEAKER_03: Data has become the infrastructure of the modern law firm.
• 29:44
  SPEAKER_03: And and, you know, if you don't want to record everything, don't collect that data.
• 29:48
  SPEAKER_03: Right.
• 29:49
  SPEAKER_03: You know, see see if that makes it for a better engagement and and better representation.
• 29:54
  SPEAKER_03: But what's wrong with recording lectures and having them transcribed?
• 30:00
  SPEAKER_03: To me, that seems really counterproductive.
• 30:04
  SPEAKER_02: It does.
• 30:05
  SPEAKER_02: It's it's strange.
• 30:06
  SPEAKER_02: I mean, uh I said I went to Cooley because I wanted to learn from practitioners.
• 30:09
  SPEAKER_02: I don't know.
• 30:10
  SPEAKER_02: May maybe these, maybe this I know they they they consulted with some big law firms, but I just don't understand the decision, honestly.
• 30:18
  SPEAKER_03: Well, let's look at the opposite side.
• 30:21
  SPEAKER_03: Do young lawyers risk becoming dependent on AI too early?
• 30:25
  SPEAKER_02: I think that law students may, but I don't think young lawyers will.
• 30:29
  SPEAKER_02: Because again, I think you learn a certain set of skills in school, everything's theoretical, and then you get out and you do the job, and you're like, whoa, now I'm learning by eating gravel, right?
• 30:39
  SPEAKER_02: I'm actually doing it.
• 30:41
  SPEAKER_02: So um, I don't think they're they're gonna be at a loss once they start working in practice.
• 30:46
  SPEAKER_03: All right.
• 30:46
  SPEAKER_03: So one last question before we move on to our practice segment.
• 30:51
  SPEAKER_03: What's one AI capability every solo lawyer should learn?
• 30:56
  SPEAKER_03: What was that?
• 30:57
  SPEAKER_03: How to prompt.
• 30:58
  SPEAKER_03: How to prompt.
• 30:58
  SPEAKER_03: How to prompt.
• 30:59
  SPEAKER_03: Just just how to prompt.
• 31:01
  SPEAKER_03: We did we did get an episode on prompting.
• 31:04
  SPEAKER_03: Awesome.
• 31:05
  SPEAKER_02: They should reference because it's not it's not Google search, right?
• 31:09
  SPEAKER_02: And a lot of people think it's like the one and done.
• 31:11
  SPEAKER_02: And when I gave my presentation, now I taught Photoshop for 15 years.
• 31:15
  SPEAKER_02: So I could take some of the images generated that have six fingers and I could correct them, you know, by AI.
• 31:21
  SPEAKER_02: I pull them into Photoshop.
• 31:23
  SPEAKER_02: But I did this image of two people, cartoons, standing next to one of those big saws that the lumberjacks use, and they're sawing that log.
• 31:32
  SPEAKER_02: And the whole point is to iterate, it's to go back and forth and back and forth.
• 31:36
  SPEAKER_02: And that's the beauty.
• 31:37
  SPEAKER_02: So the prompt isn't just, you know, one sentence or a one and done.
• 31:41
  SPEAKER_02: When I say how to prompt, I'm talking going back and forth and going together.
• 31:45
  SPEAKER_02: How to interact, how to interact.
• 31:48
  SPEAKER_01: I've been trying to preach this for a while now.
• 31:51
  SPEAKER_03: All right.
• 31:51
  SPEAKER_03: So you know what?
• 31:53
  SPEAKER_03: I stay away from politics on this podcast, but you said that you were a Photoshop trainer for 15 years.
• 32:02
  SPEAKER_03: I have to ask this question.
• 32:05
  SPEAKER_03: What do you think about the Mitch McConnell proof of life photo that was just released?
• 32:10
  SPEAKER_03: What do you what you know what I'm talking about?
• 32:13
  SPEAKER_03: The one where he's he and his wife are in a hospital room and he's got a newspaper on his lap.
• 32:17
  SPEAKER_02: Yeah, I couldn't tell you.
• 32:19
  SPEAKER_02: You know, I I have to admit, I haven't seen it because I'm in a sprint right now with one of my projects where they're wanting us to work a phenomenal amount of hours to get the project completed.
• 32:28
  SPEAKER_02: So I have not seen it, however.
• 32:30
  SPEAKER_02: One of the articles I talked about, wrote when I was an academic was how to spot manipulated photographs as trial exhibits in a court of law.
• 32:40
  SPEAKER_02: And so it was way back just talking about Photoshop.
• 32:43
  SPEAKER_02: I don't know nowadays.
• 32:44
  SPEAKER_02: I think the deep fakes in the photo generation is just phenomenally more challenging.
• 32:49
  SPEAKER_02: So I haven't seen it.
• 32:49
  SPEAKER_02: I'll have to look that up.
• 32:51
  SPEAKER_03: All right, you know what?
• 32:51
  SPEAKER_03: I'll link to it in the show notes.
• 32:53
  SPEAKER_03: Okay, good.
• 32:54
  SPEAKER_03: All right.
• 32:54
  SPEAKER_03: So now we're gonna move into our uh regular segment of practice signals, which is information we get from lawyers in the wild having posing problems that baby AI can help them with.
• 33:09
  SPEAKER_03: So they almost all come from Reddit these days because uh, you know, Redditors are out of their minds slightly, but they're not afraid of showing their lunacy.
• 33:23
  SPEAKER_03: Yes.
• 33:24
  SPEAKER_03: This one was actually a sad story.
• 33:27
  SPEAKER_03: There was a post from a law student who had accepted a summer associate position, and then life happened.
• 33:34
  SPEAKER_03: He had unexpected medical bills, his housing situation fell apart, he maxed out his credit cards, and he's got no family near where he's working for a big law firm.
• 33:48
  SPEAKER_03: So the title of the post was simply homeless as a summer associate, should I tell my firm?
• 33:54
  SPEAKER_03: Now, imagine dropping that into Chat GPT or Claude.
• 33:58
  SPEAKER_03: My question.
• 33:59
  SPEAKER_03: So my question is, how would you how would you advise this this person?
• 34:05
  SPEAKER_03: Uh and and maybe is there an AI solution for that?
• 34:08
  SPEAKER_03: See, now you're talking big law again, which is not my wheelhouse.
• 34:11
  SPEAKER_03: Well, we're talking about human beings.
• 34:13
  SPEAKER_03: We're talking about like a homeless person who's who's working for a big law firm as a summer associate.
• 34:20
  SPEAKER_02: So sort of the human side of things.
• 34:22
  SPEAKER_03: Yeah, I mean, can he you know, is there an AI solution to help out this poster?
• 34:28
  SPEAKER_02: I think one of the things that I I tell people is you always ultimately, if you have an issue, you want to connect with a human expert or a human in the end.
• 34:35
  SPEAKER_02: So maybe this homeless situation, maybe you could ask the model, where can I go?
• 34:42
  SPEAKER_02: Where's a local resource?
• 34:43
  SPEAKER_02: Full well knowing it's not trained on completely new data, but but pretty pretty well established.
• 34:48
  SPEAKER_02: You know, what's a contact?
• 34:50
  SPEAKER_02: What's what's a phone number?
• 34:52
  SPEAKER_02: And then, you know, connect offline, connect in the real world.
• 34:55
  SPEAKER_02: I think that that's always very important because again, realizing this is just, you know, in your computer.
• 35:00
  SPEAKER_02: This is this is not um in your head and it yet.
• 35:03
  SPEAKER_02: And it's it's not connecting to a lot of real world resources because it doesn't know it's out there or it can't access that information.
• 35:12
  SPEAKER_03: What do you think, Heather?
• 35:13
  SPEAKER_03: What do you think about our our homeless summer associate?
• 35:16
  SPEAKER_01: Yeah, really sad story, number one.
• 35:18
  SPEAKER_01: But I understand that associates sometimes don't make enough to live.
• 35:22
  SPEAKER_01: I've, you know, like most people in the legal world, found that out a long time ago and was shocked, like I'm sure everyone is.
• 35:28
  SPEAKER_01: But I would I would ask the AI model, but I would also use Perplexity for this one, not just uh Chat GPT, because Perplexity is up to date and it is sort of like a super Google, if you will.
• 35:38
  SPEAKER_01: But use the GPT, whatever, you know, LLMs you have at your disposal, and you can open the free Chat GPT.
• 35:44
  SPEAKER_01: You can use the free Perplexity, you know, for small amounts of usage.
• 35:47
  SPEAKER_01: Find out what resources in his county or area there are for housing, and then uh it absolutely get connected with a human person who can then help you maybe find an option if it's if it's there for them.
• 36:01
  SPEAKER_03: You know, I never really thought that this was an AI question.
• 36:06
  SPEAKER_03: This to me is very much a human question.
• 36:10
  SPEAKER_03: Right.
• 36:10
  SPEAKER_03: You know, go to if you're at a summer associate, somebody hired you.
• 36:14
  SPEAKER_03: Somebody recruited you, somebody hired you.
• 36:16
  SPEAKER_03: He he wants to know, should I tell my firm?
• 36:18
  SPEAKER_03: Not only should he tell his firm, he should go hat in hand to his supervising partner and say, you know, I love this firm.
• 36:28
  SPEAKER_03: Everything is really great for me, but these are the problems I'm having.
• 36:32
  SPEAKER_03: Does the firm have any resources to help me?
• 36:35
  SPEAKER_03: Right.
• 36:36
  SPEAKER_03: You know, if that if that firm has any decency at all, and I'm not saying that every big law firm is chock full of decency, but humans typically have some degree of compassion for others in their orbit who are struggling.
• 36:52
  SPEAKER_03: You know, they would put out an email or they slack whatever the firm uses to communicate, or they'd have behind the scenes conversations.
• 37:00
  SPEAKER_03: You know, is there another submerassociate that could use a roommate?
• 37:04
  SPEAKER_03: Yeah.
• 37:05
  SPEAKER_03: Are there other partners or associates that have a spare bedroom that they could have this uh law student there for the summer?
• 37:13
  SPEAKER_03: I think there's a solution to this problem.
• 37:16
  SPEAKER_03: But I did ask ChatGPT if there was an AI solution, and ChatGPT said, AI is not only write me an email, it could be something that completely different use of AI.
• 37:30
  SPEAKER_03: AI might suggest you ask for a payroll advance so that you could afford to buy an Airbnb for the summer.
• 37:38
  SPEAKER_03: Ask whether the firm has emergency assistance, talk with your law school's career service office, contact the student affairs dean, look into Bar Association emergency funds, see if another summer associate needs a roommate, explain the situation before performance suffers, and avoid payday loans.
• 37:58
  SPEAKER_03: And while none of those ideas are particularly brilliant, AI is a resource for all of us in all sorts of different ways now.
• 38:07
  SPEAKER_03: Right.
• 38:07
  SPEAKER_03: And I do think that that could be a legitimate use of AI to help the summer associate.
• 38:12
  SPEAKER_03: Although, in my opinion, it's not the first and best solution.
• 38:16
  SPEAKER_01: I will also piggyback that that you know I have two attorneys, they're associates that work part-time for me.
• 38:22
  SPEAKER_01: So I do know that they sometimes need a second job.
• 38:24
  SPEAKER_01: And a place to live.
• 38:25
  SPEAKER_01: Yeah.
• 38:26
  SPEAKER_02: Well, they could be an AI trainer.
• 38:28
  SPEAKER_03: Right.
• 38:28
  SPEAKER_03: That's right.
• 38:29
  SPEAKER_03: Maybe an uh maybe, you know, I think this is the gig economy and everybody's got a side hustle.
• 38:36
  SPEAKER_03: And AI training is definitely a top on the list.
• 38:40
  SPEAKER_01: It's like the like the technology DoorDash gig.
• 38:42
  SPEAKER_02: Well, and it makes me a better lawyer, I think, because the more you teach, which is your training, the model, the more you understand your own subject matter.
• 38:50
  SPEAKER_02: So it's not just, you know, a tech perspective.
• 38:53
  SPEAKER_02: I'm writing about estate planning and Medicaid and um family law.
• 38:57
  SPEAKER_02: So it's making me a better learner.
• 38:59
  SPEAKER_02: Understanding it helps me understand my own um areas better.
• 39:02
  SPEAKER_02: So you think it makes you think better?
• 39:05
  SPEAKER_03: I think it helps Or does it make or does it make you lazier?
• 39:08
  SPEAKER_01: No, it makes you think better.
• 39:10
  SPEAKER_01: It gives you ideas you wouldn't have.
• 39:12
  SPEAKER_01: Right.
• 39:12
  SPEAKER_02: It makes it makes you a better thing, even it makes you a better editor too, because you're not starting from the file new.
• 39:19
  SPEAKER_02: Again, you're you're critiquing.
• 39:21
  SPEAKER_02: And I did want to say one thing, uh if you don't mind me adding adding I want you to say whatever you're going to say.
• 39:28
  SPEAKER_02: So one of the things that you you asked was, you know, what do attorneys think about AI?
• 39:34
  SPEAKER_02: I want them to understand the limitations.
• 39:36
  SPEAKER_02: Like we said, it's not scraping the internet all the time.
• 39:39
  SPEAKER_02: But also, there is so much knowledge we have in our brains and that we learn, and that's implicit, and that's not accessible to these models ever, ever, ever.
• 39:49
  SPEAKER_02: I mean, paralegals are dealing with humans and emotions, and it's messy, right?
• 39:55
  SPEAKER_02: And law is messy, and it's hard to get a binary outcome, which the models like, because we're dealing with the the people and the counselor, your counselor.
• 40:07
  SPEAKER_02: But there's so much that you know, again, when you're practicing, that is not posted somewhere online that a model can scrape.
• 40:14
  SPEAKER_02: So when people say to me, Oh, AI is going to replace our jobs as lawyers, I'm like, that is so far from the truth.
• 40:20
  SPEAKER_02: There's so much we know that it can never access.
• 40:23
  SPEAKER_02: And there's so many interactions we have, you know, with opposing counsel, with judges, with clients, um, that there's such a human side to our work, we're safe.
• 40:34
  SPEAKER_02: This is just a great tool.
• 40:35
  SPEAKER_02: So I wanted to add that.
• 40:36
  SPEAKER_03: Thank you for for those words.
• 40:38
  SPEAKER_03: I think they're very wise because it's really true that the more you interact with AI, the more you realize how essential the human involvement is.
• 40:50
  SPEAKER_03: Because AI is really very limited in what it does.
• 40:53
  SPEAKER_03: What it does is amazing, but it's still limited to a large extent.
• 40:58
  SPEAKER_03: It doesn't have that human judgment ability.
• 41:02
  SPEAKER_03: Right.
• 41:02
  SPEAKER_03: So I agree.
• 41:03
  SPEAKER_03: So um let's move into our Flintstones, Simpsons, and Jetsons.
• 41:09
  SPEAKER_03: Welcome back to our season two AI Build.
• 41:13
  SPEAKER_03: This is our third episode of the build process.
• 41:17
  SPEAKER_03: If you've been following along, you may be wondering all right, good for you, the Flintstone Simpsons, Jetsons season two.
• 41:25
  SPEAKER_03: Where's the AI in all this?
• 41:27
  SPEAKER_03: Taking us along spoonful by spoonful.
• 41:30
  SPEAKER_03: You know, when are you gonna get to the meat?
• 41:33
  SPEAKER_03: The first week, we identified our firm's knowledge assets.
• 41:36
  SPEAKER_03: Last week we organized them.
• 41:39
  SPEAKER_03: This week, we're gonna clean them.
• 41:41
  SPEAKER_03: What does that mean?
• 41:42
  SPEAKER_03: Before you teach AI your law firm, you have to decide what deserves to be taught.
• 41:46
  SPEAKER_03: If you hired the smartest associate in the world, you wouldn't hand them every document your firm has ever created.
• 41:51
  SPEAKER_03: You'd hand them your best work and you would advise them make sure that you don't leave over anything from a prior case in these firm in these forms.
• 42:03
  SPEAKER_03: Or if you're a Simpsons lawyer, you know, and you're collecting the client questions, you'd really want to study them and say, well, are we really using two questions to ask one thing?
• 42:14
  SPEAKER_03: Uh and if you're a Jetsons lawyer, you're gonna review your firm systems.
• 42:19
  SPEAKER_03: Don't document every exception or unusual situation.
• 42:22
  SPEAKER_03: You want to write down how the promise, you want to write down how the process normally works on a good day.
• 42:30
  SPEAKER_03: So AI learns much more effectively from clean standard workflow than from a list of every possible detour.
• 42:39
  SPEAKER_03: So, Sarah, how would you go about cleaning documents before you submitted them to AI?
• 42:46
  SPEAKER_03: I mean, you you've got a million forms, I'm sure.
• 42:50
  SPEAKER_02: Well, I'm working on vibe coding, putting on my WordPress site a PII anonymizer, document anonymizer.
• 42:57
  SPEAKER_02: And it told me how to write through HTML and JavaScript this web page.
• 43:02
  SPEAKER_02: And you get to go through and identify this is a name, I want it removed, you know, this is a company, this is this is a social security, I want it removed.
• 43:10
  SPEAKER_02: So again, the human is the one that's the human in the loop, right?
• 43:14
  SPEAKER_02: Before you get this sanitized document.
• 43:16
  SPEAKER_02: But I think one of the things I heard you talking about, and I'm translating it into my AI training job, we and you talked about skills.
• 43:23
  SPEAKER_02: As an AI trainer, not only are you doing prompts, you're also writing up what is called a greater guidance.
• 43:29
  SPEAKER_02: So you have a model that answers the prompt, but you may also have a model that grades that answer.
• 43:36
  SPEAKER_02: So you're writing up this is how the golden response, right?
• 43:41
  SPEAKER_02: The gold star standard response should be.
• 43:44
  SPEAKER_02: And these are the things that it needs to include.
• 43:46
  SPEAKER_02: And maybe some of these are failures that it should watch out for.
• 43:49
  SPEAKER_02: So um, I think that's sort of kind of how I translated what you were saying.
• 43:54
  SPEAKER_03: I I think that's very apt.
• 43:56
  SPEAKER_03: Heather, you're using a lot of skills.
• 44:00
  SPEAKER_03: You're doing a lot of vibe coding as you put together SOPs for your company.
• 44:06
  SPEAKER_03: Do you have an experience in making sure that the data you're putting in is clean?
• 44:10
  SPEAKER_01: Yeah, I think it's uh goes way back to the old, you know, retired phrase gago, because it is garbage in, garbage out.
• 44:19
  SPEAKER_03: Although about 10 episodes ago, we decided not to use that for it it really calls back to that, right?
• 44:26
  SPEAKER_01: Because of what you're putting in.
• 44:27
  SPEAKER_01: Yeah.
• 44:28
  SPEAKER_01: So so yeah, I I do, I think instead of uploading like a 300-page PDF to be reviewed anymore, you know, or instead of uploading a 300-page PDF as an attachment or as an exhibit anymore, you should now put it through first and and see if those things are all done that that we're talking about, and make sure there are no blank pages, nothing sideways, that the OCR was clean, that everything is readable, and that and any handwritten notes are legible and and able to be read.
• 44:58
  SPEAKER_03: I I'm I have this massive bankruptcy uh project that I'm working on, and I'm working through it with uh ChatGPT, and I'm getting frustrated because ChatGPT keeps prompting me to do things that don't feel like we're moving the project along, but we really are, because you know, the back end of whatever technology or system you're using has to be clean.
• 45:25
  SPEAKER_03: It has to support the architecture of what you're trying to.
• 45:28
  SPEAKER_03: All right.
• 45:29
  SPEAKER_03: Well, this has been a great episode.
• 45:31
  SPEAKER_03: Sarah, thank you so much for coming on board.
• 45:34
  SPEAKER_03: Any any parting comments?
• 45:35
  SPEAKER_02: I think some of the best people to follow too are people that are building.
• 45:39
  SPEAKER_02: I'm I'm on Twitter.
• 45:40
  SPEAKER_02: It's always Twitter, never X, but I'm on Twitter a lot.
• 45:43
  SPEAKER_02: You know, I'll follow people that have 20 followers because they're building.
• 45:47
  SPEAKER_02: I I always think it's interesting to see rather than just reading the theoretical and the high-level viewpoints, to see what works and what didn't work and and and then ask.
• 45:58
  SPEAKER_02: People are just crazy about sharing information.
• 46:02
  SPEAKER_02: And that's what I love, especially like with developers and tech people in tech.
• 46:06
  SPEAKER_02: There's so much out there that that they will share.
• 46:09
  SPEAKER_02: Just just do it.
• 46:11
  SPEAKER_03: All right, that's great.
• 46:12
  SPEAKER_03: Uh, thank you so much.
• 46:13
  SPEAKER_03: We will see you next week for the next episode of AI Tools for Practicing Lawyers.
• 46:18
  SPEAKER_03: Thank you.
• 46:19
  SPEAKER_00: Thank you, Sarah.
• 46:22
  SPEAKER_00: That's it for today's episode of AI Tools for Practicing Lawyers.
• 46:27
  SPEAKER_00: Thanks for listening.
• 46:28
  SPEAKER_00: We'll see you next time.