When a lawyer messes up by using an AI platform that produces mistakes, they might get sanctioned by a judge. When a judge messes up using an AI platform, “it could become precedent. So, it’s a much different conversation.” Judge Scott Schlegel, of Louisiana's Fifth Circuit Court of Appeals, speaks from his experience as an early leader in courtroom efficiency. Today, he supports a measured judicial approach to AI with this guiding principle: “It's not our job to be first. It's our job to get it right.” Tune in to this conversation with hosts Todd Smith and Jody Sanders to hear about his newer project: the Judicial AI Consortium (JAIC), which he is developing with U.S. Magistrate Judge Maritza Dominguez Braswell of Colorado and Judge Xavier Rodriguez of the Western District of Texas. The consortium is designed to be a simple forum where judges can “ask stupid questions, talk to each other about how you're using it, what you're seeing out there. Is it helpful? Is it useful? How far should we go?” About 200 judges around the country have signed up so far.

Connect and Learn More

☑️ Judge Scott Schlegel | LinkedIn

☑️ United States Court of Appeals for the Fifth Circuit on LinkedIn | Instagram | YouTube

☑️ Todd Smith | LinkedIn | X

☑️ Jody Sanders | LinkedIn | X

☑️ Texas Appellate Law Podcast on LinkedIn | X | Instagram

☑️ Texas Appellate Counsel PLLC

☑️ Kelly Hart & Hallman, LLP | LinkedIn

☑️ Subscribe Apple Podcasts | Spotify | Amazon Music | YouTube

Produced and Powered by LawPods

Sponsored by Court Surety Bond Agency and Proceed.

Welcome to the Texas

Appellate Law Podcast,

the show that takes you inside the

Texas and federal appellate systems.

Through conversations with judges, court

staff, top trial and appellate lawyers,

academics, and innovators,

we provide practical insights to help

you become a more effective advocate.

Whether you're handling

appeals or preparing for trial,

you'll discover strategies to sharpen

your arguments, innovate your practice,

and stay ahead of the latest developments.

And now, here are your hosts,

Todd Smith and Jodi Sanders.

Produced and powered by LawPods.

Welcome back to the Texas Appellate

Law Podcast. I'm Todd Smith.

And I'm Jody Sanders.

Our guest for this episode is Judge

Scott Schlegel from Louisiana's Fifth

Circuit Court of Appeal.

Welcome to the podcast, Judge.

Thanks for having me, Todd

and Jodi. Good to be here.

We were just joking offline.

This is worth a brief mention that the

name of your court is a little confusing

to those of us from Texas who practice

in the Fifth Circuit Court of Appeals,

also in your town,

also other courts in our state or courts

of appeals and not court of appeals,

but I wanted to make sure we had that

correct, the Circuit Court of Appeal.

How long have you been in

office on that court now, Judge?

Going on three years.

I was on the trial court for a little

Right. I do want to kind of give our

listeners a chance to get to know you.

So if you wouldn't mind just

kind of introducing yourself,

giving a little bit of your background

and give a brief synopsis of what got you

to your current position.

Yeah.

I was actually a financial advisor

during the late '90s when the

market went south for about 10 years.

So my wife looked at me and said,

"You should go to law school."

So I actually went to law

school at night while I

was a financial advisor by day. And

then once I graduated law school,

became a downtown lawyer doing products,

liability defense, civil products,

liability defense. I always joke, made

great money and wasn't for me. So I

became a prosecutor,

made no money and loved life and

enjoyed my time prosecuting cases.

It's a great job. It's a tough job.

We see some pretty terrible

things as a prosecutor.

And so I did that for about five

years and just lived and breathed the

courtroom. And the inefficiencies

literally drove me crazy.

I couldn't understand

it. So I quit my job,

took out a major medical cash out of 401k.

My son was in school,

my wife was in school, and I ran

for a judge and thankfully won.

And since day one, it has always been,

what can I do to make the justice

system more efficient, effective,

and accessible? And in my estimation,

that is using simple off-the-shelf

technologies to do things a little bit

better. I always say,

run your business better.

We're not talking about human judgment.

We're not talking about replacing judges.

We're just talking about the business

side of it and how I can make it a more

efficient practice. So when I was on the

trial court, the first thing I did was,

you know that big red book

that every minute clerk has?

And in order to get a hearing,

you have to call the minute clerk

and hope that they call you back.

Or you have to send a runner to get a

date and it drove me crazy. So I took the

big red book, tossed it in the

trash, put my calendar online,

allowed lawyers to pick their

own dates, call opposing counsel,

make sure it works for everybody.

I sent you text and email reminders

with a Zoom link before Zoom was a word.

That cost 150 bucks a year.

And that's just one simple example to

modernize the justice system and just make

the process better.

So when you got on the bench, was

it a general jurisdiction court?

Was it criminal?

Yep. Handled criminal, civil, domestic,

and then I also built a

couple of specialty courts.

So then I ran a reentry court and a

switch and certain probation program.

So at any given time,

over a hundred and some odd probationers

and specialty courts that I was

managing as well.

Wow, that's cool. So how early

did you start Zoom? I mean,

I know a lot of people picked

it up for the pandemic.

2014. Wow. I mean, I was elected in 2013.

So that online calendar started

shelf products. Before AI

was a thing, my speech was,

I can modernize a justice system for

under $1,000 a year in any jurisdiction in

under 30 days and I'll do it for you. Wow.

And I have no coding skills whatsoever.

Well, that's really, I mean,

it sounds like you went in,

I don't want to say

really shook things up,

but my curiosity's getting

the better of me now.

And I'm curious about how your peers

on the trial bench responded to your

blazing the trail technologically.

Look, judges are people

just like lawyers or people.

If you're in a firm and you go to the

senior leadership, some of them are in,

some of them are not

today, not tomorrow, never.

And so you've got all of

that. And so the beauty of,

and one of the reasons I ran for judge

to begin with is the beauty is if you are

the judge and you run your courtroom how

you want to run your courtroom. I mean,

people can't tell me no.

And so I was able to modernize my

own chambers and then those who were

interested, I would help them.

Those who were not interested,

there are 99 problems and that's not

one of them. And I just need your votes.

Hey, you don't need to do it,

but I need your votes to get the budgets

to do these things. And like I said,

it doesn't cost that much money.

So it really wasn't an issue.

And I say how easy it

is, but it's not easy.

And we all know how difficult it is.

Change management is very difficult.

But you can tell me no

today, but I'll outlast you.

How did that go? I know in Texas

when people jumped to Zoom,

it was by necessity in the pandemic. And

then as things started to open back up,

there was some kind of conflicts between

what the civil procedure rules allowed

and what courts were

doing. How did that work?

Were you able to kind of navigate the

rules and manage to set everything up

pretty easily or did you run

into some conflicts there?

Yeah. So I mean, prior to the

pandemic, and we'll use that example,

I used it for pretrial conferences,

the hey, how you doing?

Why are y'all driving two hours and paying

for parking when you can just jump on

a Zoom or just say, "Hey, how you

doing? Where are we discovery?

Have you deposed X, Y, and Z? Have you

issued propounded discovery requests?

Where are we on that? " I mean,

I never understood why you

would drive to court to do it.

And I know why as a trial court judge,

because that was your time to meet your

opponent as opposed to just picking up

the phone and doing it. You

use chambers as a meeting site.

And so the rules weren't an

issue for pretrial conferences.

For hearings, we weren't having

It was typically for pretrial conferences.

And we already had remote

witness testimony rules in

the civil procedure. I'm

sure Texas did as well. There

were some nuances to it.

So as the pandemic continued,

we certainly went and modified the

rules of civil procedure statutorily,

worked with legislators and lawyers and

got the rules for criminal and civil the

way we thought that we needed them.

So certainly there were some legislative

changes that need to be made post

pandemic. But again, when

you're dealing with a pandemic,

the constitution trumps every local

rule and every civil procedure that rule

that exists.

I mean,

people certainly became more open-minded

during the pandemic about how to get

things done. And so at least now

everybody knows what Zoom is.

And so maybe you're

still a tech evangelist,

but you don't seem like maybe you're so

far out there anymore that the things

that you rolled out to the

trial court. After you left,

I want to get into your current position

a little more, but after you left,

have the changes that you made

stuck in the district court?

For the most part. There were

other things for criminal.

We integrated text and email reminders

right into the case management system to

cut down on failures to appear.

And that still exists because it's

got baked into the way we did things.

Obviously,

judges have their own opinions on whether

or not hearings should be held via

Zoom or not.

We actually had a legislator

passed a law that essentially

says that you can opt in on the front

end and the judge has to allow you to

attend virtually unless there's good

cause why not to. So they did the whole,

the reverse on it to kind of

manage those issues. And look,

you said I'm a tech

evangelist. That's true,

but it's more of the justice system.

And so I tried horrific murders and child

rapes and all these terrible things.

And the best thing to do is walk up to

a jury and hand a photograph and not try

and show it on the big screen up there.

So tech for tech's sake is not what I

do.

Technology to advance justice

is what I propose and what I'm

a proponent of. And sometimes, like

I said, no tech is the right answer.

Did you find improvements in things like

non-appearances and people that were

pro se being able to navigate? Did

you see that impact pretty quickly?

Oh, absolutely. Again, once I did

it for civil, it was the same $150.

I flipped it to the criminal. I didn't

give them the opportunity to pick dates,

but I would say, "You're hereby

ordered to appear on this date.

What time would you like

to come? Nine, 10 or 11.

I got 100 on my criminal docket. I

can't hear a hundred cases at 9:00 AM.

So why am I making a hundred people

come to my courtroom and cause mayhem?

You might have to take the bus, you might

have to drop your kids off at school,

you might have to bum a ride. Tell

me what time you want to come.

You're coming on this day,

but tell me what time you come." And

when they would make that selection,

I would then have them sit down at

the computer and put their name and

everything in there. And then I would

send them text and email reminders to cut

down on failures to appear.

And those emails would also include

the phone numbers of all the public

defenders that were

assigned to our division.

It would include all the different

forms that they may or may not need.

So it was an information,

it was pointing to a website that I

built using Squarespace for 300 bucks.

And I just had a page for everything.

If you were in this probation,

you had this page. If you

had this, you had this page.

All the information that you needed was

there. Use it to reduce the barriers.

Once you reduce the barriers, look,

no more excuses. You decide, I decide,

and I've provided you the information.

Did you, I don't want to

call it client feedback,

but you get the idea where I can

imagine that in terms of user

satisfaction, so to speak,

that those arrangements in your

court when you're on the trial bench,

it had to make it just that much

better for the participants.

Did you get feedback like that from

the folks that appeared before you?

I mean, colloquially, yes. I

mean, I didn't send out a survey.

How was your experience today? Well, you

put me in jail, so I wasn't too good.

Or you ruled against my

client. I hate Judge Lagel.

I don't know what that looks like,

but certainly I think lawyers were

appreciative of the ability to look at the

calendar online and find a date that

worked for everybody so they don't have to

file motions to continue because no one

can consulted with them about the date

that was good for everybody. I

don't want to say I built an app,

but I had an app that you could save to

your home screen that was a web-based

app and it literally was click

here to schedule the hearing,

click here to drag and

drop your courtesy copy,

click here for sample jury charges,

click here for phone numbers. And you

know where those papers went when you sent

me a courtesy copy in the trash.

So just drag and drop it on my website

and then I'll be sitting there on a

Sunday watching a ballgame and I'm getting

ready for trial because it just hit

my iPad.

Yeah. I ask you about the

customer satisfaction.

It makes me think of another

episode that we did, gosh,

probably two years ago now about

procedural justice and the feeling that if

people feel like they're heard and that

the judge is listening and treating them

fairly, that that makes a big difference

in how they feel about the result.

Oh, it absolutely does.

Your time's just as important as my time

and I'll never be late. If I'm late,

call the cops,

something's happened to me because your

time's just as important and you're not

going to like every decision I make,

but I want you to understand that I

hear you, see you, I've read everything,

I'm ready to go and how can I reduce

barriers? No reason to make people ...

You always use the DMV example.

If you're sitting at the DMV for

three hours when you get to the front,

you're just mad and you're not listening

to what they're saying anymore.

And we use that with probationers.

If the whole point of people in these

drug courts and these specialty courts is

to help them get healthy so that there

are no more victims and hold them

accountable.

And you make them sit for two or three

hours and how are they going to listen to

the counselors that are trying to help

them through the issues that they're

dealing with?

They're not.

So using all of these tools to

build a better DMV enables people to

listen. It doesn't remove

who you are, doesn't remove,

I'm going to hold you accountable.

It doesn't remove any of that.

It's simply a way to help people succeed

in life so that there are no more

victims.

Well,

you spent that decade on the trial bench

and now you've moved on to the Fifth

Circuit Court of Appeal and

have been there for a few years.

How has that transition gone in terms of

your adjustment to the new environment

and assessment of their tech for justice

situation and what have you been able

to do or needed to do in

that transition? Yeah.

Look, it's a completely different

job. You guys are appellate lawyers.

It is a completely different job

than the trial court. And thankfully,

I came into a court that had already

built its own case management system for

the past 10 or 15 years.

So everything was digitized and it's a

phenomenal foundation and a chief and a

bench that are interested in

looking for the next iteration.

I want to move at the speed of light and

it takes a little time to get everybody

to understand and appreciate these

things, but that's good. I mean,

that's the way it should be.

It's not our job to be first.

It's our job to get it right. And so

some of the ... I wrote this morning,

I don't know if y'all saw it,

build a frictionless system and then find

out where friction needs to be placed

back in. Because when I was starting

out with these gen AI tools,

friction was a good thing. It

was good for my inexperience.

It prevented me from making errors. It

prevented me from doing things. Now,

friction drives me crazy because I

know how to use the tools and I'm like,

"Give me my F1 race car that I can go.

" But I always joke,

the private sector's building

F1 cars for dirt tracks.

And so how do we as a

court look at these things?

And it should be slower than I want.

I'm not saying my way is the right way

because I probably go too fast sometimes.

So it's a good thing that

I have colleagues that want

to understand these things

before we move forward. And

now, almost three years in,

we're looking at an on- prem AI servers

where we can code it and train it and

help out with central staff on civil

cases. We'll be able to help out

with intake at the clerk's office.

Now I'm looking for the budget

to do it and going forward.

How does it work in Louisiana

for the appellate courts?

Do you kind of have your own,

your court has its own

budget and decisions on that

kind of thing, management,

or is it handled on a higher state level?

No, we're not centralized. Each

court has their own budget.

So legislature's in session right now.

The judiciary has put forth its budget.

There's a line item for each circuit.

We've asked for it and hopefully

we get what we've asked for.

We never get all of what we've asked for.

And then something pops up where the

boiler goes out and the money that you

needed for the AI server

goes, "Scott, it's good.

I know you want your AI server,

but we're not getting it now because

we have a boiler what we have to fix."

Because again, it is a,

we are in a building that

we have to manage and hire

clerks and everything else.

Well,

whenever we talk to a lawyer or a judge

from another jurisdiction who is on the

appellate side of things,

I always like to ask,

how many circuits in Louisiana

for the courts of appeal? That's.

Five.

Five. Okay. We have a lot. We have 15.

So I always like to compare notes because

I'm fascinated by the states that only

have one intermediate court of appeals.

Right. Yeah, yeah.

Yeah. Five, that seems like a

pretty good number. We have a lot,

which is good for people like me and Jody.

Yeah.

So you've made the transition now

fully over the past three years,

which just happens to coincide

with the real just blossoming,

ballooning up of GenAI being available.

So now you can do even more than you

could before in terms of coming up

with some innovative strategies for

dealing with cases and your caseload and

management and all of that.

And have your colleagues,

I know you're kind of taking the reins

it sounds like on some of these projects,

but your colleagues have

been receptive, I assume,

to the initiatives that you're bringing?

Oh yeah. Again, it's a process. It's

a process that you start on day one.

There are people that argue

for more than I'm game for,

and I push back and say no. So again,

there's always going to be that.

So I can say, yes, my colleagues

are supportive, but yes,

my colleagues think I'm crazy

at times and that's okay.

But I think I do it in a measured

approach. Again, I always,

as it relates to these tools,

if you've seen my website,

I drafted a guideline for

judges, how to use GenAI tools.

And I always start with, hopefully

you're using enterprise level tools.

Hopefully you're not using free ChatGPT

or free Claude or free products,

but if you're at least using

the pro level $20 versions,

here is a guideline to follow.

It's not a tech guideline. It's a,

here's how to be careful. Rule one of AI.

Don't use AI.

Do your job first. Get a direction,

get a feel, understand the file.

Don't just put it into

a GenAI bot and say,

"What's the answer?" And so it just

walks you through 10 steps of how to use

tools to go through the

iterative drafting process

if you decide to use it in a manner

in which it's drafting. And hopefully,

again, you're not drafting 30

page opinions using a GenAI tool.

That is not what I'm suggesting

or referring to. But again,

it's this whole friction thing. It's this,

I want you to understand how these tools

work and here is what I have learned

and by the way, you already

know how to do this.

If you view the tool as a law clerk or

a first year associate or anything else,

once you read the steps, you go, "Oh,

I get what this is. " You never

rely upon a first year associate

for anything. Your central staff memos,

you're going to read them and go,

thank you,

but you're still going to go do your

job and you're going to walk through it.

And so now that the tools are being

baked into the tools that courts

already trust, it's a different

argument. So when I first started here,

we didn't have the co-councils and

the proteges baked into these Westlaws

and Lexuses that we could even afford.

And so the previous iteration of

what I tried to build was way too

expensive.

The tokens cost way too much and there

was just not a comfortable level of using

tools that we aren't familiar

with. And so I've always said,

until the tools get baked into

the tools that the courts trust,

you won't see adoption for the chambers.

Now that that's happening with your

Office 365s and Copilots being baked in on

enterprise level, with the

proteges and with the co-councils,

you're going to start seeing adoption

going forward. And then you have other

courts that are going, "I get

it. This is too much friction.

It's driving me crazy.

Give me the private

sector's version of this

tool." And so you're seeing that with

the clear briefs and the learning hands.

And this is not me pushing any

products. I'm just talking out loud.

The enterprise level co-count, I mean

Claude, enterprise level ChatGPT,

enterprise level Gemini. All

of these things are the next,

and I'm not saying it's next iteration,

but it drops some of that and enables

you to do more than necessarily

some of the friction tools. And

that's the way it should be.

I always joke that you should have to

level up before you touch the next tool.

You got to play the game and it

tells you, "All right, Judge Lagel,

you can now go to level two." I.

Think that's a great analogy.

I'm going to borrow that when I'm trying

to explain it to some people who are

working on some of these things.

It really is. Well,

so it seems like what you're seeing

and taking part in leadership of on

the advancement of use of

these tools in the judiciary,

there's a lot of parallel between that

and what's going on in the private

sector, or suppose it could

be public sector lawyers too.

And I'm sure judges face the same

basic issues in terms of we've

got hallucination problems still. You

point out don't use the free tools,

use the paid ones for privacy reasons

that get you that extra level or A level.

Preferably enterprise, that's even better.

But I wanted to focus on this newer

project that you've been a part of

called the Judicial AI Consortium.

I think you told me that can

be shortened to Jake, J-A-I-C.

Why don't you tell us about it and kind

of tell us your involvement in it and

look forward to hearing more about it

and how we can help you spread the word

about it.

Yeah. So Judge Maritza Braswell, she's

a magistrate judge out of Colorado.

She came up with the idea. I mean,

she and I have spoken on

panels together over this AI.

Who are the judges who are

using AI type thought process?

So she called me and federal

judge, she's a district judge,

Xavier Rodriguez out of Texas. So she

called the both of us and said, "Hey,

look, what do you think of this idea,

Jake?" And the idea is very simple.

Judges only ask stupid questions,

talk to each other about how you're

using it, what you're seeing out there.

Is it helpful? Is it useful? How

far should we go? Is that too far?

Is it not far? What rules are being

in your state, in your courts?

Everything from, do you have

an AI policy? If not, why not?

And the AI policy can be as simple as

you can't use it. I hope that's not your

policy, but your staff,

people are using it.

So our policy is very simple.

You may use AI in these situations

with these approved products

and you must tell your judge

that you used it. That's it.

So we don't want to be those two federal

judges that got hailed in front of

Chuck Grassley about their

hallucinations. And they said,

"My staff used GenAI

and we didn't know it,

and there were some mistakes

in there." So again,

it's our job to get it not to get the

first, it's our jobs to get it right.

And so when lawyers mess up, no

big deal. I mean, it is a big deal,

but it's no big deal. We could

sanction you. We can strike pleadings.

You can be embarrassed in front of your

clients. When we mess up, it's the law.

It could become precedent.

So it's a much different conversation.

And so I encourage people to play around

with it before they start using it in

their own workflows. And that's the

point of Jake. The point of Jake is,

here are my experiences.

There are other people like

me out in the judiciary,

and it's a great learning

tool for all of us. Again,

we're not recording anything. It's

just a conversation for judges.

So we have had a first meeting.

We had over a hundred judges

from across the country,

and then we are having what

we're calling pop-ins, little 30,

45 minute come if you can conversations.

The first one occurred last

week. There was about 20,

which is a perfect

number for these pop-ins.

And then we'll have a couple of more

before our next big meeting in April.

And so Judge Braswell

led one, I'll lead one,

Judge Rodriguez will lead another one.

And again,

it's just a great time for us to talk

amongst colleagues and figure this thing

out.

I mean, I guess it's

still early days for Jake,

but do you envision it turning

into a nationwide network of judges

or is it just going to- Yeah.

That's the hope.

Okay.

I mean, I think it already is. I think

we've already got over 200 people,

over 200 judges who have signed up.

We've had international judges who have

won to sign up and we're keeping it to

the US only for now. But yeah, I mean,

I think 200 judges with a couple of

emails is a pretty good starting.

Point. Absolutely. Did you use AI to

write your prompt to send out to them?

Absolutely. Why would you not?

That is a perfect use case.

Yeah.

Your point about the gravity

of judicial AI use is well

taken because if a lawyer, if

Jodi or I do it and we mess up,

you cite a hallucinated case and we

get sanctioned, that's embarrassing,

but that's more or less

the end of it for us.

But we're talking about

embedding it into institutions,

the extremely important third branch.

And so I hadn't really quite

thought about it that way,

but we're all kind of numb at this

point to seeing news reports about

sanctions of lawyers who

continue amazingly to me

to not vet their documents properly.

Although as you point out, I mean,

there are things that can happen and

I've seen it happen where the first year

associate is using ChatGPT in the gray and

not disclosing it to his partner or

her partner and that's a problem.

But boy, yeah, when they start showing

up in judicial orders or opinions,

which has happened, I mean,

the instances of that,

I've only heard of a few. Are we

seeing that becoming actually more of a

problem? I know you're trying to

educate so that that doesn't happen,

but do you know about how many

instances we've seen of that happening?

I only know of three, not

that it hasn't done more.

You've got the two federal judge

opinions that had to be recalled,

and then you have the state court

order, I think it was in Georgia,

a lawyer prepared an order for

a judge that order contained

hallucinated cases.

The judge signed it without checking

it and it went up and actually

their higher court just affirmed

the lower court's signature on that

order saying, "Y'all didn't

bring this up." I'm sorry.

Oh, wait, what?

You know you have to object.

I don't think it affected the final

outcome. It's probably the answer.

I haven't looked at it too hard,

but I thought it was interesting

that the Georgia Court of Appeal,

I don't know if it was the Supreme

Court of the Court of Appeals,

just came out with their ruling

on that hallucinated order.

So just the mere fact that a hallucinated

case was cited doesn't affect

the validity of the order is

the end result on that one.

That makes sense to me,

but there still has to be some

teeth in the fact that the

lawyer presented a proposed form of order

to the court with a hallucinated case

in it. And I suppose that's

handled obviously outside of,

can be handled outside

of the judicial process,

certainly outside of the

process of the merits.

It's a good reminder of the courts when

you're presenting an order, run it,

site check it first before signing.

Yeah, for sure.

Look, that's all that we did at the trial

court. After make a ruling, it said,

now prepare a judgment and have it

to the court within 10 business days.

Now we're going to have

to look a little closer.

Yeah. I mean, that's very common

in our state court practice.

Our state court judges are not

preparing their own orders.

And so this is a real risk.

And I hope that if nothing else

comes out of this conversation,

trial court judges in our state and

elsewhere would really benefit, I think,

from tracking what y'all

are doing with Jake,

because I'm sure that the judges are

hearing about this from other sources,

but just the idea that they can plug

into a resource like this and be informed

or better informed about the risks and

about things that are happening around

the country, to me, it just seems

like it brings a lot of value.

I think your ranks may explode and you

may wind up with more people than you

want. You're going to have to start

regional chapters or something.

Yeah, something. But

look, aiforjudges.com.

If any judge is listening and you

want to sign up aiforjudges.com,

there's a few articles on there

and a link to join and then receive

emails from us to attend meetings.

Oh, I think that's awesome.

I know you're talking about NJ,

the chamber side and the judge facing

side and how you handle your policies and

the tools.

Are you also talking about the attorney

and kind of other side of that? I mean,

do you have policies in your court or

statewide about how attorneys can handle

AI or what they have to do?

Yeah, the game plan. Again,

we've had two meetings so far.

The game plan is to find those issues

as they pop up and discuss them.

And AI related, of course,

is plenty of other organizations you

can learn how to do other things,

but anything AI related

that's hitting the courts,

whether that's how do we use it to

assist with access to justice issues?

How do I use it to speed the process

along? How do I use it to draft?

How do I use it for administrative orders?

How do I use it for all these and what

tools are we using and why are we using

them and what are the settings

and how do you do this? I mean,

anything related to AI in the justice

system from the judicial's perspective

will be discussed.

Has Louisiana created rules

on attorney use of AI?

I know some courts in

Texas have local rules,

but there hasn't been

anything at a system level.

Yeah. So I was the chair of

the Supreme Court's AI task

force, if you will. And

so I wrote a letter,

open letter to every judge

across the country saying,

"Please don't do these rules,

these orders saying you cannot use GenAI."

I did this as soon as MATA came out,

I was like, "For the love of God,

do not issue orders." It was,

"You cannot use AI." And I

was like, "Well, then they

can't use their computers.

You can't use GenAI." Well,

all right, you can use it,

but you got to tell me about it.

And I wrote an open letter to the entire

country saying, "For the love of God,

please don't do that.

" We have enough rules out there to

address these situations. And thankfully,

the Louisiana Supreme Court agreed and

took the approach and sent a letter to

the members of the bar reminding

them of all their duties,

all the various rules

that apply and say, "Look,

today we think the rules that exist

cover every issue that we're seeing."

We could change that, which is great.

That's the way it should be.

There might be coming time

where we actually need to

pass a rule because you're

not stopping the hallucinations. For

the love of God, just stop, just stop.

And if you're not going to stop, maybe

now is the time to come in and say,

"Well, we're going to mandate one hour

of AI CLE for every lawyer." Again,

I'm not a rule for rule sake guy,

I thought the hallucinations

would stop by now,

but now I'm on the bandwagon of we

should start mandating one hour CLE,

not jazz hands CLE,

an actual CLE that says

here are the risks and

benefits of using AI. Here are some of

the tools, here are some of the settings.

Again, we're not trying

to make everybody experts,

but hopefully you'll pause after

hearing that mandatory one hour AI

CLE and the fact that things

change every single hour,

you're always going to have good material

for another one hour CLE the following

year.

Oh, no doubt about it. Well,

I'm glad to hear you talk about

standing orders and local rules.

I assume that's something that's been

discussed within Jake or will be shortly

because I was so glad to hear what you

said about that because it drives me

insane when we have too many rules

that are very issue specific.

And I just saw on LinkedIn this week,

some judge, I can't remember who or where,

it actually issued or

required a certificate in

pleadings saying you did not

use gen AI or AI. And you point out,

you can't really say that. Do you use

a spell checker? Do you use anything?

Yeah, briefcatch. I mean, good grief.

If you're almost committing malpractice,

if you don't use briefcatch or something

like that. And so boy, I really favor,

and I've said this before,

but our ethical obligations are what

they are and they tend to cover all the

situations. I mean, they've

been in place for how long?

Model ABA rules in the states that have

adopted their own variations of them.

So I'm very happy to hear you talk

about discouraging the adoption of

specific local rules or standing

orders, which is even worse, I think,

addressing those issues because it doesn't

change anything about the ethics of

our job that we have gen AI.

Once you sign the dotted line,

I can hold you accountable.

And I don't care if ... I always joke,

how many of you all ever had a paralegal

or a first year associate and signed

the dotted line after

they handed you a copy?

Did you disclose to the court that your

paralegal prepared this or that the

first year associate? No, no, you

did not. You signed the dotted line.

That means you own it.

Well, and that seems like the

cases that I keep reading about AI,

it seems like it's a knowledge

issue rather than a rules issue.

They didn't understand, oh,

I thought when it spit out these cases

that those were good cases because it had

checked them. And it's just

a misunderstanding that no,

it's answering your question with words

that it's put together and they may or

may not be accurate words.

One thing that the folks committing

those errors need to know is you can

actually use GenAI to

avoid those problems.

There are specific apps that

will check you and make sure,

or you can use the tools that

are already there to avoid that.

And it's a great first use case for a

judge. Proofread this document for me.

Don't change anything, just

proofread this thing for me.

Just make sure that are

there any spelling errors?

Are there any misstatements of the facts?

Are there any misstatements of the

law? Site check it for me. I mean,

these are simple things that you're not

going to get yourself in trouble by site

checking something or by

proofreading something. It's great.

It'll catch things that

the human eye cannot catch.

Once you've read the same brief

or the same opinion 20 times,

you can't see it anymore. You go blind.

Oh yeah. We've all been

there for sure. Well,

I think that's a great overview of

what Jake is doing. And like I said,

it's still early days. And

so there's a lot yet to come,

but you gave the URL for Jake and we'll

certainly encourage every judge we know

to check it out at every level.

It sounds like it's something

that's going to be worth tracking.

I hope you do get to the point where

you have to create local chapters.

That would be awesome

because it means that people,

judges are interested and they want

to learn about it and they want to

potentially use it to help.

I know you talk about AI and we've

talked about it on this show too,

about AI as an access to justice issue.

And more efficient justice on the

judiciary side plays into that,

dovetails into that.

So I think it's very commendable

what's happening. And we're certainly,

I'm not familiar with Judge Braswell,

but Judge Rodriguez is just down

the road from me in San Antonio.

I'm in Austin and certainly

very familiar with him.

And he has written and spoken a lot

here locally in our state on these same

topics.

And so it's great to know that we've got

leaders in the judiciary standing out

front and really encouraging other judges

to speak up and pay attention to this.

Is it fair to say, I mean, well,

maybe you don't know what the age

distribution is among Jake folks yet,

but you've got the digital natives

who are comfortable with this.

And now we're all old enough

where we see digital natives,

not necessarily us growing up,

but digital natives are entering this

phase of their career where they might be

judges.

You probably can't make any generalization

about the distribution of age among

the folks that have expressed interest

in Jake so far. But I assume that the

ones who are older are just generally

trying to educate themselves about AI and

figure out what's going on.

Yeah. I mean, it is to

some extent because look,

I think I'm older than Judge Braswell

and Judge Rodriguez is older than I am.

And I don't think it's necessarily

because you're a certain age,

you can do this better than the

other. Again, Judge Rodriguez,

if you're listening,

I'm not calling you old,

but Judge Rodriguez knows how

to do this just as well as I do.

And he's at the end of

his judicial career.

I think he's been on the

bench almost 30 years now.

So it's not like I'm the old person

in the room and I can't do this.

And if you frame it properly,

then you can tell people you already

know how to do this. And frankly,

younger people using these tools scare

me more than older people using these

tools because the younger

lawyers and judges may

not have the scarring necessary to

understand the nuance of the law.

And it's this ... I use this example

and we talked about war stories,

so I'll give you one right now.

When I first became a prosecutor,

I was asked to third chair, not first

chair, not second chair, third chair,

a horrific homicide, multiple victims,

DNA everywhere, mixed DNA samples,

fire covering everything up. It was awful.

I had no idea how to try that

case and I had no appreciation of

how to present the evidence. But you

know what I knew? I knew PowerPoint.

This was right after the phone

board and I was like, "Look,

that phone board stuff is nonsense.

Let's do PowerPoint." So I convinced

the senior prosecutor to allow me to.

I used the diagram of the scene.

I had all the evidence laid out.

I had all of the different mixtures of

DNA color coded on every single piece

of evidence.

And he would say something and I would

push a button and you'd get the next DNA

sample.

It was horrible.

It was the worst presentation of evidence

ever because the senior prosecutor

knew the timing, but he didn't

know the tech. I knew the tech,

but had no idea of the timing and it fell

very flat. It confused the prosecutor.

It confused everybody.

And that was not a good thing because

I didn't understand enough about trying

the case. I didn't understand the

nuances of a multiple homicide.

I didn't understand any of that. So it

didn't work well. But on the flip side,

there was another homicide.

This was before cell tower

records were really a thing,

but I was the young buck, so they

threw the records to me and said,

"Figure this out.

" And I remember the Saturday where I

figured it out and I could pinpoint the

defendant was here at the time of

the murder right on the murder scene,

and I was able to track his flight to

another state using cell tower record. So

that was an example of when the

youth, if you will, was helpful.

And so I think it's this back and

forth that we need to think through.

And the senior lawyers need

to be intentional about not

just turning over to the

juniors,

and the juniors need to be intentional

about learning the law before they just

start using PowerPoint

or GenAI in this case,

because you're just

going to hurt yourself.

Yeah. I think about some people

I know and who are older than me,

who have really made the effort to

learn this stuff. And it's really all,

it seems to me just about mindset and

understanding the big picture of how it

... You don't have to

know exactly how it works,

but you just have to know how to use the

tool, what the limits of the tool are.

We hear all the time the idea that, well,

you've got to have the human in the loop.

And I think you talk about that some.

And I think you even maybe drilled down

a little further on that and say the

right human in the loop, not

just any human in the loop.

What's your philosophy on that, Judge?

I mean, you hit it. I hate the

terminology human in the loop.

It's the right human in the loop.

How old were you, Todd and Jody,

when you got your driver's license?

I got a hardship license at 15.

So let's use that as example. My son's

20 now, so when he was 16, he got a car.

Assume that I can retire at some

point and the dream car is the

Turbo S911, right?

And so let's assume my 16 year old

came up to me when I had the nine

eleven Turbo S and said, "Hey,

dad, I got a hot date tonight.

Can I take the Porsche?" And I'm

like, "Sure, buddy, here you go.

" And I tossed him the keys.

What's going to happen?

He's going to put that

car into a tree or a wall.

He's going to hurt himself and he's going

to hurt the little girl that he's took

on a date. My point is,

just because you have a law

license doesn't mean you

should get the nine eleven

or the GenAI tool. You

don't know how to drive yet,

even though you're a licensed lawyer,

even though you're a sitting judge,

you don't have the skills necessary to

handle that much power. And that goes to

my friction conversation.

And so just because you're a labor lawyer

doesn't mean you're going to become

the best med mal lawyer that ever lived

because you have a gen AI tool. No,

you're still going to hurt yourself even

more so because you're going to be so

confident that I'm a great lawyer that

you think you know that skillset as a med

mal lawyer. And it's the same thing

with the court and everybody else.

Just take it slowly, do

something you know really well,

learn the strengths and weaknesses of

the tools so that you can go from the

Buick to the Cadillac,

maybe to the Honda and then to the Porsche

and the Lambos and everything else.

So start small, start slow,

do something today in a

very controlled way so that

you can learn the

strengths and weaknesses.

And once you learn how to use

ChatGPT and you go over to Claude,

they're completely different and Gemini

is completely different. They act

differently.

Well, and this kind of dovetails I think

too with something you've talked about,

about drift.

The tendency historically in judicial

chambers is you've got the judge and

staff, can be administrative

staff, lawyer staff, and the judge,

depending on the court, is

going to vary a lot by judge.

Their level of reliance on staff

for original thought and drafting

varies a lot. But if you've got

a judge that relies on their,

say their law clerks a lot,

that sounds like to me,

a situation where that drift might come

into play. There might be some risk.

And I think by drift,

what you mean is the tendency

toward over delegation.

And maybe in this is perhaps

the friction area for how AI and

chambers is going to really work.

I guess what I'm coming to on this

is it seems like, yes,

judges can dip their toe in

this and try to understand it,

but there's probably going to

have to be some level of training,

whether it be just in chambers, chambers

by chambers, or say in your court,

court wide, if there's a big

rollout of using AI tools,

if you've got an AI

server in your courthouse,

what is the plan or what

would be the ideal plan,

I guess is the way I would ask it,

for how you ensure continuity and AI

use and that there's not folks who are

getting behind the Porsche, the wheel of

the Porsche and driving it into a tree.

Take an iterative

approach, go slow. Again,

build your dream world and then back

off and put friction back where fiction

belongs. So dream big, build

it, don't implement it.

And then once you've built the

dream world, go, "All right,

we can't allow that drift to occur.

Put that friction point back in right

there." We're not allowing this thing to

draft an opinion.

We're not allowing people to not

struggle with the white page problem.

I'd never, ever want people to

hand this over and say, "Hey,

give me a ... " I'm not a first

draft guy. I don't agree with it.

I know some do agree with it. I don't

agree with it. I don't ever want ...

We hear this word bias all the time.

I'm not talking about that kind of bias.

I'm talking about confirmation bias so

that if something is already on a piece

of paper,

it's going to be the tail wagging the

dog and you're going to go in that

direction and start down a line

where, no, you do the first draft,

you struggle with the ideas, you struggle

with the concepts, fight through it,

get something, and then look,

use it to clean things up,

use it to say something better,

use it to finish a thought,

but don't ever allow tail to wag the dog.

And so I am a

big opponent to first

drafts. I don't like them.

And I'm not talking about

administrative orders. I mean, look,

if you do something wrote all day every

day and you're just replacing names,

sure, use it to draft administrative

orders. And even on, again,

like a writ that is a motion for summary

judgment denied on the showing made.

Something very simple, one or two,

that's not white page problem

that I'm talking about.

I'm talking about handing over the

pleadings and saying, "What do you think?

" Grock. No, let's not do that.

That's not the social contract.

It's not anything that we signed up for.

I would hate anybody to ever hear that

they lost their kids because Grock said

so. If they were to use Claude or

ChatGPT, they would've had their kids.

That's not the social contract. If

you want to do that, go mediate it,

go arbitrate it, that's fine,

but that's not the justice system

and it's not something that I'm.

For. So let's talk, just

flesh out a little bit for us,

the idea of the social contract.

What do you mean by that?

I mean, you and I signed up for human

to judge an issue, period the end.

If you want the bots to do it

because it's going to be faster,

all these buzzwords,

consistency, transparency,

all these buzzwords that we can have

an entire hour debate on those terms.

Whose singularity are you building

for? I'll just leave it at that.

That's not the social contract. I mean,

we signed up for, we have a dispute,

we bring it to the courts, we

ask the courts to decide it.

We then can say Judge

Schlagel is horrible or not.

Part of the feel we talked about earlier,

Todd, part of the feel is to say,

"I hate that guy.

He is a terrible judge and I'm taking

him up and you can take me up." Can you

imagine somebody losing

their kids and saying,

"I hate ChatGPT?" There's no feeling

there. That is an algorithmic

decision that was made and that could

change tomorrow because it was asked a

different way or the algorithm changed.

That is not the social contract.

And again, I'm not saying don't go use

these tools to mediate and arbitrate.

That's a decision that was

made by parties consensually.

There's no consensual decision

when you're in the justice system.

I got sued or I have a problem

and I'm bringing you into court

to handle this issue. And I think

that your voice matters, Todd,

your voice matters, Jodi,

and my voice matters.

Voice matters and it needs

to be handled by a human.

Got it. That's like the

ultimate case of drift.

If you're just handing

over everything to the bot,

to the robot over lords,

I understand. I get it.

And it takes me back to

something I read about a lot.

And I think this is always a good way

of looking at what AI is and can be

as thought partner. And you

describe it perfectly, Judge.

Start with your blank first

draft as we hear it described,

starts with an S. And get input on it.

And is it logically consistent?

What errors are in it?

Things like that are

great uses of the tool.

The idea of your discussion of that

overall concept made me think about

something I read about not too long

ago too, that I think it was the AAA,

American Arbitration Association

now has this chatbot generated

arbitration process,

but the parties are consenting to it.

So there's not a social contract problem,

right? And there are very

close parameters, type

parameters built around it,

as I understand it.

And that's not what we have in terms of

the general use of general AI or gen AI

and to generate,

go from zero blank page to opinion

with no input from the judge,

just what do you think?

I'm still curious how that AAA

thing's going to work out, but

we shall see, but that's not the way

the judicial process ought to work out.

So appreciate your comments on that. Well,

we're getting close to the end

of our time together today,

but I do want to ask you,

in terms of the practitioners

that come before you as a judge,

do you have some concrete

things that lawyers who are,

let's say coming into your court,

tips that you might give to them in terms

of the use of GenAI? It sounds like,

as I said earlier,

there's a lot of parallel between the

way that judges use it and the way that

Eric shouldn't use it and

the way that lawyers should.

Is there anything specific to the lawyer

use case that comes to mind that you

want to bring to folks' attention?

Lawyers should be using this

to write better arguments to

the court. I don't know why

you wouldn't use it. Again,

this always premises you

understand how to use it,

but if Todd is a great appellate lawyer,

Todd will write a better brief after

he's written his brief personally or in

tandem with that will make it clearer

for the courts to understand what he is

trying to say. Again,

you're not turning it over.

The bot can make a suggestion.You're like,

"That's terrible." I'm not telling the

judge that. If you use it critically,

then you're going to write

a better brief to the court.

And this is an appellate podcast. So

you might be a great orator, Todd,

but I've pretty much read

everything. I'm pretty good.

By the time you get there, there

are some issues that are very,

very confusing that it's important for

oral argument and I want to hear about

it and everything else.

But I mean, for the most part,

you're writing your

opposition, your reply,

that's already forming the ideas

before we even get in there.

And so I would absolutely use it to

tear your opponent's brief apart.

Use it to prepare for oral arguments.

Use it to prepare your own briefs.

I would use, they have Google Notebook LM,

which is a sleeper that

nobody talks about. I mean,

you can feed briefs into it and literally

create podcasts for yourself to listen

to arguments, to get ready for your own

oral arguments. Again, go for a job,

use it to walk,

delete it and do it again and see

what they are actually talking about.

It's just a great way to get

you thinking through arguments.

Ask these tools to be a panel

and what questions are you in?

Feed at the pleadings.

"Hey, you're my panel.

I'm about to go for oral arguments. Ask

me a bunch of questions and just go back

and forth with it.

Just get hit by the panel of AI bots

so that you can get ready for oral

arguments." I mean, these are

great ways to use these tools.

As an advocate, I would use it all day,

every day because you

want to go a certain lane.

You want it to confirm everything that

you're saying. That is a great use case.

And my job's to get it right. Your job

is to convince me that you're right.

So use these bots to sharpen your tools.

I think that's great.

It's our tradition I always wrap up with

a tip or a war story and you've given

us some great tips,

but I don't know if you've got more or

if you have a war story you want to share

as we close.

I won't give you another war story, but

leave it at this. It's going to be okay.

We've been here before. We're

going to get through it again.

I remember about 20,

25 years ago when you were in that first

CLE and that lawyer walks up to you

and says, "Hey, Todd, Jody, come

over. I want to show you something.

I got my client files in

the cloud." And you're like,

"The cloud. What are you talking

about the cloud?" You're like, "Yay.

All my client files are up there

in the air. And if I want it,

I just pull it up and be

like, you're going to jail.

You're going to lose your law license.

What are you talking about the cloud?"

But what did we learn from that?

We learned a couple of things. Today, if

you're not using a cloud-based storage,

you're probably the one

committing malpractice. Today

we learned that you don't

use free because free is bad.

We learned if you're

going to use cloud-based,

you need to make sure that you have the

right license for the right types of

files. We understood if

it's a medical record,

it needs to be a HIPAA compliant

cloud-based storage and all of the same

lessons are here today

for GenAI. Don't use free.

Make sure the settings are right.

Don't let it to be trained off of you.

All of these different things we

understand. So we've been here before.

We'll get through it. It's all

good. That's what lawyers do.

We adapt. We keep moving forward

and we figure things out.

So you'll be good. If you

need a hug, ask Todd or Jody.

I'm sure they'll give you one.

We'll put up a booth at the

next appellate conference, Jody.

Free AI stress hugs.

Yeah. Good luck with that. Trying to

get appellate lawyers to hug each other.

We'd make sure that there's

about a 30-foot barrier

around us if we tried that.

I love it. That's great. Well,

Judge, this has been great.

Sure appreciate your time. I personally

know of places I can find you online.

Do you want to just recite

for our listeners the best

places folks can find you

to track what you're doing?

Judge Lagel.com. All there.

I'll also commend, for those

of you who are on Substack,

I'll also recommend your Substack,

cover a lot of great topics there

and things that really make me think.

So I appreciate everything you do there.

And thanks for being

willing to come on with us.

I know that when some strange appellate

lawyer from Texas reaches out and says,

"Hey, Judge, would you like to come

on our podcast?" I'm sure you're like,

"Well, who are these guys?"

But you've been great.

We certainly appreciate your time.

Thanks for having me.

Thanks for listening to the

Texas Appellate Law Podcast.

If you enjoyed this episode,

please share it with your colleagues

and rate and review the show on your

favorite podcast platform.

To connect with us,

suggest a topic or inquire

about being a guest,

visit textaplawPod.com or

find us on LinkedIn and X

@textappLawPod. Produced

and powered by LawPods.

The views expressed by the participants

on this podcast are their own and not

those of their law firm's

courts or employers.

Nothing you hear on this show establishes

an attorney-client relationship or is

legal advice.