Latest / Coopers' Code / Is AI Making Lawyers Lazy?: This Month in Legal AI

Transcript

Is AI Making Lawyers Lazy?: This Month in Legal AI · Coopers' Code

0:00
Welcome to Cooper's Code. We are doing the AI overview in
0:04
terms of AI and the law. It's shows up about every month
0:09
or so and I'm thrilled to be here with Co host Sherrod Milan
0:13
Farr and Marshall Cole. Sherrod's up from the Bay Area
0:16
as people know, plaintiffs personal injury lawyer, a reform
0:21
defense lawyer. Marshall Cole is from the LA
0:23
area and he's the the kind of lawyer that most lawyers want to
0:29
make sure they never never need esteemed legal Mal defense
0:33
lawyer. So all of us, though, have a a
0:35
vested interest in technology and use of AI and, and keep
0:40
current and want to make sure that everyone out there has the
0:42
opportunity to keep current as well.
0:44
So Sharon, I might kick it to you.
0:47
I know we've got some bullet points to cover, but the
0:52
hallucinations in everything from proper loss to proper win
0:57
to mega firm taking it on the chin and where do we go from
1:02
here? So it seems like there's a range
1:05
of issues that have occurred that kind of change the
1:08
perspective of this kind of mistake is being made by, you
1:11
know, solos or improper plaintiffs.
1:15
So we have a case in Alameda County where the judge dismissed
1:18
the case when the proper plaintiffs and then for those of
1:21
you who are not lawyers are people who are representing
1:24
themselves, submitted fraudulent evidence to the court and the
1:28
court basically dismissed their case because of that.
1:32
We also have a very esteemed firm, Boise Schiller, that
1:37
recently made some filings that at least one filing that was
1:43
containing hallucinations and they're in hot water because of
1:47
that. So now there seems to be a range
1:50
of people who are using AI for legal tasks and are not
1:56
necessarily doing the betting. But on the other hand, I
2:00
recently saw a story about an improper who beat her landlord
2:04
in a landlord tenant case, likely an eviction case using
2:08
AI. And that was kind of a
2:10
fascinating approach that she had used to minimize potential
2:13
hallucinations when she was filing documents with the court.
2:16
So it's kind of few things to throw around.
2:19
That last one was the one I think that that might deserve
2:23
the most unpacking. I, I think that the take away
2:25
from, from my perspective on the hallucination side is if
2:30
everyone is getting tagged by it, it's a it, it means that we
2:35
need to do our best on quality control.
2:39
And if a firm like Boy Schiller is, is getting tagged, that
2:44
tells me that I need to make sure that if somebody's giving a
2:47
brief to me that I've taken a look at it.
2:49
It that if there's a case, it seems like it might be a little
2:52
bit too good for us that it gets vetted.
2:54
But Sherrod, I, I thought what this proper did was, was
2:59
brilliant. Do you want to?
2:59
Do you want to share that? Yeah.
3:01
And going to your point, it doesn't seem to be an issue of
3:06
resources, right? Because if this mega firm that
3:09
has all these esteemed lawyers in it falls into the same trap,
3:13
It's an issue of our people just stressed and just wanting to get
3:18
something off their desk or they don't have time to vet it.
3:20
And that's where I think the pitfall is, is our human nature
3:25
to not necessarily pay as close as attention as we should.
3:28
And I think everybody has the potential for that mistake that
3:32
we just need to be hyper vigilant about.
3:35
With somebody with the depth that Marshall has, when a lawyer
3:38
comes to you and says, you know, I'm worried about my, my firm
3:43
and, and my exposure. How do I have best, best
3:48
practices to reduce my risk for AI use?
3:51
Because I don't know if a paralegal is using it.
3:52
And, and I do try and do my best for checking things.
3:56
That's one and and two is. If I do end up making mistake,
4:00
what should I do to fix it? Did you have any recommendations
4:04
in that regard, Marshall? Yeah, on the first one, when it
4:07
comes up all the time, I say you should have a policy in place
4:11
that that creates boundaries for your staff as to when they can
4:15
use it, how they can use it, etcetera.
4:17
And and if they are doing it, they should provide training and
4:21
force people to read these articles because it seems like
4:25
since we've been doing this each month or so, there's a new firm
4:28
that this happens to every month it's a new one.
4:32
And I agree, it's kind of now run the gamut from, you know,
4:36
mill Pi solo to, well, reputable, you know, very
4:40
reputable firms. And I think that it's, you know,
4:46
and as as Sherrod was talking, I was wondering, you know, when
4:49
the lawyer who did that in the Bo's Schiller matter, you know,
4:55
was it just sheer laziness, like really trying to get something
4:58
off their desk? Because if you're actually doing
5:02
it right and trying to get correct and accurate case law, I
5:06
wonder how much time you're actually saving than just doing
5:10
those searches through W law, which is still really efficient
5:13
legal research. And you know, sure, you can
5:17
definitely get cases quick, but if they're not accurate, then
5:21
forget the risk component. But you're also just not
5:23
actually saving any time. You, you know, you're just you,
5:27
you've created something that's fake.
5:28
So that's worthless. But if you're going to do it
5:31
right, it might have taken more time, right?
5:33
You might have to go and vet the cases.
5:35
And also, if you have to go and vet the cases anyways, again,
5:39
are you saving time versus just doing it properly through,
5:44
through Westlaw or Lexus or whatever it might be?
5:47
So I, I, I think those are the, the interesting parts of what's
5:52
going on here. And people I don't think are
5:54
thinking thoroughly like as is this really beneficial to me?
5:59
Sherrod will come back to the the the pro probe.
6:01
He's haven't dropped that line, but this reminded me of there's
6:07
a personal injury firm down in Los Angeles does very high end
6:11
work and when I was a young lawyer I had heard that they had
6:15
in essence a a New Yorker style fact checker.
6:18
So they have an editor whose sole job is before any meet and
6:24
confer any brief, anything goes out the door, that person has to
6:28
go through it and review it and any citation in it gets QCD.
6:33
It seems like a lot of effort to have one person on your staff
6:37
who's just doing that. But that to me sounded like a, a
6:40
great threshold. And it is something we we use a
6:43
a variant of that where if somebody writes A brief,
6:45
somebody else goes through and tries to poke holes in it to, to
6:48
try and find something wrong with it before it goes out the
6:49
door. Yeah, just as a side note, I was
6:52
talking to a friend of mine who told me that one of our mentors
6:55
had a job with a very prominent defense firm.
6:59
And his job was when they needed to cite to an article about
7:03
something, he would write that article and then they would cite
7:06
to it. So.
7:10
Go to the flip side. Of that.
7:11
Coin, I like that. Yeah, that's a good one.
7:16
I buried the lead a little bit, Sharad, in terms of the the pro
7:18
per who's using a tool that might put us all out of business
7:22
if if somebody keeps doing the anti hallucination method.
7:25
So this was on a news video that I saw.
7:30
Basically this lady supposedly beat her landlord, I'm assuming
7:35
in an eviction case and she did it by using AI.
7:41
In this way, she would go to 1 system, put in her prompt, get
7:46
the response back, and then would take it to go to the
7:49
second system and enter the response and have it vet the
7:54
response and then take that response to a third one and have
7:58
it vetted by the third one. So that, it seemed to me
8:01
significantly reduced the chance of hallucination and made it so
8:06
that whatever she was submitting passed muster and she apparently
8:10
won the case, according to the news report.
8:12
It's an interesting approach, Marshall.
8:14
I think that approach is great for someone who doesn't have a
8:17
law license, right? They don't have a way to know
8:21
cuz cuz they're the only other thing they could do is Google
8:23
it, right? So if they're just Googling
8:25
these cases, you know, presumably they exist, but maybe
8:28
they don't at this point, you know, and they don't know one
8:31
way or the other. So she's still taking a
8:33
calculated risk regardless of just having one system bet the
8:37
other and then another vet those.
8:39
But you know, she did it to a level where she's comfortable.
8:41
And also there's the repercussions if it there were
8:46
hallucinations are much different for her than there
8:48
would be for any of us or any licensed lawyers out there.
8:53
And then lastly, it kind of goes back to the point I made
8:56
earlier, which is if that's what you have to do in order to make
9:00
sure you're not getting hallucinations, you probably
9:04
could just do the research in the same amount of time from,
9:08
you know, just a legal research perspective and ensure that
9:11
you're getting accurate information.
9:13
I mean, I assume she was having a draft entire briefs for her,
9:16
which is great. And I think it does provide a
9:20
huge benefit to people who don't have access to a lawyer, who
9:24
can't afford a lawyer from what they would have been filing
9:27
before to what they can file now, excluding case law.
9:30
Just just draft me an argument, you know, on these subjects or
9:35
on this issue or here's my 5 facts.
9:37
I'm going to do a phenomenal job for those people and I do think
9:40
that they're going to have, they're going to see better
9:43
results against these types of matters, you know, landlord,
9:47
tenant matters, small neighbor disputes, whatever it might be.
9:51
I think they're going to be very well off.
9:55
I'm bouncing around with a, a very unsmooth Segway to do a, a
10:01
different topic. One of the notes I have in terms
10:03
of conversations that we had was something related to frontline
10:06
managed services. And I, I have a note, Sherrod,
10:09
that this may have been a entity that you mentioned at one point,
10:15
an AI tool that produces billables by sitting in and
10:18
monitoring all e-mail and workflow.
10:20
Does that ring any bells to you? I don't know if that's the
10:22
company. I think the company I saw, it
10:26
was something like mean time or something like that, and they
10:29
log into every system that you have and then they can monitor
10:35
every second of your life while you're working.
10:38
And I haven't tried them. I don't know anything beyond
10:42
that. But it seems like they were
10:45
pitching to people who do hourly work, saying we'll make it a lot
10:48
easier for you to keep track of your time by automatically
10:53
essentially creating these time entries for you.
10:56
Marshall, have you come across anything because I know you do
10:59
some time billing, Yeah. I do and I was reading about it
11:03
and we don't use it and I, I would not be going to rush to
11:08
sign up and pay money A to allow this company just unfettered
11:12
access to my firm's e-mail system.
11:15
And I think any of these, when you start allowing that type of
11:17
access, you got to do your first question is security.
11:21
But, but beyond that, I think it presents an issue for me as far
11:25
as the overbilling, right? I mean, just 'cause I get an
11:30
e-mail doesn't mean I bill a .1 or a .2 or a point, whatever.
11:35
You know, it may not warrant a time entry.
11:40
So for it to just create entries because it sees emails and you
11:46
know how long it would create that time entry for, it seems
11:51
really suspect as well because some people read faster than
11:55
other people. I mean, we have to build the
11:57
amount of time that we spend on the particular task.
12:01
And so those are my issues with that type of, you know,
12:06
advertisement and allowing it to do that.
12:09
Sure, I, it probably could go through my emails and create
12:13
point ones all over the place, but I don't think that's how I
12:16
bill my time. Most of those emails don't get
12:19
billed. Well, I recognize that the
12:22
majority of our listeners tend to be more on the contingent fee
12:25
side and so time billing is less interesting.
12:29
I do know that some contingent fee firms keep track of ours to
12:34
help them understand how much time they put into a matter,
12:38
what cases create value, what cases don't create value as as
12:41
they go forward. And so I know that when people
12:44
are looking at how to be efficient on that, efficiency is
12:48
important. But as you point out, Marshall,
12:50
so is security. We keep time because some of our
12:54
cases are cases where we can file attorneys fees motion on
12:58
right. So it's important to have
12:59
accurate time and be able to do that.
13:02
But again, it every phone call is not something that is billed
13:07
and every e-mail isn't billed. We try to put our time entries
13:11
in, in a way when if a judge looks at it, they're looking at
13:15
it that this is billing for serious work that's being done
13:17
to advance the case to the point that it got to as opposed to
13:23
just, hey, we're nickel and diming this with every .1 entry
13:27
that that's conceivable. And presumably you could create.
13:31
I'm sure they can set it up to try to be as efficient as
13:35
possible, but it's one of those things, and maybe this is a
13:39
little bit naive to say from an AI perspective, but it just
13:43
seems to me it's one of those things that requires a personal
13:46
touch. You know, I read my emails
13:49
different than my partner right next door reads his emails or my
13:53
associate reads her emails and they all are expected to bill
13:57
based upon that. And so it's so personal and it
14:01
and it requires an exercise of judgement whether I'm going to
14:05
bill it or not that that I don't think can be captured through an
14:10
AI platform. That's how I look at that one.
14:13
That makes sense. The, the security piece brings
14:16
me to a, a separate but related, and I don't know that I'd
14:20
necessarily categorize it as security, but the, the idea of,
14:24
of having data about one's clients accessed by other
14:29
sources and without using names of, of entities.
14:33
I, I know that it has come up. And I believe, Sharad, you
14:36
brought this to our attention that some of the companies who
14:40
are offering services out there, say, for demand packages or
14:43
other places are both doing a good job of providing the demand
14:47
package drafting, but also offering a kind of a back end
14:50
where they're gathering data on what the resolutions are with
14:54
the expectation that at some point they're building some sort
14:56
of Colossus variant. Do you have any more insight on
15:00
that, Sharon? Well, I actually did speak with
15:03
an executive at one such company who is now told me that they are
15:07
discontinuing or they have discontinued that practice
15:11
because it wasn't really getting them the data that they need it.
15:14
So I don't know what that means, but as of right now that company
15:18
is no longer offering that incentive of tell us what your
15:22
case settled for and we'll you know, give you some additional
15:25
credits toward our our service. That I think answers the
15:29
question, but I, I don't know that long term that it, it
15:32
necessarily closes that, that arena.
15:35
And, and, and I think we've talked about this a little bit
15:37
in the past, Colossus being the database that, that insurance
15:41
companies use to tell you that your case is worth $1347.62
15:46
based on similar incidents across the, the country.
15:51
The more granular this data gets and the more AI can do data
15:59
analysis and and data science, the more it seems like we're
16:03
going to start getting big data feedback on valuation on cases.
16:11
That's an interesting concept because there are now firms out
16:17
there that are doing big data analysis on jury results.
16:22
And there's apparently research that was done that indicates
16:26
that AI is actually notoriously bad at predicting the results.
16:33
And one of the researchers who was giving a presentation at a
16:37
conference that I went to indicated that there was one of
16:40
the AI models initially was fed the information and said this
16:44
case is worth $250 million. And then they said, well, can
16:49
you tell us how you got to that number?
16:51
And the response came back, oh, thank you for pointing that out.
16:55
I re evaluated it in the case is worth 1.2 million.
17:00
So that's a huge variance. And he used that as a very
17:04
extreme example of how it got it wrong.
17:07
But for the most part they were saying that the the AI itself is
17:12
really poor at determining what the case value is.
17:16
And quite frankly, you know, we've talked about this, Miles.
17:19
I think the idea that somebody else got a $100,000 verdict on
17:24
the same case and that somehow drives my value, your Marshall's
17:29
value, is absurd because there's too many variables to say yes,
17:34
this case is instructive in terms of what the result is
17:37
going to be. And in the instances that I've
17:39
actually done it, I've had some trials come up and I've just
17:42
been curious and using it as an opportunity to sort of try and
17:46
learn, hey, what did you be likely to do in these
17:49
circumstances? And without coaching it, it's
17:53
really bad. But the second you start to say,
17:55
well, a verdict of this number or a verdict of this number or
17:59
what about these defenses, it's like a golden retriever.
18:02
It then just starts to want to please you and it and it
18:05
understands where you're going. And so then you're just getting
18:08
kind of what you want. But I don't think it's accurate
18:12
at all. And so, you know, 'cause it'll
18:15
be like, Oh yes, I could see these defenses being really
18:17
strong after I've pointed those defenses out.
18:22
And then I can ask for the weaknesses of those defenses and
18:24
then it will point out the weaknesses of those defenses.
18:27
But it didn't do it beforehand. And so, you know, sort sort of
18:32
like just telling me things I already know and wasn't very
18:35
useful in helping me try and learn something new or come at
18:40
something from a different direction, like as if I would be
18:43
dealing it with the jury. And I think a lot of that is
18:47
just like trial lawyer experience and that true human
18:51
connection that you try to make with jurors.
18:54
To your point, you're odd that like just cause someone got
18:56
100,000 in some jurisdiction doesn't mean on the same facts
19:01
same case could be like everything identical.
19:03
You bring a totally different flavor to your case.
19:06
With 12 different people you can get a completely different
19:10
result. Yeah, I just get tired of
19:12
hearing from, you know, opposing counsel.
19:14
You're not Miles Cooper or Marshall Cole when you're trying
19:16
cases, so. A variant I think of, of, of
19:21
what we're talking about. At one point our firm tried
19:24
using a Monte Carlo simulator for analysis for our portfolio
19:29
for the bank. And, and I think most people
19:33
know this, but on the contingent fee side, if you're using a line
19:35
of credit, your bank wants to know what the value of your
19:37
cases are. And so, you know, usually once a
19:40
quarter, once every six months, you provide an estimate to the
19:44
Bank of these are the X number of cases that I have and by name
19:48
here they are. And this is when I think it's
19:50
going to resolve and this is how much I think it's going to
19:52
resolve for in terms of the net fee.
19:54
And when we a Monte Carlo simulator was it was a very
19:58
fancy spreadsheet that was pulled together.
20:00
They kind of gave a high end and a low end on what those
20:03
estimates would be run through multiple times and, and what
20:06
have you. And our our take away after
20:10
having played with it for about a year and a half was nothing
20:15
beat our. Whether you call it gut
20:18
instinct, limbic system, you know, 25 years of, of seeing how
20:22
cases turn out, we were always pretty much spot on on
20:27
valuation. We were always about six months
20:30
to a year off on how long it was going to take.
20:33
We knew those, those, those, those challenges.
20:35
So we, we had, we adjusted for them.
20:38
And my take away from this was after doing this for as long as
20:43
we've done it, we seem to be better than machines on the gut
20:48
instinct valuation. And I don't see anything so far
20:52
that tells me that that's going to change.
20:56
And, and I think I'm, I'm just saying a variant of what what
20:59
you Marshall and and and trod both said.
21:01
I think you're absolutely right on that because there's too many
21:05
variables, right? You change a few jurors in that
21:08
same trial and you may have a different outcome.
21:11
And we know that from talking to our jurors at the end of our
21:14
trials. The one thing I found that it's
21:17
interesting to do with the AIS, our practice management system
21:21
just had a new release and roll out of an upgraded AI system and
21:25
it's a significant upgrade. So I have a case going to a
21:29
mandatory settlement conference. So I took both our statement and
21:33
theirs that I put it in and I said, all right, review this as
21:38
an expert, do an analysis of the strengths and weaknesses of both
21:43
sides. And we have what are what can
21:46
each side do to like bolster their arguments and that sort of
21:49
thing. And he actually did a pretty
21:51
good job of giving me an honest and balanced analysis of the
21:55
weaknesses in our case and weaknesses in their case.
21:57
And at the end of it, it said, you know, which maybe was the
22:01
part of the people pleaser piece of it.
22:03
It said the defense seems to be underestimating the value of the
22:07
OR the severity of the damages that or the injuries that your
22:11
client sustained. So it was an interesting piece
22:15
of feedback. But in my opinion, it's only a
22:18
data point, right? Because to me, I want to talk to
22:22
other human beings who look at my case and say, well, OK, So
22:26
what? Or Oh my God, that's horrible.
22:28
How can this person be put through that?
22:30
Those are more instructive, in my opinion, than having a
22:34
machine kind of give me what I I may want to hear.
22:38
Sometimes when you are in cases and you keep hearing from
22:42
defence counsel and adjusters about how terrible your case is
22:45
or how they're going to win or how they've had eight other
22:47
cases like this and they defence them all you, you start feeling
22:51
a little bit like, oh, well, I must, I must be seeing this
22:54
wrong. And, and the answer is no.
22:57
If, if you, if you're in the game, if you're, if you're
23:00
working with your colleagues, if you're spitballing the case and,
23:03
and round tabling it with friends and everyone's saying
23:05
like, no, they're missing something, that each, each case
23:09
is unique, that these these are, these are humans going through
23:11
human experiences. Fortunately, defense counsel
23:15
will be able to put that in writing in that same case,
23:17
saying that he was going to win and quote, see you at trial.
23:21
So I, I, I really thanked him for putting that in writing and
23:24
guaranteeing the win on an open policy case.
23:27
Yeah, yeah, yeah. I, I think it's just, it's
23:30
important outside of the AI side of things on the, the trial
23:34
where the, the, the litigator side of things is to have faith
23:37
in one's ability to, to analyze the case and move forward on it,
23:42
even when you know the defense is, is telling you how they're
23:46
going to crush you eight ways from Sunday.
23:47
And if you don't have an idea of that and and haven't learned it
23:52
on your own and are relying upon AI for that, I think like get
23:56
ready to lose or get ready to have a number that you're not
24:00
going to be happy with because you without the ability to
24:04
independently analyze it, take it as a data point, like you
24:08
said, and, you know, consider all the other factors, including
24:12
who the lawyer is on the other side making that, you know,
24:16
chest pounding statement. I mean, it could be the worst
24:19
lawyer ever that you're not worried about it or the best and
24:22
say, well, he, you know, he's going to do a really good job or
24:25
she's going to do amazing. So whatever platform you're put
24:28
it isn't going to know or understand those.
24:31
That's an excellent point because if the AI or one data
24:36
point is a determining factor in how you value your case, maybe
24:39
you need to refer that case to the lady who beat her landlord
24:43
as a pro per. I mean, you're basically taking
24:48
the human and your experiences as a human being out of the
24:52
equation, which to me is the worst possible approach in
24:56
evaluating and advising your clients.
24:59
That's what I really like about what we do versus other
25:03
industries, at least it from litigation, right?
25:06
We're in court different than transactional lawyers and things
25:09
like that. And it's certainly different
25:11
than people at finance where it's very data-driven.
25:14
There's a lot of objectivity to it.
25:16
It's black and it's white. I mean, we, the three of us deal
25:21
in so many Shades of Grey and never is something black and
25:26
white. I mean, it's just very, very
25:28
rare. And so, you know, once you
25:30
decide it's grey, right, it's just what shade of grey.
25:33
And I think that's where the skills that you develop as a
25:38
trial lawyer, if you're going to be in court or writing legal
25:41
briefs requires the understanding of all that nuance
25:46
that goes along with it, which is very, very important.
25:49
And I think in the times I use AI to try and get that type of
25:54
experience, it, it tries its best, right?
25:56
It does, you know, I get some value out of it, but it's
25:59
because I know what's valuable and what's not.
26:01
And I know how to input the data in the 1st place.
26:05
All of which is built upon those years of experience, you know,
26:09
reading judges, being in front of judges, understanding all
26:12
that nuance that it certainly can't do because it just doesn't
26:16
know those people or how they play a role in whatever we're
26:19
doing. And going back to the idea of,
26:24
of knowing your cases for younger lawyers who are
26:27
listening to this, who haven't had that experience and, and who
26:29
may be more inclined to rely on ChatGPT or or Claude.
26:35
A reminder that one of the great things about the plan of Spar is
26:39
that almost all of us are willing to help each other out.
26:43
Meaning a phone call, having some people over to lunch, doing
26:47
doing it in person and spitballing a case.
26:50
I mean I learned valuation on cases from hanging out with
26:52
great lawyers and now I feel I have a pretty good valuation.
26:57
Even with that, I'm likely to call up or get together with a
27:03
few of my friends to to spitball case if I've if I've got
27:06
questions about it. I appreciate the inference there
27:09
that the defense bar does nothing well.
27:12
It. Which is.
27:13
Search it and this I've never worked the defense side.
27:18
What I hear from the defense side is the plaintiffs bar is so
27:21
collegial and so willing to share things with each other
27:24
that you make it hard for us on the defense side because we we
27:28
don't always do that in the same way.
27:30
We don't have the listeners, we don't have the the deep social
27:34
connections in the same way. And maybe I'm wrong, Marshall,
27:37
but that's that's that is what I've heard.
27:39
No, no, it's totally accurate. And we, we try our best.
27:42
I think you, you all are a much tighter organization.
27:45
The listservs are much larger. I completely agree for younger
27:49
lawyers to take advantage of all that stuff.
27:51
We certainly have it on the defense side.
27:53
It's just not as expansive. I mean, the membership is not
27:58
there, but we, we certainly do our best and, and share what we
28:01
can. But no, I agree with you on on
28:04
your side of the aisle. It's tremendous.
28:07
I will put a little caveat out there about the listeners and
28:09
listeners are a great place to gather information.
28:13
I I've shared this story more than once, but it's worth
28:15
sharing again. They are more porous than one
28:18
might think. And so do not list your entire I
28:21
have a case with blah, blah, blah, blah, blah against so and
28:24
so and so and so. And my client is willing to
28:26
settle for X. If you do that, there's a
28:28
reasonable chance that they that might find its way into the
28:32
wrong hands. And, and I share this only
28:33
because there's probably 20 years years ago, I posted on the
28:36
list of asking for feedback on judges in San Mateo County.
28:40
The next day at trial call, I see the defence counsel flipping
28:44
through some pieces of paper and it's printouts of the emails
28:46
from our list or with with the conversation of the judges.
28:50
Yeah, yeah. A friend of his knew he was
28:52
going to trial. It was, it was no malicious
28:54
piece. A friend of his knew he was
28:55
going to trial in San Mateo and says like, oh, well, this might
28:57
help you out. So.
29:00
Well, having been on both sides of the aisle, I agree that there
29:04
are certain factors on the defense that make it a lot
29:07
harder to collaborate and help each other out, and that's a
29:11
much longer conversation. But what I found on the
29:14
plaintiff side is you're right, there's so many people willing
29:17
to help. But I also find that I can call
29:21
my friends on the defense and run stuff by them, and they call
29:25
me to run stuff by me. Just a few days ago, a friend of
29:29
mine, he has a very serious case on the defense side.
29:31
He called me up. He's like, what do you think?
29:33
And I give him my honest evaluation in terms of the
29:36
strengths and weaknesses of his case.
29:38
And those are things that the individual relationships built
29:43
for us where a computer can't simulate that sort of
29:48
interaction. And for me to call you guys and
29:51
say, hey, I got this issue. What do you think?
29:53
Am I just totally off base or do I have a reason to believe that
29:58
this is a good approach that we're taking, are hugely
30:02
valuable and for young lawyers, build those relationships, get
30:06
mentors and call us, call other people because people want to
30:11
help younger lawyers get better because that's how we end up
30:15
with the profession sustaining and being a solid profession as
30:19
opposed to everybody keeping their knowledge to themselves
30:21
and not really helping the younger generation come up.
30:25
I do think that as Marshall referenced, I think the golden
30:29
retriever earlier, the if, if you are running this by your AI
30:35
thinking partner there, there is a a substantial risk that it's
30:38
going to give you false hope because it wants to please
30:44
anything else that people think is is particularly timely that
30:48
we should cover before we wrap for this this month's session.
30:52
One, one other thing is and we, we, we touched on it earlier
30:57
maybe before recording or I don't recall, but it's sort of
31:01
the AI fatigue and you know, obviously we're hearing about
31:05
it. We're hearing about it.
31:06
We're using it, we're using it. But I think it was Miles.
31:09
You were saying that most people are not finding it actually to
31:13
be effective and either saving time or they just don't know one
31:18
way or the other and they're just using it blindly.
31:20
There's a, a study that people have been citing to that 95% of
31:24
the industries who have implemented it in some fashion
31:26
are not finding a meaningful increase in productivity.
31:29
Forgive me, I don't have the site off the top of my head.
31:32
And so that's I think something that's percolating in the back
31:35
along with the conversations about AI bubbles and and
31:40
circular investing. So, but I, I cut you off,
31:42
Marshall, I'm sorry. No, no.
31:43
That, that I, I, that was it. And I, I just think it's really
31:46
interesting and for our industry when we see at least a once a
31:50
month, almost a new firm or incident with, with people doing
31:54
something wrong. You know, how long until people
31:58
say, I'm not going to risk it, right?
32:00
I'm not going to do it for at least litigation purposes or, or
32:06
B, it's just actually not saving us any time and and it's not
32:10
worth the monetary investment. And so it'll be interesting to
32:13
watch this develop over the next couple episodes because I'm sure
32:18
it's going to just morph into a large topic just as these things
32:23
continue. Because I sort of optimistically
32:26
expected those horror stories of, of hallucinated briefs to go
32:31
down. Not like there's enough journals
32:35
out there, daily journals, newspapers, friends talking
32:38
amongst friends, listservs, right where this gets brought
32:42
up. And it's just a reminder of how
32:44
many lawyers there are out there.
32:46
But I just would have thought you would, it would have been
32:49
nipped. That would it be it?
32:51
You wouldn't see it anymore. And for us to be here at however
32:54
many episodes in continually talking about a new one is
32:57
surprising to me. I do think it, it is, while it's
33:03
wise to be mindful about the uses, I do still like using
33:09
Claude as a thinking partner. I do still like throwing things
33:12
in there, iterating on on strategies about cases.
33:17
It's sometimes there are things that get thrown up that I don't
33:22
know, I I necessarily would have thought of, but but it's it.
33:26
There is a risk of parking your brain if if you if you just rely
33:32
on AI. The AI summary of this episode
33:35
is going to say that we are down on AI because we've spent most
33:37
of our time talking about getting together with colleagues
33:40
and, and how AI is not as reliable a resource.
33:43
I do think it still has utility and and relevance and you need
33:49
to make sure that it's it's being done in a right fit way.
33:52
I think if you read the would read the summary from most of
33:54
our episodes, it would seem it would read that way, but but it
33:58
really isn't the cakes, you know, it's just sort of like
34:01
just exercise extreme caution and it can do amazing things.
34:06
Well, it's kind of like exercise, right?
34:08
I want to be able to do 50 push ups in a row, but I don't want
34:11
to do the work that leads to being able to do 50 push ups in
34:15
a row. And with AI, people are saying,
34:18
well, it's not increasing my productivity.
34:20
Well, the question is, how are you using it?
34:22
How are you investing some time in the beginning to set it up so
34:26
it actually does what you wanted to do in the way you wanted to
34:30
do it? And those are all important
34:32
things that we have to consider when we're evaluating whether or
34:35
not AI does help us or doesn't help us.
34:39
But like you said, you know, the MIT study that you pointed us to
34:43
about how that's affecting, you know, people's mentality and how
34:47
we're looking at this, This isn't new because there was
34:51
another study that was done and they found that if you have your
34:54
phone sitting in front of you on your desk versus behind you
34:59
versus in another room, you're more focused and effective when
35:04
the phone is no longer in front of you and it's in another room.
35:08
And that just tells us that our brains are wired to constantly
35:12
seek this stimulus that the phones and technology provides.
35:16
So how do we use that knowledge to help put our process in place
35:24
in a way that allows us to gain maximum productivity versus
35:28
allowing us to spin our wheels and say, oh, well, you know
35:31
what? I put this prompt in, it gave me
35:33
garbage. And I it, it's, it's terrible.
35:35
I don't want to use it anymore. Just a a a tangent on your
35:40
comment about the phones. I don't, I won't be able to
35:44
credit the original origination on this, but somebody referred
35:48
to as an adult pacifier and I thought that that was about the
35:53
most apropos thing I think I've heard in a long time.
35:56
I I'm not feeling great. I'm going to grab my phone.
35:59
All you got to do is look at the stats at how many times a day we
36:02
pick up our phones. It's insane.
36:05
Yeah, as always, I, I, I love our conversations.
36:09
I love the work put put into beforehand because it it helps
36:13
me stay current on all these things.
36:15
So thanks for joining us everyone out there, for all of
36:19
you doing justice out there. Happy hunting.
36:23
If you're enjoying what we're doing here with Cooper's code,
36:25
please help us extend our reach by giving us a five star review
36:28
on Apple Podcasts and consider sharing an episode with a
36:30
friend. Please e-mail us at podcast at
36:32
Coopers dot law with questions, comments, feedback, and
36:35
suggestions you have for who you'd like to see on the show.