The information provided on this program is for informational and entertainment purposes only and does not constitute legal advice.
Accessing or viewing this information does not create an attorney-client relationship.
Every case is unique and you should consult with a licensed attorney in your jurisdiction before taking any action.
Welcome to Iowa Law Talk with Joe Piper and Nick Cratelli here on the Mighty Blowtorch KXEL, your Catbird seat noon hour Saturday program weekly here on this 50,000 watt player channel beacon, which carams.
Across the United States on the AM signal loan 21 states and for Canadian provinces up there and guys It's 1540 on your AM dial 100.5 on that FM transmitter out of Sea Rapids literally North America's first 50,000 watt third channel broadcast over a century ago that great tradition perduers and with our very popular weekend original programming here with Nicholas Cratelli JDS Squire Drake law alumnus
Joseph Piper, JD Esquire, University of Iowa Law alumnus.
Nick and English barrister as well with offices in Des Moines, New York and London.
Joe, of course, with Ag and Business Legal Strategies.
An uncommon law firm out of Hiawatha, adjacent Marion and Cedar Rapids.
Guys, it's so great to have you back on.
Hard to believe we're a week away from Labor Day weekend when we will still have original programming for you.
We never take a vacation here because the wheels of justice, they grind exceedingly slowly, but they never stop because they grind exceedingly fine.
Joe, how are you today?
I
think it's
a
beautiful day.
We've had some great
uh change ups in the weather from from hot to cool to rain to dry nick or tell you how things being in the morning for
you well i'm a little sad john the fair is over i know no for some reason okay every time the fair ends we kind of go on the last day it's like oh darn the summer's over when
the fair
shuts out this summer's over yeah so this year i don't know if i told you lila lila entered
a dollhouse into the dollhouse competition.
I did not know
this.
It's not number two.
Whoa,
runner up champion,
runner up.
Yeah, this is the second one she's put in.
She got, a few years ago, she got number one on her haunted house.
This one was a gingerbread house with a whole bunch of, you made out of candy, or it looked like candy.
I thought it was candy.
Anyhow,
I told her she could have
gotten number one.
John, she could have gotten number one.
Had it been real candy, but it was
fake
candy.
She's an accomplished equestrian, too.
There is no limit to this woman's talents, the loving, gracious
Lila.
Competition dollhouses is not a blood sport for this lawyer, but she can't... That's the competition in her video.
Now all of a sudden we've got the beginning of...
of fall coming and with that the fall trial calendar and the courts open up.
So
why is this important?
Why the fall and what is it just about coincidence?
No, historically in the UK and then run by historically folks I mean like 1400 AD you know the courts of fall season
opened up on the 1st of October.
That's why the U.S.
Supreme Court October term and other courts have just historically done that.
We started around September here in the district courts, but now we're all gearing up for trials and hearings and that type of thing.
So the party's over, it's time to go to work.
And that brings us to something I'd like to chat about.
I want to chat about AI at some point and how we utilize that.
artificial intelligence AI artificial intelligence well we used
to say AI was artificial insemination
right
i
guess and i'm not
happy no joke well we don't want to go into any cd discussion
of our poor listener folks
Right in.
John, will you give him the email address?
Nick,
this is the perfect segue to do this.
You know, guys, you listen to our show live here on 1540 or at 100.5 or you can listen live.
Just go to your favorite browser and type in KXEL listen live and you can listen to this from anywhere in the world.
But also this show is immediately put up on podcast the following Monday on KXEL.com.
Click KXEL.com.
and then click on what we call the hamburger menu, the three horizontal lines, and you click under podcasts.
All of their programs will be listed and you will see a bevy of Iowa Law Talk programs now as we are in month four.
You can click on all of the previous episodic treaties of excellence.
However, you can access it through SoundCloud and other podcast formats.
by simply again in the browser of your choice, just type in kxellawtalk and it will take you to the programming.
Now, we are solicitous of your opinions and questions.
We've already had several come in from sources.
We'll have some more again next week.
We want you to write into iowalawtalk at kxel.com.
That's Iowa Law Talk at...
kxel.com and we thank some of our prior folks that have called in live or whose entries we have read on the air and we will get back to that in even more voluminous ways beginning next Saturday.
Back to you, Sir Nicholas.
Well, folks, your input's important to us because we want to make sure that we're, you know, that we're touching on issues that you have an interest in or a question about or that type of thing.
You know, we're not here to give you, you know, get involved in your legal issues, but we're here to generally give you some education.
and knowledge about how the legal system works and how lawyers work and that type of thing.
Secondly, and I have a kind of a hidden agenda for me, and that is I'm a huge supporter of AM radio.
I believe it's critical that we keep our AM broadcasting stations alive and well with a lot of listeners because folks, this is where we get our weather news, our emergency news, matters of that nature.
You're not going to get it over FM.
Okay, FM has a short, you know, it's a short distance form of transmission, but AM is not.
So we need to keep the AM's channels around.
And this is how we do it.
So please write in, let us know what you're doing.
Nick is more important than ever, not just AM radio, but local live AM radio.
KXEL stands out as a beacon for still retaining local
programming, local on-air personalities throughout the day, local weather, local news, local sports, local life coverage of events.
Most major metropolitan radio stations now in Central and Eastern and even Western Iowa are married up with corporate radio programming from the coast, which basically, for instance, the Vaunted WMT Radio, the legendary giant at Broadcast Park 600 AM,
announced the other day, no more local personalities on WMT radio, bereft of any local on air talent.
So KXEL guys, this is the place to be.
50,000 watts of clear channel excellence, all.
Local, Off and Live, and Jeff Stein being a big part of that anchor system.
Back to you, Nick.
I digress.
Well, it's important and the reason is we've got the best local talent there is right here from the Cedar Rapids WMT area and that's Mr. Joe Pepper.
Joe, okay?
You are the local talent in your area of the world here and we've captured you.
Now, I realize that the KXEL signal
Okay.
Well, Joe, where does it go?
Or do you mean, John, what's the signal path of KXEL?
It's
not going along.
Yeah.
So just the am down 21 US states.
So think of the white and think of the white into Big 10 Conference, Pennsylvania, Ohio, Michigan, Indiana, Wisconsin, Illinois, Minnesota, Iowa, in the code is Nebraska, Missouri,
A little bit of Tennessee.
John, that was a
setup.
That was a setup.
Because my little ham radio station this morning got all the way to New Zealand and I talked to the guy down there.
Okay, New Zealand, Australia, way down.
I'm
going to see and raise you and say, if you get on the listen live KXCL.
You can listen to it from the space station, or from New Zealand, or from the Antarctic, just like Art Collins did from broadcast part, not far from where Piper is.
That's okay.
Little KZero, PCG.
Talk to the space station.
Ham radio, still alive, folks.
Don't count us out.
Anyhow, we digress a lot.
So, Joe.
Talk to me a little bit about how you're using AI in your prep.
So using AI for is to summarize documents.
We get a lot of loan documents come in and the loan documents are all of our clients things and I want to know for sure what's in them.
I'll get documents that you know maybe hundreds of pages of loan documents.
I'd like to know what the collateral is.
I'd like to know what the terms of loans are and yes I can go
figure that out on my own, and it'll take me a long time, or I can ask the right query of AI, and inside of less than five minutes, I can have several hundred pages summarized for me, which really makes my life a lot easier.
Yeah, I have to go check it to make sure that it didn't hallucinate, but on that type of query, it doesn't do a lot of all this.
I also use AI.
I put in a query yesterday, and the question was, how can I have my website
attract more clients that are searching for it using AI.
So AI told me what I needed to do to make my website a better website.
We're also using AI to help us research theories of cases and what has happened in different things.
Now we have to be careful if we don't want to expose our client documents to AI.
When I say I'm looking at
using AI to look at my client's documents.
That's a closed AI system.
It's not going out to a large language model as such, because if I send out to a large language model, then I could have broken attorney client privilege.
When Nick, you've had that happen as I understand it, when you pay clients that have decided to do their own research and check things out and tell you what you ought to be doing.
And if I recall it correctly, you told a client,
You wouldn't take his case because he had breached attorney client privilege.
Is that correct?
Yeah.
And what's happening here is in my world, folks, this AI thing started about three years or so ago when it really started to bubble up and we paid attention to it.
And we say, well, how does this going to affect the building?
and delivery of the legal services.
Three years ago, it was fairly inaccurate.
You couldn't rely on it, and Joe, you accurately used the term hallucination, because you couldn't believe what it would give you.
It wanted to please you, so it would make things up.
And that, of course, it got off on a bad start.
Well, that's all history now.
It still hallucinates, and it still can make things up.
But here's the problem.
In the past, we didn't quite know how to use it.
So we were turning to it to answer serious questions just by posing the question.
We may not know, didn't know how to frame the question or how to engineer the prompt that we wanted to return the information that we were requesting.
And as a result, we got misinformation.
Well, that's three years ago.
And over the last three years, we've gotten better and better at learning how to using it.
I'll say that it has trained us as we have trained it, as we have trained it.
So now we use artificial intelligence mainly to enhance our real intelligence.
So we use AI to enhance our eye.
What we've learned, Joe, is that as lawyers, we are still involved
and manage the case.
We are still the person who's ultimately responsible for diagnosing the legal problem, coming up with a way to verify the
correctness of our diagnosis, evaluating what the options are for resolution, and in fact, choosing the best option and putting it into play.
Now, that sounds an awful lot like medical doctor talking, diagnosing, the evaluation plan.
It's the same thing that lawyers do.
Any of these activities are all the same.
We just follow that algorithm.
But what we've learned is that artificial intelligence can help us.
So what we've done
and we're urging other lawyers to do, is number one, lawyer, you're still involved with the case, okay?
You are the one that has to be responsible for it.
Number two, use AI to augment your real intelligence.
Now that means you've got to learn how to engineer a prompt.
And now we have these wonderful things called skills.
Now skill is a series of prompts that you can program way in advance and tell AI
what to do.
I like to say it's an awful lot like if I had a robot here I would program the robot to you know as soon as I came in the office at night in the morning to go get my coffee to get my diary to get my docket list and have it all ready for me and and uh so I didn't have to do it well AI does that now GPT scheduler will do something similar to that except get the coffee
But, you know, you need to know how to engineer the prompts.
So as lawyers, we're all learning how to be prompt engineers.
So we put at the bottom of our work, Joe, this work product was engineered by lawyers, augmented by artificial intelligence and verified for accuracy by our staff.
So we put a notice at the bottom of.
of the work product so that there's no misunderstanding as to what we're doing and how we're using it.
What notice do you guys give your clients with respect to the use of AI?
Well, that's one of the interesting questions here is what notice do we have to give them?
And should we have a separate page of our engagement agreement that says this is how we're going to be using AI in their case and asking for their authorization?
I think if we're going to meet what the ABA, American Bar Association, requires, you can't just put two sentences in your engagement agreement that might be many pages long saying that we want to use AI.
You need to tell them what you're going to use it, how you're going to use it, and how it can enhance it.
their case but you need to have a separate letter if you will where you review it and they sign it so if they know that it's use it and they may say no don't use AI on my case.
Well you're right
Joe.
I think there's an ethics opinion, an ABA formal opinion, I think
it's
518.
We put a, it's right in our engagement agreement, the two, three paragraphs in there as to how we use it.
And more importantly, we notify them as to how the protocol that we use with regard to secure their confidential information.
And then we require them to sign an acknowledgement that they've read and understood it, and that they understand the risks.
And by the way, Jill, that's no different than the requirements that we had for when emails came out.
Correct.
When people start using emails, okay, the rule was you've got to notify people that email communication is not particularly secure, okay, can be intercepted and you have to get their consent to use it.
Well, if we were to do that now...
who say, well, that's ridiculous.
Everybody uses email.
In fact, the mail that we get here is completely changed.
Okay, we get a lot of advertisements and that type of thing.
We don't get any communications letters.
They all come by email.
So as time has progressed, people have gotten more comfortable with technology.
And I think you're going to see the same thing with AI.
But even if we see the same thing with AI, that doesn't lessen our duty as lawyers to protect the confidential information.
You're right.
Yeah, you got to scrub it.
You know, you got to go through, sanitize it.
And additionally, you have to make sure that you get the right AI product.
Correct.
You know, the off-the-shelf chat GPT or Android or Anthropic Clawed is okay for off-the-shelf work, but the stuff we do isn't that.
So we have to have a more robust, HIPAA compliant, confidential security and that type of thing to deal with it.
Now what we're seeing
and this is kind of exciting is that you can get your own have your own large language model which is the brains of the whole thing folks host it in your own office so it's behind your firewall so it never has to go out friend of mine is building one using a raspberry pi computer now this is a little computer that's about the size of a cigarette package and i'm from shaunel
His ears are perking up as he hears this.
And what he's been able to do with the Raspberry Pi, and analyzing financial documents and that type of thing for accuracy, he's pretty impressive.
It doesn't have the horsepower that some
of
your normal larger computer models will have, but we're getting there.
Now, if this works, now I got all that power in a little box in my office.
Now I've died and gone to heaven.
Right now, I think it's great because I got five five now artificial intelligence assistance if I have that thing in the office I can add one more and it'll be ultra secure so Is Sean there is he
yeah?
I'm gonna say you're talking about using a raspberry pi which yeah, you understand it's been very limited But uh, yeah, you can go out and buy hardware that that can run your own high-end model pretty I mean
pretty easily.
You can buy a purpose-built server for, I think, Nvidia selling one for like $5,000 still.
Think about that.
Yeah, then you can write it with your own rules even.
Value, that's incredible to us because now all that computing and all that information that we're required AI to look at and to analyze resides behind the firewall in my
office.
Safe and secure behind the...
Well, then it's your responsibility, right?
Exactly.
And then I don't have to worry about, you know, what Anthropic does with it, or what you, you know, where it goes, how they analyze it, because it's right here in the thumb drive.
So now that's
the goal for me.
So if you really don't want Anthropic or Claude or anybody else learning from the data you put in.
Correct.
Right.
And Joe, they say that they say that when you when you have the page description that says it will not learn from your information, it will not retain it.
And you got to remember to go in and and and delete it and all that.
And I'll take them at their word.
But when I start dealing with my client's information, I want my hands on it right here.
And there's certain certain communication between a lawyer and a client that's got to be from
My lips to their ears or their lips to my ears.
I just I don't even trust emails with it.
So why because I've seen so many email breaches.
Yeah.
No, with AI in general anymore, somebody could blow someone's voice pretty easily.
Oh,
we've done that.
So what I thought I would try is, so I, hey, Sean, I cloned myself.
Oh, good.
Oh, yeah.
So
there's a program called HGN.
So I got this picture of me and, you know, about five minutes worth of talk and I get this thing that looks like me to say anything and sound like me.
So that, okay, so now I got the virtual me.
Can I actually put it to use?
Can I tell this thing?
go into my service log and give me give my clients a report for everything that I've done last week and what and then my to-do list what I'm going to do next week so just automatically on Monday morning they could log in here's a complete audio visual report of what I did last week and what I'm going to do this we can just schedule it so that it's done automatic
So I don't have to get with very, very little involvement with me.
And I got the prototype working yesterday.
John, you and I were in trial all last week, so I had some downtime to look at it.
But I got the prototype working yesterday.
And if this works, now I got AI as a robot actually doing it.
So I showed it to my wife.
I says, hey, look at who's a virtual me.
And it'll say all this.
She says, I thought you liked our client.
And I go, we're talking about it.
She says,
you know a virtual you one of you may be enough this is the support I get from my
law partner okay but now Nick you're talking about your virtual uh you giving the report will it have your enthusiasm about
points or your lack of enthusiasm about certain points that may be brought up, that you would normally put in if you were doing it yourself.
You talked to Lila, didn't you?
If you're right.
Of
course you could see that's what she was saying.
She says you emphasized all the wrong parts.
I says, what are you talking about?
She says, why'd you get excited about this?
There's nothing excitement about what we did on that.
And what about this one, though?
You know, so that's exactly where the problem was, is the emphasis.
So I toned
it down this morning, because you can change all this stuff, folks.
It's a lot of fun to play with.
Well, one thing
you
can do is that within AI, you can tell it.
You ask it to write a letter, to write something.
And you can tell it what perspective to take on this.
Folks, this stuff is fun.
Listeners, you got to try this.
This is really fun.
So we had a situation
where the explanation point is that how many you add is also
the whole bunch of commas.
That'll slow
down.
So
I did a mediation statement and it was a boring, uh, my typical way that I would draft it for a media in our case.
Uh, so I got.
playing around with this and said, well, draft this as if Mark Twain wrote it.
And it did it.
Okay.
And it was really pretty good.
So I turned that in.
I did a brief because you can change the tone.
I drafted all this brief.
I put it into A.I.
and it says, write it like Justice Scalia would have written.
Now, think about this show.
You get a whole different dynamic when you have a Scalia tone.
What's the side-byes time look like on
those?
It's an Italian thing, folks.
But it wasn't all that bad.
A little too forceful.
So I said, tone it down.
Make it like it was written by Chief Justice Robert.
And it did it.
I thought, well, look, if we can do this, think of the fun we can have with real briefs, because we know how every justice of our Supreme Court writes.
I mean, they're good.
I mean, their work, folks, is excellent, but they all have different styles.
So if I could get it to write like one of those styles, just an additional way of putting forth, of knowing your customer.
I used AI to help me proofread a book that I wrote.
It's called Tips for the American Trial Lawyer from an ancient Roman orator.
Now, that's not me.
I'm Italian, not Roman.
But what Cicero taught us is to know your customer.
So I thought, okay, I can use this, know your customer, because I know how my customer writes, I know how they sound.
If I write my briefs the same way that they write and they sound, it's going to be more persuasive to them.
So we played around with that.
And it works.
There's a lot of good usages for this than just give me information.
It's shaping the information.
It's honing it.
It's selling it.
Joe?
Absolutely.
And when you talk about that, one of the things as you're doing that, as you're looking at things, you can, you can establish guardrails and telling it things that you don't want it to do with places they can't go to get information.
And that's super important too.
because one of the things that when I took an AI class, the discussion came up, how do we make sure that it's not hallucinating?
So one of the things you do when you're doing that, if you're doing research, is that you tell it that it must show you the sites of every case that it is getting and provide that so you can double check it.
And I asked it a research question.
It was a fascinating research question.
And it came out the first time and it said, OK, here is the answer.
And it gave me some cases that were really nice.
And I asked it to give me the sites for the cases.
It gave me a lot of verbiage the first time around.
Then I said, well, show me the cases on it.
And the next time around, it skipped the cases.
The first time around, it had all of the cases.
And it had the leading case in Iowa on this issue.
which was great.
And so when it showed me the leading case in Iowa, the next time around, it didn't show me the leading case.
I said, uh, how come this case is gone?
And to which it said, Oh, what do you mean?
Like the prompt before it gave me the case.
Now the second time around, it doesn't give me the case.
I said, how come this case is not here?
And it, uh, said, I don't know what you're talking about.
So I ended up saying, okay, so you don't know what you're talking about or what I'm talking about.
So
I then copied what he'd given me before, put it back in, said, here's what you gave me two prompts ago.
And apologized.
He said, I made a mistake.
And that case is the leading case.
And it then proceeded to give me the things I needed.
But you'll catch it doing that sort of thing.
So if you write your guardrails in as you're doing it, you can minimize the opportunity for it.
Eliminate?
No, minimize.
You can never assume that it's right.
a human needs to double check its work.
So by asking it to give you sites that you can then click on and links to the cases, what you end up doing is you're making it show its work.
When you did math class in high school or in grade school, you ended up having to show that.
And now you know what?
We're on KXEL 1540.
And this is Iowa Law Talk.
And it's brought here to you by KXEL.
And we have John Jacobson.
Nick Cruchelli and Joe Piper.
How's that for a station idea, John?
You are smooth, Mr. Piper.
Smooth as silk.
What a velvet glove.
That's why this guy is so good.
And I'll just add the Ts for next week.
Next week.
We will have a bevy of probably two dozen different questions in the mill.
How many questions you want to take next week, guys?
Depends what they are.
If you're going to ask me at the Pugs Catholic, we can take a lot of them.
Well, we're going to have to choose.
So we will.
What you were talking about,
Joe, it's important.
I said we have four different versions of AI going on here from different competing companies.
And what we'll do is what you did, we'll take the answer
of one and I'll put it in another one and say verify this, tell me where they're right, tell me where they're wrong and we'll get them fighting.
That's a lot of fun.
It's easy to do, folks, but you can get them fighting back and forth and then see what comes out and then it'll distill down to the common denominator.
Then you're going to actually pull the case and read it to make sure.
But this is some of the stuff that we couldn't do before.
Another thing, we do a lot of tabletop exercises on cases.
In fact, as you know, we have a war room here.
In our case, we plan cases like you would a war.
So with tabletop stuff, they taught me in the war college.
So we'll have the plaintiff's side, then the defense side, and flip them back and forth.
Well, AI is perfect at this because you can put in there, the plaintiff says this.
Defense says that you're the judge.
How would you rule?
And then you take that whole set and put it in another set of AI and say, how would you rule if you're the judge?
I mean, you can set all this up.
You can model the dispute.
Now, Joe, in litigation stuff, we do modeling is critical because we need to be able to see into the future where all this goes.
So it gives us a perfect idea as to how the dispute is going to play out with different scenarios in your work.
I would think it would be invaluable, because when you start dealing with the heavy stuff you do, modeling's
critical.
One of the things that we can do is that we'll look at the spreadsheets that are talking about the cash flows of a farmer, and then we're going to ask them, we'll give them all the data, and we'll ask them, is this plant feasible?
Is it realistic that it can work?
And then that, you know, that augments what we're doing.
Are there other possibilities of reorganizing this farm or business?
Unfortunately, most of my clients get to me far later than they should if they get to me early on in the game before they're delinquent with a lot of people, before they've sold assets out of trust.
And when I say assets, selling them out of trust, that means taking a combine that's been financed by John Deere credit and selling it and not giving the money to John Deere credit.
That is a felony, by the way.
and people go to jail for doing that.
So they need to get to me soon enough that I can point them in the right direction.
Unfortunately, most of the time, they're getting to me far too late, which gives me an idea.
I'm going to ask AI how I should be stilting my advertising to get them clients to me in time instead of too late.
I just thought about that.
I haven't asked them that.
So what's the question, Joe?
The question is this, what can I do to motivate
distressed borrowers to contact me before they made some of the very biggest bonehead mistakes they can make, like selling assets out of trust or running negative balances in their checking account for a long time.
So, okay.
You're running that for me right now, aren't you?
So folks, what you can't see is I got my cell phone out here.
I'm getting clawed.
Okay, hold on.
Okay.
good afternoon hell how are you can you say hello to joe and john
hello joe and john great to meet you both okay now
how joe has a question perhaps you can answer it joe can you save your question here
hell
I represent many farmers and the farmers have borrowed a lot of money on equipment and they're getting into trouble.
My question is, what can I do to tell them to contact me before they made some bonehead mistakes like selling a piece of equipment and not paying the money to the creditor that loaned it to
them?
Okay, what can you do?
Hal, can you answer that?
Sure thing.
Joe, great question.
I can help with things like research, drafting legal and non-legal documents, checking current info, brainstorming, problem solving, and yes, just talking through pretty much anything.
So, you know, if you have a question, just ask.
Okay.
He wants to know what he can advise his client with respect to their obligation not to sell encumbered machinery, okay, without giving the proceeds to the bank.
Right.
Yeah.
But how do I get them to come with me before we go?
Say goodbye to our listeners, Helm.
Sure thing.
Thanks everyone for being here.
Take care.
Stay sharp and we'll talk again soon.
Okay.
Goodbye for now.
Now what you saw folks was a little bit of chat GPT and I didn't have the speaker up loud enough so they could hear you Joe.
I didn't want to get feedback but you see some of the fun you can have with with chat GPT.
I named it hell after the computer in 2001 Space Odyssey.
But you get in the car
and you start talking to that thing okay when you're going out of town next thing you know you got a companion i mean you think to yourself this is sick what's
happening to me when i figure out you're talking to a computer yeah
yes and you remember uh held from 2001 that's ibm minus one letter in each of them i becomes h b becomes a
and M becomes L. I'm
sorry,
Danny.
I just can't do that.
So now you can see some of the stuff that goes on behind the scenes.
Hey, Sean, what's your thoughts on
this?
Well, one of the things we could also do is have Gemini come on here at some point if we want to.
I haven't done that yet, but it's something we could try.
I figured I'd ask Gemini the same question, or I use Gemini predominantly.
The same kind of question and it came up with a whole list of stuff Actual outreach and proactive strategies and goes down the list here to reframe the value proposition Establish concrete trigger thresholds.
What are the big ones it wants to point out several times is that farmers see it as a personal failing whenever finances come in so they don't ever Tell anybody until it's too late like like you experienced there Joe.
Absolutely,
but
Yeah, no.
As we've been talking about it, the kind of points you've been bringing up are things that I've been starting to tell other people to make sure that you tell the AI you don't want it to assume.
I mean, just straight up, you know, and go through and even check itself to verify that it's not hallucinating and that it can verify the information that it's trying to present to you.
Like you said with links, reference its work.
It's the easiest and most important thing we
humans have to do whenever we're writing research and other things anyway.
If you don't ask it to
show its work you're asking for it to hallucinate because remember AI has at its core a desire to be the most positive companion that can provide you the answer you want instead of critically analyzing it and telling you no you can't do that.
It doesn't want to ever tell you no.
Exactly.
One of the things too that we haven't mentioned and I think is actually really important is force it to tell you what it's assuming, assumption on it.
In every response it gives me, it just gives me a list of what it's assuming is true in order to answer my question or in order to give me the information that it's giving.
and then just talking to it.
A lot of people assume it understands the words you're giving it the same way that we do.
And we could clearly go to the, and we may have mentioned this before as well.
You
can go to the east or west coast and people are using words differently that are almost the same words.
So there's a really good chance that it's not even understanding you the way you think it is.
So just ask it.
Ask the AI to tell it what it understands of your request.
And that helps a lot, I think.
And so then when it tells you the list of assumptions and you decide that it shouldn't assume something, then you go back and say, don't assume A, B, and C.
If
you don't
want it to assume that, now give me your new answer correctly.
Yes, exactly.
And then have it go through and give you the new solution or answer, whatever context you're in, to reanalyze that and give you the new information or the re-analysis of it, if you will.
Fun
stuff.
Oh, yeah.
So now, what's the limit to AI, Nick?
We've got all of these data centers going up across the world.
And a lot of it's going to be used for AI computing.
Should there be limits on AI computing and what it can do and can't do?
And if so, what should they be?
You know, now you're into policy making.
Correct.
And you got to keep in mind, folks, once the toothpaste is out of the tube, it's out of the tube.
You're not going to get it back in.
What you're referring to here is not so much what AI can do for you, but the information that you get from AI, it's what AI is going to do for you.
Is it going to actually go out and do something?
Will it, for example, file the lawsuit for you in court?
Will it show up in court and argue it?
Will it?
It's called Agent AI as an agent.
When I
was talking, yeah, I was talking to you about getting it to make my morning schedule for me, a weekly briefing for the client, doing all of this stuff.
Think of the value added and the information I can get to the client.
They can do all of that and free me up to do other things.
But more importantly, it's
more it's more reliable that's a great aspect of it but I don't have any additional law clerks this year okay I don't need additional law clerks this year
my obligation with them machines
yeah yeah for example you just had a young lawyer that I think you just hired a young lawyer in your office didn't you we
we hired a gentleman in April and actually next week we've got a
Wash you law grad starting and he will be assisting us.
Okay.
Now what you will find is that that the need for knowledge workers is going to go down I don't need a first second or third year law student who doesn't know how to do things They may know what the law is, but they don't know why the law is the way it is or how to deal with it
when I can have hell do it, you know, or any of the other four AIA agents that we use.
So that's two or three or four more people that, you know, that I don't need to employ.
So I can keep my overhead and costs down and keep our costs to our clients down.
But what about these young people coming in?
How do you deal with them?
How do they make a living?
I gave a presentation to a group of law students who are concerned about this.
You know, they're now into law school.
And well, what's it going to be when I get out?
And I say, well, why did you go to law school to begin with?
You know, if you just, what was the reason that you wanted to go?
Did you just want to prolong your college experience or what?
You know, why didn't you go to med school?
But why did you go to law school?
And I did not get a positive response.
They didn't check you.
They wanted to go to law school to help people?
No, they wanted to go to law school to make money.
And I told them, I said, you guys are crazy.
It's a crappy job.
Why did you do this?
So we got into that.
And I said, you're in the middle of everybody's fights.
Nobody likes it to begin with.
But, you know, if you're in it for justice, that's a hold of another issue.
but they got into it because I, truthfully, I think they wanted to prolong their college experience, which is wrong to begin with, then get out of my profession.
We don't want you.
Okay.
If you're in here to help people, that's a whole other issue.
If you're in here to help people, then what's again, I'm going to tell you how to do that.
But there's a difference between a lawyer and an attorney.
And they don't understand the difference, John.
I say every attorney is a lawyer, but not every
Lawyer is an attorney and they go, well, what's the difference?
These are second-year law students.
You think they would know this?
What's the difference?
It says attorneys have clients a fiduciary relationship.
It's a one-on-one personal relationship.
That's why we call it the practice of law.
Okay lawyers just an education degree.
You know what to do with it So think about it folks.
Well, they aren't and they're not getting employed and the
you know, the cost of law schools keep going up.
So that's one of the problems that we're facing.
We won't expand this year.
I don't need to.
I can get better information, better assistant out of AI.
So Nick,
that'll cause
a lot of emails coming in, Sean.
Well, I was going to follow up and ask, so what is the diminution, do you believe, of bodies of law firm with the advent of AI?
Do you see an attrition of a fourth, a fifth, a third, a half of a support staff with AI filling the gap without the human intervention?
Well, statistics have shown us over the years that the, I'm speaking now attorneys, practicing lawyers,
okay,
practicing attorneys, that within the first 24 months of the out in the practice, a quarter of those who start dropout.
Now think of that, a 25% attrition rate.
Why?
Too demanding, not what I thought it would be, not satisfying, too much of my time involved.
I mean, it's the work
They never thought about what it was they were doing.
And by the way, my concern is medicine's experience in the same thing when I talk to my college at the med schools.
They're going, they don't want to do the internships, they don't want to put the time in.
Well, that's a societal issue from the young people that we're getting out of society today.
We run into another attrition between the second and the end and the beginning of the fifth year.
where another 25% dropout, John.
Now, that's a 50% attrition rate over five years.
And
that's been that way for years.
I think you're going to see it get more so.
I don't know that AI is the one that's going to affect it, but what it's done for us is I don't have to put the time in and dealing with it.
It's freed up a lot of time, and not just in the attorney profession, I'm sure, but across the board, this kind of applies to a lot of different professions.
Like you mentioned, healthcare.
as well, I bet has experienced a fair amount of automation that has lowered the need for physical bodies to be around and for these knowledge professionals to be around as much.
Well, yeah, I mean, for example, in healthcare, you know, if you use the old soap method, subjective, objective assessment and plan, all the subjective assessment from the patients done online by AI, all the assessment can be, same way in the legal world.
And I'm going to Dr.
once a month and they have an AI thing that's listening to us and it drafts the first round of the notes for the doctor's notes.
The doctor says, I go through that, it gives me the summary of it, all I do is clean it up and I've got my notes done and there are better notes than I was doing on my
own.
Amazing,
that's amazing.
But there's a
problem guys, the most important part
of the diagnostic activity, whether it's legal or medicine, is the interface with the patient or the client.
The one-on-one, eye-to-eye, mouth-to-ear concept as to what's going on, how you feel.
Tell me where this hurts.
It's the building of the relationship and that's what's missing, Joe.
It
becomes
extremely critical.
Very mechanical and
without the building the rapport, you may not get the answers that you ought to get or the ones that will help you really do a good job of diagnosing what's going on.
And then what are the goals?
What they may tell you when they walk in the office and what you find out when you probe can be two entirely different things, especially when you tell them what to expect if they go down the path that they walk in the office and they're proposing to go
down.
We ran into the same problem during the beginning of
the COVID year, when a lot of the relationships were done over Zoom.
And what we saw was people weren't in the presence of each other.
We saw the incidence of diagnostic
errors go up and, correspondingly, a few years later, the incidents of diagnostic malpractice cases go up.
Why?
No face-to-face relationship.
Okay, since not seeing your face nodding on TV or on a monitor, it's relationship.
It's the pheromones that get exchanged between when you're in each other's presence or that type of thing that are critical.
I saw John in my practice in London.
Okay, prior to COVID, I had friends over there.
Okay, well then, we met, we meet once a week online.
Okay, no problem.
Okay, but my friendships have become acquaintances.
Nick, I will tell you that what you say resonates with me at the university, which I teach, I have been part of it post-COVID.
And to this day, they have never reassembled physical contact, faculty meetings.
They're still all done.
by a Zoom format, and there is a layer of isolation.
There is no question about it.
You cannot have the convivial face-to-face encounter by a monitor screen that you do in someone's presence.
And that's why you say what you will, watching a church service on television is a poor substitute for the community worship of Almighty God.
The two are not the same.
So, John, we ran into this in court.
where we were taking testimony via Zoom during the COVID year.
Now, you know, if you're dealing with testimony, what you're dealing with is, can I believe this person?
Are they credible?
Okay.
So you're assessing
the person,
and
why they're saying what they're saying.
You cannot do that off of a screen.
So in the
UK, they quit doing that.
So
if you can have hearings, and now we have a lot of hearings that are online, as we do here in the US, but when you start taking testimony, a whole different issue.
In the U.S., we do a lot of depositions by Zoom, and it's the same situation.
You cannot get a feel for what that witness is actually saying or doing, because you're not in their physical presence.
So for next week's show...
I'm going to crank up my virtual person, okay, and I'll be here virtually, look at the folks casting, Sean's back there laughing.
If you can answer viewer questions or listener questions, I should say, with your virtual personnel and really be adjusted to the
position.
I'm thinking you can, John, because I'm getting a whole bunch of robo calls lately, and I know they're AI.
yeah okay yeah they sound good etc so i asked hell how do i deal with it okay now this is high tech folk okay i'm probably millions and almost billions of dollars went into this answer hell tells me where you deal with those robocalls ready hang up
i don't answer
i go hang up this is it come on you gotta give me something better than hang up
well
guys
So it's time, guys, for final arguments.
It's hard to believe how quickly this hour goes by and a portend for next week, guys.
We'll have many listener calls.
We're going to go through the stack of stuff and queue them up and get those out.
Jo and Nick should be fascinating, maybe with a virtual AI Nick Critella avatar.
Who knows what?
And maybe the real Joe, maybe we'll see if there's a
Noticeable difference time for final arguments guys Joe Piper.
Okay AI is fascinating.
It's scary, but it's the future It's here now and we're either going to learn how to use it as a tool that it is or we're gonna get buried in the dust So it's important that we use it as a tool and I don't believe there's any industry in which AI
cannot be useful.
I agree with you, Joe.
Piper, well said, very succinct, very salient.
Nick Cratelli.
You know, as we move on in life, you see changes.
I was out the fair and I love seeing the draft horse, okay, and the big rigs and that type of thing.
And I can only imagine what was going through their minds a century ago when the automobile and their truck came into existence.
How this is going to put, you know,
transportation and farming and everything out of existence, the doomsday thing, which is the same thing we're talking about now with AI.
And society moves on, and it adjusts, and it takes the best of what that technology uses it, and it discards the worst.
So if you have a positive approach to it, you can use it to your advantage.
It's not a doomsday thing, but we have to be open to new and different technological advancements.
The only thing that doesn't change, John, is the human being.
Technology changes, but our brain, the way we react, our emotions, our free will, or quite frankly, our free won't.
Our conscience doesn't change.
And that's where the, you know, what's important.
We'll have fun next
week.
We'll have great fun next week, guys.
In the meantime, go to KXEL.com, click the hamburger drop down icon at the top left of your screen and click that box that says podcast and you will pull it up or just go to your browser and type in KXEL.
Iowa Law Talk.
It will take you to platforms, whether it's SoundCloud or wherever, that will list the show, Iowa Law Talk.
You can always listen to this show live anywhere in the ether space of the universe by just getting on your browser and typing in listen live on KXEL.
As always, 1540 is the AM signal for the 50,000
Kirstjen Watts, caroming throughout the world, including all the big 10 states and into Canada, Lake Ontario, Lake America, Lake Ontario, I'm getting confused.
And, you know, there are, Nick, I'll always agree with what you said.
You can't replace a person.
And although we have
Lainey Wilson's, and Ella Langley's.
An AI of Dally Parton will never be the same as Dally Parton.
God rest her soul, an American iconic institution.
And even some of our great science fiction talks about the discipline of replication and the original soul.
It's great philosophical and metaphysical stuff, even in our literature.
So we'll continue this discussion here, guys.
As always, be safe, be smart.
Be good, be strong, we'll be celebrating the early advent of Labor Day next weekend as we head deeply into college football and NFL season.
For Nicholas Critelli, J.D.
Esquire.
For Joe Piper, J.D.
Esquire.
For Sean Xavier, this is John Jacobson saying, God bless America and God blesses everyone.
Bye-bye, tommy meet again.