Latest / Coopers' Code / Lawyers Sanctioned for Fake AI Case Citations: This Month in Legal AI

Transcript

Lawyers Sanctioned for Fake AI Case Citations: This Month in Legal AI · Coopers' Code

0:00
Welcome to Cooper's Code, the AI show.
0:03
I'm thrilled to be doing this monthly session with Shrod
0:07
Millenfar and Marshall Poll. We're going to be talking about
0:10
what's new in AI since the last time we had this episode.
0:13
So with that and, and how much further, let's go ahead and kick
0:17
it off. I wanted to turn it over to you,
0:19
Marshall, because you were the one who initially surfaced the
0:24
Sylvia Nolan V Land of the Free Second District case.
0:29
I don't think there's a citation report yet, but the docket
0:32
number is B331918 for anyone who wants to take a look.
0:35
You want to just kind of dive in and and give us a little bit of
0:38
background of of what happened here again.
0:41
Sure. So, so the the case came to my
0:44
attention, it was circulated by a professional group that I'm a
0:47
board member of the Association of Southern California Defence
0:50
Council, and I'm just going to read the quote.
0:55
The quote is outstanding because it it just kind of send the
0:59
message that people seem to still not quite learn.
1:02
Right. And then we can talk about some
1:04
craziness beyond that. But so here, here's what the
1:06
Court of Appeals said, which is as follows.
1:10
The quote, what sets this appeal apart and the reason we have
1:12
elected to publish this opinion is that nearly all of the legal
1:16
quotations in plaintiff's opening brief and many of the
1:19
quotations in plaintiff's reply brief are fabricated.
1:23
That is, the quotes plaintiff attributes to published cases do
1:26
not appear in those cases or anywhere else.
1:29
Further, many of the cases plaintiff sites do not discuss
1:33
the topics for which they are cited, and a few of these cases
1:37
do not exist at all. These fabricated legal
1:40
authorities were created by generative artificial
1:43
intelligence tools that plaintiff's counsel used to
1:46
draft his appellate briefs. The AI tools create fake legal
1:50
authority, sometimes referred to as AI hallucinations, that were
1:54
undetected by plaintiff's counsel because he did not read
1:57
the cases the AI tools cited. And then what the Court of
2:02
Appeal did was it sanctioned $10,000 to be paid by the
2:07
attorneys to the court. And the court did not impose the
2:10
sanctions which would have been paid to opposing counsel or the
2:13
opposing party's fees because the declaration was not filing
2:17
file detailing those. But what it also did was it
2:19
directed the clerk of the court to send the published decision
2:24
to the State Bar for reporting. And I assume that those lawyers
2:29
are in the process of getting into a fair amount of trouble.
2:33
And so for me, what's just unbelievable is this court took
2:38
an opinion that by all accounts would never have been published.
2:41
There was nothing new about it, not a new issue of law, nothing
2:46
undecided. They just published an opinion
2:50
to show what this person did and how bad it is.
2:55
And so now to my knowledge, this is the first published opinion
3:00
where you've got a court basically chastising and
3:03
sanctioning council for using fabricated and hallucinated AI
3:10
cases, which obviously is something we've talked about
3:13
here and what we were talking about a little offline.
3:17
To me, it's just astonishing, despite all the articles that
3:22
exist online and in print of lawyers doing this right,
3:26
falling into this trap where perhaps earlier they may have
3:32
been able to claim some sort of innocence by way of naivete, now
3:37
it's everywhere. But yet these lawyers are still
3:39
doing it. And I in in an arbitration I had
3:44
our opposing counsel did it and he got burned.
3:46
And I just find it fascinating that that the word has not
3:49
reached all these lawyers to just not do this type of thing.
3:54
It brings up something for me and and that is I am mindful
3:58
that the people who we tend to associate with are people who
4:07
tend to stay current. What I mean by that is, you
4:10
know, Marshall, I met you at an ATC presentation, Gerard, we got
4:17
each other the the greater legal community.
4:20
I think actually I may have seen you presenting at what's new
4:23
important trial. We were both attending, I think,
4:26
but yes, that's where we met and that's what exactly what your
4:30
point is. Exactly.
4:32
So I think that there are a lot of lawyers, the type of lawyers
4:36
who who listen to a podcast like this who stay current.
4:39
I also think that there are a large number of lawyers who
4:45
unfortunately probably become your clients at some point
4:47
Marshall for the legal Mal defense or or State Bar issues
4:52
that they're facing who don't do these kinds of things.
4:57
And I've come across lawyers who don't even have an axe have a W
5:02
law or any sort of whether it's Lexus W law or one of the other
5:07
also Rams don't have access to that.
5:11
And so while it is, I think, surprising at an appellate level
5:16
for someone who could get so checked like this, I think we'll
5:20
see more of it. Actually, I recently tried a
5:22
case. My opposing counsel didn't own a
5:24
cell phone. And you know, nicest guy in the
5:27
world tried a great case, just a pleasure to deal with.
5:31
But I just thought that that was fascinating to me, you know,
5:34
because there were a couple. I learned it because the court
5:36
was trying to reach him because he was late and we couldn't do
5:39
it because he doesn't have a cell phone.
5:40
And, you know, so I agree that there's just a lot of lawyers
5:44
who, who don't have the access that we are privileged to have
5:48
and, and have the, the professional networks that we
5:51
all have that we think it's just commonplace.
5:53
And I guess maybe I certainly take for granted the fact that
5:57
it's so easy to find the stuff out 'cause that opinion was sent
6:00
to me, you know, by an organization I'm involved with,
6:02
you know, so easy for me to say, oh, it was right at my
6:05
fingertips when I otherwise, you know, certainly would not have
6:08
seen it absent being published in perhaps some other way or I
6:11
might have missed it all together.
6:13
So I mean I get. Well, and I'm not excusing a
6:17
lack of knowledge, I think Sherrod, you're, you're the one
6:19
who one of our presentations brought forward the duty of
6:22
competence under the and and I don't have the the citation at
6:26
at my fingertips, probably because I rule 1.1. 1.1 yeah.
6:31
See, I lack the competence there, but but the idea that we
6:38
are supposed to stay competent and yet at the same time when
6:41
you're running a busy practice, it is challenging I think to
6:45
stay on top of every single thing that is happening out
6:47
there. Let me just pose this question.
6:50
How many of us have heard the words?
6:53
I'm not a technology person. This is a common retort by
7:00
lawyers all the time. Well, I don't know about that.
7:03
I'm not a technology person. Well, the rules require you to
7:07
be a technology person. And how do you become a
7:11
technology person is by checking things out, you know, trying
7:16
things out and figuring out what what works best for your
7:20
practice and your approach. You know, I've been trying cases
7:23
using an iPad since 2014 and people would look at me like I
7:28
was crazy when I would suggest that all we've done this with
7:31
pen and paper. We use butcher paper in our
7:34
trials. We don't do this fancy
7:36
technology stuff. OK, great, you bring your
7:38
butcher paper. I'll bring high resolution
7:40
images on an iPad and let's see what the jurors pay attention
7:44
to, right? But this is, I think, a constant
7:46
battle in our field with people who either refuse to use the
7:52
technology or rush in headlong and don't understand what
7:57
they're doing without doing a little bit of due diligence
8:00
because they want to just get it done and move to the next thing,
8:05
but because they don't want to be bothered by it.
8:07
I'm not sure which is more challenging, the ones who rush
8:11
in and don't understand what they're doing or the ones who
8:14
almost take pride in, in the Luddite approach.
8:17
I will say I am somebody who likes butcher block.
8:20
I I think switching them the modalities up in trial, I I'm
8:24
not going to be on butcher block the whole time, but when you
8:27
have an expert create something, you know, point by point in
8:31
front of the jury and market, it may not go into the jury room.
8:35
But when you reference that in closing argument, the jurors
8:37
like, Oh, I saw that thing get built.
8:38
So sometimes that can be, I think a useful tool, but it's
8:42
got to be a tool in a toolbox, not not only thing you're going
8:47
to. Yeah, I think it's like the
8:49
fluidity like you say, I mean like you got to be able to do
8:52
everything. And I think the people who are
8:53
falling into these pitfalls of, of just using it and not
8:57
checking it are the people who are like, yes, my job just got a
9:03
lot easier, you know, And that's, I think the problem that
9:09
that's existing. And I'm sure there's people
9:12
doing it right now and briefs being filed and, and motions
9:16
being filed right now that unfortunately are, are falling
9:20
into this. And you know, I, I'm, I, I bet
9:24
it also just, I'm totally speculating, but just runs the,
9:28
the spectrum of first year lawyers to 55 years, 60 year
9:32
lawyers. You know, it's because think
9:36
people just look at it like they don't have it makes their life
9:39
so much easier and they can do so much more.
9:42
And the irony of it is, it does. And you can, it's just if you
9:47
use it effectively and efficiently, it can be that much
9:52
more useful than just going in headlong and then having to face
9:56
something like, you know, these lawyers that in this appellate
9:59
case. And look, we're all human.
10:02
We all make mistakes. But looking at it, it was not
10:05
only in the opening, but the reply brief.
10:08
I mean, I can forgive the opening brief.
10:10
You made a mistake that, you know, something got by.
10:13
But also in the reply brief he just says I was mailing this
10:16
thing in. And again, maybe I'm a little
10:20
too paranoid, but mailing in an appeal is a pretty serious issue
10:24
in my book. Yeah, the referees have a lot of
10:28
power. They, it's not like we're
10:31
sending this in to, you know, a corporate board that there's
10:36
some memo that's got raw erroneous information and maybe,
10:40
you know, someone gets a slap on the wrist, maybe someone gets
10:43
fired. You know, this gets reported,
10:46
there's monetary sanctions as we know, and then obviously State
10:49
Bar issues that come into play. So it's there's real, real
10:53
consequences with respect to the people who are overseeing what
10:56
we're doing. Rightfully so going back to and,
11:00
and, and there's the, whether it's the ABA or other NGS that
11:04
have come out with model rules on how to engage with AII think
11:09
that California Rules of Court also recently had one.
11:12
I, I think I kind of boil it down to if you're signing it,
11:16
you own it as a summary of, of what the rule is.
11:20
And I know you, Sherrod and I believe you Marshall, also use
11:25
some AI tools in assembly. So if somebody here who is, is
11:30
thinking of using AI but is not sure how to do the best quality
11:35
assurance on the, if you, if you sign it, you own it front.
11:38
What methods do you have to double check things to make sure
11:42
that what's going out actually is what it purports to be?
11:48
So miles before we jump into that, it is the Rural
11:52
Professional Conduct 1.1 and it's footnote one that talks
11:57
about technology. You're going to get our our
12:00
ethics credit for for the MCLE for this episode with all your
12:03
navigations. Nothing brings viewers like
12:07
citing the rules of professional conduct.
12:11
You'll see the uptick in this moment of the of the.
12:15
Podcast. I mean one thing I have started
12:18
to do for other firms and I recommended that they do this in
12:23
in seminars that I give, is to create a policy within the firm.
12:27
Of. If you're going to allow it and
12:30
if so, when it's going to be used, who can use it and how it
12:33
can be used. So that way at least if
12:36
something does happen, right, If you have a larger firm and you
12:39
have newer associates, unless you block specific sites, which
12:42
obviously you can do, but they can easily access them from
12:45
personal computers. So but, and again, I don't think
12:49
and I blocking and not using a is I don't think that's the
12:52
right approach here. So I think it's using it, but
12:55
clearly identifying again how and when it can be used so that
13:00
at least you have that framework so everyone understands.
13:03
And then from there it's training, right?
13:05
You've got to train if you're, if you're having younger
13:08
associates do it, you got to train them on what they can do
13:12
and how they can do it and who they can provide the information
13:15
to and then confirm that it's been checked etcetera.
13:18
So if you use that as the starting point, as I've
13:21
recommended and I've, and I've reviewed a lot of firms policies
13:25
and there's firms approach it very different ways.
13:28
And I appreciate the differences and concerns that each and every
13:31
one of them have 'cause they dramatically differ from
13:34
practice type to size of firm to age of the lawyers.
13:37
You know, they're, they're all different concerns.
13:40
And so they address them. There's no one-size-fits-all
13:43
policy. And so from there, I think that
13:45
you can add value for your clients.
13:48
And I mean, I have on my other screen right now all day, I
13:51
probably got 15 Westlaw tabs open and then three different AI
13:55
platforms open because I'm bouncing back and forth between
13:58
themes of a case I got coming up for trial and then legal issues.
14:02
And then once I get them through Westlaw, I'm plugging them into
14:05
AI to, to get to get additional information and sort of
14:08
streamline arguments. And so I'm just constantly going
14:11
back and forth, which is amazing.
14:13
And if I was just using Westlaw, I would not be as efficient as I
14:19
have been today with respect to the work that I'm providing on
14:23
behalf of my client. And nothing I've done is going
14:27
to result in a fabricated legal issue.
14:31
I'm not, I'm not submitting anything to the court.
14:33
Everything I'm doing is for my own information and but nothing
14:36
is incorrect because it's all BeenVerified and I'm doing it
14:41
diligently. If I was, you know, just on
14:43
Westlaw, that's a little old fashioned.
14:45
You know, I think I'm hampering my efforts if I'm just on AI but
14:49
think I'm a little too risky. So that's literally my screen
14:52
right here is what's I'm looking at.
14:54
I mean, obviously having a policy now we've had some new
14:58
people that are come that have come on board.
15:00
So we're working on a written policy because you know, my
15:03
approach was to talk directly to our folks about what to do and
15:07
what not to do. So we're having to change that a
15:09
little bit at this point. But having a policy I think is a
15:13
great idea. Also, knowing what tools to use
15:17
for what task, because if you're looking for case law citation,
15:24
right, maybe going to Chad GBT or Claude is not the best
15:29
approach. Maybe you want to go to Westlaw
15:31
and get that silo to do the work for you.
15:36
Or if you get something from Chad GBT, take that citation and
15:41
dump it into Westlaw and see, all right, is this a legitimate
15:44
citation? I've actually done this with
15:46
discovery responses that we got and said, well, give me the
15:50
strength and weaknesses and tell me whether or not there is any
15:53
case law that supports this. And to my surprise, Claude was
15:58
able to give me a research memo and he gave me 3 cases on a
16:03
premises liability case. I knew two of them, so that was
16:06
encouraging. The third one I wasn't from
16:09
unfamiliar with. So I took it and I put it in
16:11
Westlaw and vetted it and it was a legitimate.
16:16
So that's the way you look at it as opposed to saying, oh, here
16:20
are three cases. That third case could very well
16:22
have been a hallucination and it could have gotten by me if I
16:26
hadn't taken that extra step. So what Marshall is doing is, is
16:30
spot on in terms of making sure that you're using different
16:35
platforms to vet the work and the information that you're
16:39
getting. And having talked to some folks
16:41
who work in AI, that's exactly what they say that they do in
16:46
their approach is not just take wholesale with what somebody
16:49
says, just like if we talked about a case, maybe I'll run it
16:52
by another attorney just to see what the other point of view is.
16:56
It's really not that different. But the reality of it is, from
17:00
our point of view, if we're looking at all this stuff and
17:04
we're saying this is great, you can't have the expectation that
17:09
when you put the information in there that the end product is
17:11
going to be perfect. And I think that's what's
17:14
getting people into trouble is not remembering dial up right
17:19
when you had to connect to the Internet.
17:21
Now, if that web page doesn't load within a millisecond,
17:24
you're going, my God, why is it so slow?
17:27
When in reality, if you go back to when dial up was a thing, it
17:32
really was not something where we would sit there and be like,
17:35
oh, it didn't load in a millisecond.
17:37
You know, it may take a minute for it to load.
17:39
So now we just have to readjust our expectations and, you know,
17:43
fight our own lazy tendencies. I don't think it's made me do
17:47
anything faster. I don't think I have speed up
17:50
the would I having implemented AI tools that I use, I'm not a
17:55
completing a task quicker. What I'm doing is it's, I think
17:59
it's expanding my thought process and how I approach
18:02
things and allowing me to sort of open up a whole new Ave. of
18:06
allowing me to think about different things and synthesize
18:10
those thoughts in a different way.
18:13
So I don't think like today, what I've been working on.
18:15
If I didn't have AII, don't think it would have taken me
18:19
longer. I think I've just been able to
18:21
look at more things and just tackle different issues than I
18:28
otherwise may have not thought, maybe have not thought of.
18:30
So it's sort of open more doors for me.
18:33
So I'm, I mean, I suppose I'm I'm looking at more information
18:38
in perhaps the same amount of time, but it's not like I'm
18:40
sitting here being like, oh, now I because of AII accomplished
18:44
the task in one hour when it would have taken me otherwise 3
18:47
hours. I'm still spending all day on
18:49
what I've been doing. It's just another way of adding
18:53
a value. But that's the difference
18:56
between your mentality of seeing this as a tool to enhance your
19:01
abilities, your skills and training as opposed to using it
19:05
as a crutch so that you can go sit on your porch and have a
19:08
drink at the end of the day, right?
19:11
So there there is that approach because you're looking at it as
19:16
something that, hey, this can make me better as opposed to,
19:19
well, this can just get me to get away from this and not have
19:24
to deal with it. I would say your your two
19:26
approaches are very different from mine.
19:27
I I got 3 appellate briefs done this morning and and spend the
19:30
rest of the day surfing. And, you know, I think that's
19:35
amazing. And I hope the.
19:38
Waves were great, but Miles, there's a lot of firms who are
19:43
doing that. And we probably touched on this,
19:45
but you know, on the plaintiff side, you, you don't get paid by
19:49
the hour. And so there's the incentive to
19:52
do that. And I think that that is
19:55
dangerous. There's less of an incentive for
19:58
me to do it even if I wanted. But on your side, you know, the
20:01
more time you spend on your case, the return on that
20:04
investment just goes down, down, down.
20:07
So I I just think on the play of side there's a lot more appetite
20:12
to just have it be done so you can be on the beach surfing.
20:16
Each side, whether you're a billable hour or a contingent
20:19
fee, there's a pro and a con to the approach and there is a
20:23
subconscious bias that you need to, I think of that on the
20:27
contingent fee side. We, we, we make sure, as I'm
20:30
sure Sharad, you do with, with your clients and, and, and the
20:35
people that you work with. We don't care if it's 10 hours
20:38
or you know, thousands of hours into a case, the the case is
20:42
going to get the same attention. And some of them we make more
20:46
per hour and some of them we make less than minimum wage.
20:50
And such is life. And in the long run, it all
20:52
leaves out. That's an important point in the
20:56
sense that the model that your firm runs and my firm runs is
21:01
very different than volume firms who are just churning cases in
21:06
and out. And that same case is being
21:08
settled for a fraction of the amount that you or I can get for
21:12
the same client. So the incentive for those firms
21:16
may be very different in terms of, all right, let's just get it
21:19
in, get it out, get a demand letter, whatever they want to
21:22
throw at us. Maybe we'll take as opposed to.
21:25
No, I'm going to work this up and get a better result for the
21:29
client. There clearly are and I've
21:31
spoken to some people who have done what you jokingly said,
21:36
Miles, which was create a brief in 20 minutes and gone to the
21:40
beach and gone surfing and guess what?
21:42
The brief was perfectly fine. It's not to say that every time
21:46
you do do this, you're going to get fabricated and hallucinated
21:49
cases if it's a pretty straightforward issue.
21:52
And certainly if you know the cases that once you see them,
21:55
right, which it's happened to me a lot of times, Sherrod, you
21:58
just mentioned you do two of those cases, boom, you're done.
22:01
And so it can be really helpful. And I'm sure hundreds and
22:06
hundreds of briefs have been filed based upon just 100% AI
22:12
generated with accurate case law and very thoughtful and thorough
22:17
analysis and probably have prevailed.
22:19
And those lawyers are sitting on the beach with a big smile on
22:23
their face. And that's great.
22:25
I don't, I think it's, you know, risky, but if they did it
22:29
responsibly, I'm not going to sit here and say that you
22:32
shouldn't have done that because they know No, Marshall, I, I
22:38
knew the cases, the analysis was good.
22:40
I checked it and I was out of there.
22:42
Good for you. That is what AI can offer.
22:46
But that lawyer is a very intelligent lawyer and still
22:52
being responsible because he has a he or she have a has a deep
22:55
knowledge of the law and their particular field.
22:59
You're enjoying what we're doing here with Coopers Code.
23:01
Please help us extend our reach by giving us a five star review
23:03
on Apple Podcasts and consider sharing an episode with a
23:06
friend. Additionally, we want to know
23:08
who you want to hear on the show.
23:10
Have an interview suggestion, e-mail us at podcast at Coopers
23:12
dot law. I'm going to pivot now to
23:18
Sharad. You submitted for our discussion
23:21
Nvidia's $100 billion investment in Open AI.
23:25
And I thought this was going One Direction.
23:27
And then you, you clarified for me it's utility in another way.
23:31
So do you, do you mind unpacking that for us?
23:34
Sure. So in the last week and 1/2 or
23:37
so, we've seen NVIDIA announced that it's investing $100 billion
23:42
with AB into Oakland AI. Microsoft's investment in open
23:48
AI is reported to have gone from 13 billion in value in terms of
23:53
what they invested is now worth $150 billion.
23:57
And we've seen Larry Ellison's net worth in one day jump by
24:01
$100 billion. So a lot of defense firms and
24:07
insurance carriers are complaining, right, about these
24:11
nuclear and thermonuclear verdicts.
24:14
This is ridiculous. Why is this happening?
24:17
Well, when we look at this, if the jurors are exposed to these
24:21
kinds of numbers and seeing the kind of money that's being
24:25
thrown around, when a plaintiff stands up and says, hey, award
24:30
my client 30 or $40 million, the jurors psychologically, in my
24:35
opinion, are not seeing those numbers be as large as they
24:39
would have been thought to be in the 80s or 70s or or 90s.
24:46
So it is a very different dynamic and atmosphere.
24:49
And there was a, you know, sex abuse case in Los Angeles where
24:52
the jury awarded $114 million in that case just recently.
24:58
So when you see those numbers as a defense firm, you have to ask
25:02
yourself, all right, when I'm sitting there saying no, this
25:05
case is not worth anything, that the medicals are not there.
25:09
The real risk is that if the jury is willing to award large
25:14
numbers on non economic damages, it could end up being a massive
25:18
verdict, which is what we were seeing significant increase
25:22
since 2020 in those cases. Sherrod, I fully expect that the
25:27
next opposing counsel who you go to trial against, who is doing
25:30
their due diligence will be filing a motion lemonade on
25:33
referencing NVIDIA and open AI and investments.
25:36
And reptile and reptile of. Course, yeah, No, no
25:39
reptilement. That you guys don't even know
25:41
what that is. You've never heard of reptile?
25:43
I would love. I love getting the opposition to
25:46
that MILI don't even know what that is.
25:48
You know, I, I, I will. This is a little bit of a
25:51
tangent, but the last time we opposed it, we acknowledged its
25:54
existence and said that that that is so far back in the
25:57
iterations of development that is now referred to as the
25:59
dinosaur theory. We've got our new things so.
26:02
I'll bet if I look hard enough at the books behind you, Miles I
26:05
they might see the reptile theory book somewhere.
26:09
Yeah. I don't actually have that one,
26:14
so I think the only one you can really make out is Dune, which I
26:19
do not use for for legal argument very often.
26:23
But you have a copy. I didn't have to go far.
26:28
Yeah, well played, well played. Joking aside, do you have
26:33
thoughts on on that as in terms of dealing with it on the other
26:37
side? Because I know that some of the
26:39
some of the work that you do is defence and some of it is
26:43
defence where people are asking for money.
26:45
If someone were to use that as an argument for gosh, it's not a
26:51
lot of money, just look at this out of the other.
26:55
I think that these types of analogies are pretty commonplace
27:01
whether you you know people show jet airplane, you know fighter
27:06
airplanes, the Monet right from your classic cases in in closing
27:10
argument to anchor these jurors is nothing new.
27:14
I think the difference, what is new is the dollar figure and the
27:20
numbness that we have to big numbers with, with not just
27:26
Larry Ellison, with all those super, super wealthy people.
27:30
And also I think with social media and the ability to see how
27:34
they live and how it looks so easy to get a big yacht and get
27:40
a few jets. And so I think from a societal
27:44
standpoint, jurors don't view those numbers as, Oh my gosh,
27:49
they're huge. Like you said, mentioned the
27:51
$114 million verdict last week or the week before in LA
27:54
Superior Court. I mean, that lawyers gotten a
27:57
few of those to my knowledge. So he knows how to do it and
28:01
good for him. I just think it's more a
28:06
numbness that we that that that everyday people have to those
28:09
big numbers and the willing to understand that injuries can and
28:15
perhaps should result in reward of a of a substantial number
28:20
like that. I think it's on the defense side
28:22
and I'm I'm 95% of the time on the defense side.
28:26
It's our job to be as creative as the plaintiff side is in
28:31
coming up with ways to, you know, negate all that.
28:36
You know, with the fighter jet example, I always laughed
28:38
because if the fighter pilot who died, who was in that jet died,
28:44
the federal government has a set limit on what that person's life
28:47
is worth. And it's not a lot of money.
28:49
And so anytime that reference has been made, it certainly
28:54
opens the door to say that. I think at the time, last time I
28:57
remember bring it up, it was like $180,000 or something that
29:01
the family got. And so, you know, when you're
29:05
valuing a life, well, the federal government certainly
29:08
values it and it's not a 7 figure number.
29:11
So it's things like that that keep it fun and entertaining.
29:14
It's. Chess, not checkers.
29:18
Yeah. Well, switching gears a little
29:21
bit, I, I did want to bring forward, I was playing around
29:25
with Claude over the past. I think it's been within the
29:27
past two weeks that a couple of new things got unlocked.
29:31
And one is that Claude, if you, if you toggle a button, can now
29:37
access all ones past chats. And it used to be Claude could
29:41
only work in the chat that you add in the moment.
29:44
But now if you have a, a subscription, it can look at all
29:48
the prior chats and use that to inform it's discussions going
29:52
forward. And the second one is it's apps.
29:57
It can interface with apps and create Word and Excel files.
30:00
And the third that just showed up I think on Friday was it can
30:03
schedule, meaning it can, if you allow it, it can access your
30:07
calendar and add to your schedule, but it can't delete.
30:12
So if it adds something, it can't delete it and it can't
30:15
move things. I bring these forward in in part
30:19
because this is a long entry to to our conversation.
30:23
But it seems like just two months ago that Anthropic was
30:29
talking about a research exercise that it did in a
30:35
sandboxed environment where it allowed an AI to go into a
30:40
sandboxed environment, work in the server where there was
30:44
information in various emails about.
30:46
And I'm going to overly simplify this, but in essence, one of the
30:49
people who had the power to pull the plug on the AI, the IT
30:54
person had emails that indicated that he was having an affair.
30:59
And in that study, the AI threatened to blackmail the IT
31:04
person if the IT person shut it down.
31:06
So, but my springboard here is, I find it interesting that these
31:12
were issues a couple months ago and now it's starting to make
31:15
its foray out into the world. So that's a springboard for you,
31:18
Sherrod Marshall, in terms of are you using these tools?
31:22
What do you think about them? Do you recommend people play
31:25
with them not and go? So I'm only allowing the chatbot
31:29
to blackmail my hologram. So that's one safeguard.
31:35
The other piece of it is that when you look at these things,
31:39
the chats are supposed to be, if you have the premiums version
31:43
are supposed to be in their own silo.
31:45
So it's if it's accessing it. And I believe the last time I
31:48
looked, when you delete a chat from Anthropic, it basically
31:53
gets deleted out of their system within 30 days of you hitting
31:57
delete. So unless they've changed that
32:00
policy, and I'm not sure if they have, that was the policy before
32:05
and now they have it so that it can generate Word and Excel
32:09
documents. And I have not used that feature
32:12
yet. And frankly, at this point, I'm
32:14
not inclined to give anybody access to my calendar.
32:17
So that hasn't been an been an issue.
32:20
But the Word and Excel documents seem to be useful because before
32:25
when you would download the chat, if you asked it to
32:29
generate something, it would be in a format that basically you
32:33
had to play with in order to open.
32:35
And it had all these weird codes in it.
32:37
And now I think they're trying to solve that issue.
32:41
For me, it's playing with each of them.
32:45
I agree with regarding the calendar issue.
32:46
Some things I like to wait a little bit until I'm just going
32:49
to give access to my Outlook, which is for me by far and most
32:53
the most important program I use is Outlook.
32:58
But I think it's trying all these different things out and
33:03
trying them in different ways. I also agree regarding like the
33:07
Word and Excel files. I've noticed when I've tried to
33:10
migrate things from Word to Excel, there's sometimes it's,
33:13
it's not quite seamless or as seamless as I want it to be.
33:17
And I think it's just getting better and better as you go to
33:21
do that. As far as saving chats, that's
33:24
not some, I, I generally have looked at it starting fresh each
33:29
time. I haven't had sort of this
33:30
continuing and ongoing conversation really day by day,
33:34
but I think that that's super helpful because you can just
33:39
build on it. And I really like calling it a
33:41
conversation because you can get tired and you can say, let's
33:45
pick this up tomorrow where we left off and you can just do it.
33:49
And it remembers exactly all the points that you were discussed,
33:54
all the nuances, and you can just keep going from there.
33:58
So it's just really just as such a great extension in the human
34:02
perspective and how we interact without the memory lapse of real
34:08
life humans that's going to occur.
34:11
One other thing to keep in mind is that Copilot, if you're use
34:14
Outlook, there's now a button at the top right corner that says
34:18
Copilot and you can use that and then it gives you the other.
34:23
They have options like prompt coach, writing coach, idea
34:27
coach, career coach, learning coach.
34:30
So those are all things that are being rolled out and I haven't
34:33
had a chance to look at them in any detail, but they're now
34:37
putting copilot into our outlook, which could or could
34:42
not be problematic depending on, you know, how you look at it.
34:45
But that's another thing that's coming out.
34:48
And I know W Law's Co counsel also has a plug in that you
34:52
could put in where it can help with e-mail writing and that
34:56
sort of thing as well. Full honesty, I just clicked the
35:00
copilot button for the first time.
35:02
I was. I'm sitting right here.
35:03
A little bit with Copilot and one of the things, if you have a
35:07
large body of, of e-mail communication, it will start
35:11
trying to mimic you and offer drafts in your style.
35:15
So, but it's, it's still, it looks like it's pretending to be
35:21
you instead of being you. If you if you read the content,
35:24
I feel like I'm still pretending to just be me.
35:26
So. Probably pretty good so far
35:30
you've. Been doing a very good job and,
35:32
and the conclusion of this, we'll, we'll tell everyone that
35:34
this entire podcast is actually Google LM using our avatars,
35:38
right? Well, I I'm pretty sure you have
35:40
a claim. Against them because the voice
35:42
is remarkably similar to yours. So maybe that's an angle that we
35:46
could take. Maybe I should be looking to
35:48
somebody for IP? Representation as far as I would
35:51
say on the. Copilot side and and.
35:53
This is something that I, I would, I would hope to see and,
35:55
and maybe there are people who've played around this with
35:57
this more. My schedule was always a game of
35:59
Tetris. And so if, if copilot reached
36:03
the point where it could start suggesting time blocks and
36:07
moving things around and getting to know my schedule and offer
36:09
improvements on how I do my work, that would be, that would
36:13
be huge. I think that would be amazing.
36:16
Just to see how you can be more efficient with managing your
36:20
inbox and, and things like that. It's that's a whole area for.
36:26
Me, that would be such. A stress reliever because for me
36:30
it's, it's the emails in the night and things like that that
36:34
are that keep me up and, and scheduling and remembering if
36:37
you did things. And I'm sure there are so many
36:40
gaps in my efficiency, even with assistance and things, physical
36:45
human assistance and things like that.
36:48
So I'm going to dive into this and report back for next
36:51
episode. Well, I know, Sharad.
36:55
We're going to talk about Google Notebook, but I think I think
36:58
we're getting close on as far as time for our our recording
37:01
window for this. So I think we'll put a pin in
37:03
that, save it for next time, tease a little bit in terms of
37:07
of that for anyone who's. Listening.
37:11
If you have found some cool new stuff want to share, drop it in
37:15
the comments, send us an e-mail at podcast at Coopers dot Law.
37:19
And as always, for all of you doing justice out there, happy
37:23
hunting.