1
00:00:00,000 --> 00:00:06,280
Welcome to Threaded Minds, Law and Beyond with Sohaib Zada.
2
00:00:06,280 --> 00:00:12,040
This podcast is your space to explore the intersections of law, business, personal growth
3
00:00:12,040 --> 00:00:13,880
and modern life.
4
00:00:13,880 --> 00:00:20,440
I'm your host Sohaib Zada, a legal expert, educator, practitioner and lifelong learner.
5
00:00:20,440 --> 00:00:25,580
Together we will unravel the threads that shape our world.
6
00:00:25,580 --> 00:00:31,600
From legal principles to career strategies, self-development and the challenges of our
7
00:00:31,600 --> 00:00:33,200
interconnected world.
8
00:00:33,200 --> 00:00:38,320
We dive deep to uncover stories and insights that matter.
9
00:00:38,320 --> 00:00:42,680
Let us connect the dots and discover new perspectives.
10
00:00:42,680 --> 00:00:47,960
This is Threaded Minds where every thread leads to a story and every story sparks a
11
00:00:47,960 --> 00:00:52,640
new idea.
12
00:00:52,640 --> 00:00:54,160
Welcome to Threaded Minds.
13
00:00:54,160 --> 00:00:59,120
In today's episode we have a special exceptional guest that I promised you I'm going to have
14
00:00:59,120 --> 00:01:01,080
in this show.
15
00:01:01,080 --> 00:01:09,720
I have Erika Wernemann-Ruth, a legal consultant with expertise in AI, privacy and compliance.
16
00:01:09,720 --> 00:01:10,720
Welcome Erika.
17
00:01:10,720 --> 00:01:11,720
Thank you.
18
00:01:11,720 --> 00:01:13,960
It's a pleasure to be here.
19
00:01:13,960 --> 00:01:16,880
I'm happy to have you here in the show.
20
00:01:16,880 --> 00:01:25,520
Alright so today we are going to explore Erika's career journey and how her work in consulting
21
00:01:25,520 --> 00:01:27,880
has evolved over the years.
22
00:01:27,880 --> 00:01:32,160
There are very interesting key themes that we will cover in this episode.
23
00:01:32,160 --> 00:01:39,800
For example, what early career lawyers ask, what are the evolving challenges that we have
24
00:01:39,800 --> 00:01:43,640
in privacy and compliance.
25
00:01:43,640 --> 00:01:47,800
Before we begin I would like to start with just a simple question.
26
00:01:47,800 --> 00:01:51,300
I have a question about your career journey.
27
00:01:51,300 --> 00:01:57,360
Could you walk us through your career journey and how your work in privacy and for example
28
00:01:57,360 --> 00:02:00,240
in consulting has evolved over the years?
29
00:02:00,240 --> 00:02:02,200
Yes of course.
30
00:02:02,200 --> 00:02:05,760
So I am going way way back.
31
00:02:05,760 --> 00:02:11,000
I actually studied economics and then I did a conversion degree to law.
32
00:02:11,000 --> 00:02:15,800
I trained and qualified with a city law firm in London.
33
00:02:15,800 --> 00:02:22,280
Originally actually a facilitator so I worked in disputes and then in the run up to GDPR
34
00:02:22,280 --> 00:02:29,280
a lot of my clients were technology clients and I just thought the topic was quite fascinating
35
00:02:29,280 --> 00:02:34,840
so I slowly pivoted across rather than sort of an abrupt change.
36
00:02:34,840 --> 00:02:44,080
I then spent some time in the Middle East, I was out in Dubai for about six years and
37
00:02:44,080 --> 00:02:50,200
in that capacity I supported the wider region so I worked across the Middle East, Africa,
38
00:02:50,200 --> 00:02:56,440
sort of bridging or linking the different threads between Asia sometimes and Europe
39
00:02:56,440 --> 00:02:59,120
and UK which is great fun.
40
00:02:59,120 --> 00:03:06,440
I returned to the UK, kept working in data protection primarily but increasingly also
41
00:03:06,440 --> 00:03:12,440
in artificial intelligence so I spent some time on government boards looking at AI ethics
42
00:03:12,440 --> 00:03:18,800
before the EU AI Act came out and generally I always thought emerging tech was an area
43
00:03:18,800 --> 00:03:26,040
that I found fascinating so early on I looked at cryptos, blockchains, various forms of
44
00:03:26,040 --> 00:03:29,520
AI facial recognition and driverless vehicles.
45
00:03:29,520 --> 00:03:34,560
It seemed like if there was something quite techy I tended to try to get involved from
46
00:03:34,560 --> 00:03:40,400
a data protection perspective and then I so I did that I ended up spending a decade in
47
00:03:40,400 --> 00:03:46,440
private practice I spent some time on remote environments to major global players I worked
48
00:03:46,440 --> 00:03:51,760
with some early stage companies and I think it's such a common story isn't it but I loved
49
00:03:51,760 --> 00:03:57,160
my time at the firm I was there for like I said a decade and quite happily on a partner
50
00:03:57,160 --> 00:04:02,920
so that's what I saw in my future and then had my son and realised actually I could do
51
00:04:02,920 --> 00:04:07,920
it with a different pace of life so I was headhunted I was recruited for a start-up
52
00:04:07,920 --> 00:04:14,360
so I did that for a while it sort of dawned on me relatively early on that start-ups perhaps
53
00:04:14,360 --> 00:04:18,520
don't need someone like me full-time they need me to come in and put the infrastructure
54
00:04:18,520 --> 00:04:26,600
in place so after a bit of time with the start-up I took the leap and I set up my own company
55
00:04:26,600 --> 00:04:32,800
so I've been working as a consultant via my own business for about a year and a half.
56
00:04:32,800 --> 00:04:36,940
That's very nice actually this is fascinating when you look at your resume and you see like
57
00:04:36,940 --> 00:04:43,060
for example you study economics and then you moved into the law I think that's what's important
58
00:04:43,060 --> 00:04:51,000
to combine multiple areas multiple majors and try to come up with a niche area where
59
00:04:51,000 --> 00:04:56,800
the clients need your legal services or what you could offer them and it's actually interesting
60
00:04:56,800 --> 00:05:02,120
I noticed that your new resume worked with EverShed Sutherland and then you moved into
61
00:05:02,120 --> 00:05:06,920
your own practice so probably you have good knowledge about the differences between working
62
00:05:06,920 --> 00:05:13,520
for a law firm and starting your own I would say boutique law firm or your own business
63
00:05:13,520 --> 00:05:17,200
providing legal consultancy and services.
64
00:05:17,200 --> 00:05:22,600
It's interesting worldwide in many jurisdictions nowadays you can actually practice law without
65
00:05:22,600 --> 00:05:30,040
a law degree for example in the UK they opened a route SQE for people to practice law you
66
00:05:30,040 --> 00:05:35,920
just need two years of legal practice and then two exams SQE 1 and 2 and you can be
67
00:05:35,920 --> 00:05:40,960
qualified as a solicitor I'm not sure but I heard about California there is another
68
00:05:40,960 --> 00:05:48,700
route for that so you can see there is a shift in the legal education or the competencies
69
00:05:48,700 --> 00:05:57,760
that law societies or regulators expect in the future legal professionals so we'll see
70
00:05:57,760 --> 00:06:04,760
more updates we may see updates with regards to the required qualifications so it doesn't
71
00:06:04,760 --> 00:06:10,000
need for some people to start with a law degree and even if you start with a law degree you
72
00:06:10,000 --> 00:06:16,680
need to prove that you're competent in that specific area by training and having multiple
73
00:06:16,680 --> 00:06:20,960
experiences as you did in different locations it's very interesting.
74
00:06:20,960 --> 00:06:25,600
I think just sort of building on that for a second I'm actually increasingly working
75
00:06:25,600 --> 00:06:32,040
with people that are not in the legal profession so in particular around AI there are I think
76
00:06:32,040 --> 00:06:37,600
multiple paths into the profession so I'm seeing people coming with a philosophy background
77
00:06:37,600 --> 00:06:42,280
they're looking more towards the ethics aspect obviously in compliance you don't have to
78
00:06:42,280 --> 00:06:47,360
be illegal although many people are you don't have to be a lawyer so you can come from a
79
00:06:47,360 --> 00:06:52,840
compliance background a governance background a risk background so I'm sort of interacting
80
00:06:52,840 --> 00:06:56,640
with people from many different backgrounds.
81
00:06:56,640 --> 00:07:02,440
I know especially with AI I'm doing my thesis in crypto trading platforms and the interconnection
82
00:07:02,440 --> 00:07:08,960
with securities compliance and I find out with these new fields we need experience people
83
00:07:08,960 --> 00:07:17,240
from different backgrounds whether it's law ethics social science economics but the problem
84
00:07:17,240 --> 00:07:22,720
the people from public law or human rights or some areas they don't they don't understand
85
00:07:22,720 --> 00:07:28,360
the terms or our specialty so they think we are creating something innovative or something
86
00:07:28,360 --> 00:07:29,720
that doesn't exist.
87
00:07:29,720 --> 00:07:34,160
I find challenging I find it challenging talking about my field sometimes.
88
00:07:34,160 --> 00:07:35,160
Interesting.
89
00:07:35,160 --> 00:07:44,020
All right I know that you often get approached by students or early career lawyers or aspiring
90
00:07:44,020 --> 00:07:52,640
lawyers or solicitors whether it's on LinkedIn on social media so what are the main if you
91
00:07:52,640 --> 00:07:57,880
can summarize for us like the main questions that those people ask with that with regards
92
00:07:57,880 --> 00:08:04,280
to their profession to their practice in the future.
93
00:08:04,280 --> 00:08:10,320
I think that's broadly speaking two different themes there are people are asking about particular
94
00:08:10,320 --> 00:08:18,280
skills or what they can do to stand out certifications generally around upskilling and I guess being
95
00:08:18,280 --> 00:08:24,400
an attractive either employee or somebody who could perhaps make a change into the profession
96
00:08:24,400 --> 00:08:30,480
from a different practice area and I think the other section of queries I tend to get
97
00:08:30,480 --> 00:08:36,000
is more around I don't know like a crystal ball like what do I think about the future
98
00:08:36,000 --> 00:08:41,640
is it worth going into this area or that area what are the risks are the profession going
99
00:08:41,640 --> 00:08:45,680
to is the profession going to change are the roles going to be very different.
100
00:08:45,680 --> 00:08:52,120
So I guess those are the two sort of broad thematic issues people tend to raise out about
101
00:08:52,120 --> 00:08:55,280
and I try to respond as much as I can.
102
00:08:55,280 --> 00:08:59,040
What do you typically respond to them what advice you offer to them.
103
00:08:59,040 --> 00:09:05,880
It's a good question right so I think it depends people when people look at my career and they
104
00:09:05,880 --> 00:09:10,920
a lot of people assume there were some strategic decisions I made along the way I didn't I
105
00:09:10,920 --> 00:09:16,200
was just along for the ride I made different decisions based on what felt right at the
106
00:09:16,200 --> 00:09:21,640
moment there wasn't some overarching plan to my career and if I had to say anything
107
00:09:21,640 --> 00:09:27,000
I think your early career is for testing different things and seeing what you enjoy what you're
108
00:09:27,000 --> 00:09:32,720
good at I think people try to plan 10 15 20 years and advance and that's just not how
109
00:09:32,720 --> 00:09:38,920
careers unfold sometimes I've stayed in position longer positions longer than I intended to
110
00:09:38,920 --> 00:09:44,000
because I enjoy the people sometimes I just put my hand off on randomly got sent off on
111
00:09:44,000 --> 00:09:46,360
an assignment to discover that I really enjoyed it.
112
00:09:46,360 --> 00:09:52,120
So I think you can try you can try to pick up new skills you can certainly be a little
113
00:09:52,120 --> 00:09:59,240
bit sensible about it but I can think you can really plan a 20 30 year career as an
114
00:09:59,240 --> 00:10:04,600
early stage professional to find what you enjoy what you're good at and in terms of
115
00:10:04,600 --> 00:10:08,680
the skills I think that now you mentioned some of the different areas right increasingly
116
00:10:08,680 --> 00:10:14,160
I'm seeing roles open up in legal tech companies I think that would be a fascinating space
117
00:10:14,160 --> 00:10:20,400
to be in I think going straight in house will give you a different kind of skill sets that
118
00:10:20,400 --> 00:10:24,760
is increasingly I think into a bond in private practice as well if you can sort of bridge
119
00:10:24,760 --> 00:10:31,880
the gap between the technology and law in terms of the long term view nobody knows and
120
00:10:31,880 --> 00:10:36,720
if they tell you they know
121
00:10:36,720 --> 00:10:43,880
you see trends and you can sort of you can make some good informed guesses about how
122
00:10:43,880 --> 00:10:49,680
things are going to unfold but if I had to make predictions five years ago I don't think
123
00:10:49,680 --> 00:10:54,960
anything would have been really spot on so I think it's rather than trying to future
124
00:10:54,960 --> 00:10:59,560
proof your career by taking certain steps again think it's better to find out what you're
125
00:10:59,560 --> 00:11:04,720
good at and what you enjoy and sort of try to follow those paths.
126
00:11:04,720 --> 00:11:09,920
I do believe actually this is a great idea I do believe you need to start getting the
127
00:11:09,920 --> 00:11:15,400
training the hand on approach to see well where you find your passion because I remember
128
00:11:15,400 --> 00:11:20,400
myself when I graduated from law school I didn't have that passion for technology law
129
00:11:20,400 --> 00:11:25,600
and securities law and then I found this path so some people may take years may take months
130
00:11:25,600 --> 00:11:31,520
to find out where they can where they can find their passion or there they want to practice
131
00:11:31,520 --> 00:11:39,880
and most importantly I think we won't see as much as in the past with regards to general
132
00:11:39,880 --> 00:11:45,680
practices like in the morning family law in the afternoon criminal law and in the evening
133
00:11:45,680 --> 00:11:50,600
one case in for example commercial law or contracts I do think we'll see more specialized
134
00:11:50,600 --> 00:11:56,000
people more niche areas and I do believe that's what the legal profession requires at the
135
00:11:56,000 --> 00:11:57,000
moment.
136
00:11:57,000 --> 00:12:03,160
Yeah I mean I guess there's different views on that as well I think you're probably right
137
00:12:03,160 --> 00:12:08,840
and sadly from this sort of city perspective that I've always worked in that's always been
138
00:12:08,840 --> 00:12:15,560
the case but in a lot of scenarios it's also understanding having a broad exposure understanding
139
00:12:15,560 --> 00:12:21,360
different kinds of implications and how they might unfold and then speaking to the passion
140
00:12:21,360 --> 00:12:27,480
point I think this is something I don't really see trainees or younger professionals really
141
00:12:27,480 --> 00:12:33,720
think about as I get along in my career and I work with these really impressive people
142
00:12:33,720 --> 00:12:39,720
a lot of the time they are they might work in an area that to an outsider would seem
143
00:12:39,720 --> 00:12:46,360
quite dull but they enjoy it because they're so good at it so I think there's two sides
144
00:12:46,360 --> 00:12:50,840
to the passion that you can pick an area that you're passionate about and develop skills
145
00:12:50,840 --> 00:12:56,680
in it but I think if you stick in an area as you become more proficient and better I
146
00:12:56,680 --> 00:13:01,680
think you also find passion in those areas so it's sort of the depth of knowledge and
147
00:13:01,680 --> 00:13:07,400
the mastery of a topic makes you passionate about it if that makes sense so I don't think
148
00:13:07,400 --> 00:13:11,920
you have to go around chasing your passions necessarily to have a fulfilling career you
149
00:13:11,920 --> 00:13:16,480
can also just build your understanding and your knowledge I think a lot of people then
150
00:13:16,480 --> 00:13:19,400
become passionate as a result.
151
00:13:19,400 --> 00:13:27,160
Okay and let's say as an early career professional or aspiring lawyer I found my path in intellectual
152
00:13:27,160 --> 00:13:33,680
property or AI or compliance do you expect over the years of your experience do you expect
153
00:13:33,680 --> 00:13:38,560
from that lawyer or professional to know everything about that field or what's required what's
154
00:13:38,560 --> 00:13:42,720
the main skill that lawyer should have?
155
00:13:42,720 --> 00:13:49,800
They know everything they really really don't and I think it's one of the I think it causes
156
00:13:49,800 --> 00:13:54,640
a lot of anxiety and not just in early stage professionals people that have been established
157
00:13:54,640 --> 00:14:00,520
in their career because the pace of change it feels like everything is speeding up and
158
00:14:00,520 --> 00:14:06,080
I've seen and heard people struggle a little bit with it they feel like if they can just
159
00:14:06,080 --> 00:14:10,600
read a few more articles attend a few more conferences read up on all of those things
160
00:14:10,600 --> 00:14:17,280
they've saved in their inbox all of the different LinkedIn posts etc then they'll finally arrive
161
00:14:17,280 --> 00:14:24,440
and know what they need to do for their profession and that is just it is wishful thinking because
162
00:14:24,440 --> 00:14:30,280
you're never going to know everything certainly not in the early stage of your career but
163
00:14:30,280 --> 00:14:36,760
also not when as your career progresses there's too much information in the world so I think
164
00:14:36,760 --> 00:14:41,480
the key here is knowing what to prioritize and it's quite liberating when you take that
165
00:14:41,480 --> 00:14:45,400
approach because it was never the case that you can know everything in a domain and I
166
00:14:45,400 --> 00:14:50,000
think increasingly now we're just seeing how unrealistic that is which is in itself sort
167
00:14:50,000 --> 00:14:55,080
of agency producing because you get to pick what you're going to focus on you get to pick
168
00:14:55,080 --> 00:14:59,680
and prioritize and you you're never going to get across like even though I spend all
169
00:14:59,680 --> 00:15:05,560
day every day pretty much reading about AI it's a snippet of all the information that's
170
00:15:05,560 --> 00:15:11,840
out there so I think you need to get good at realizing where you can find information
171
00:15:11,840 --> 00:15:16,240
when you need it and what to prioritize and I think those are really the only decisions
172
00:15:16,240 --> 00:15:23,440
you can make yeah I agree I think also with the fields of AI or some unprecedented areas
173
00:15:23,440 --> 00:15:28,520
we don't have frameworks in some jurisdictions so even if you try to find legal frameworks
174
00:15:28,520 --> 00:15:34,280
you don't have you have to make analysis and try to compare with existing frameworks or
175
00:15:34,280 --> 00:15:40,080
look at some best practices internationally I do believe it's very important for legal
176
00:15:40,080 --> 00:15:44,840
professionals to look into the technology aspect and to get some training in the business
177
00:15:44,840 --> 00:15:49,400
field for example for people who start their own business they need to get training in
178
00:15:49,400 --> 00:15:57,160
the marketing in accounting in how to build clients and all these aspects and that doesn't
179
00:15:57,160 --> 00:16:02,280
happen for people who work under a law firm they just focus on the legal work and meeting
180
00:16:02,280 --> 00:16:08,760
the clients so I do believe it depends on varies from an individual to an individual
181
00:16:08,760 --> 00:16:13,720
what's important is to know where to get the information from and what are the available
182
00:16:13,720 --> 00:16:21,040
resources in your area yes I think that's spot on and you know with the increased prevalence
183
00:16:21,040 --> 00:16:28,120
of people using AI to generate content that sounds plausible I think one of the key skills
184
00:16:28,120 --> 00:16:35,840
will also be to be able to tell quality from AI generated content and that is actually
185
00:16:35,840 --> 00:16:42,360
slightly trickier than it sounds because what platforms like ChadGPD what they produce it
186
00:16:42,360 --> 00:16:48,320
sounds credible and unless the domain you might not spot the errors so I think one of
187
00:16:48,320 --> 00:16:54,400
the skills is coming back to basics what we were taught at university right references
188
00:16:54,400 --> 00:17:01,680
checking quality checking etc I think that's also quite important so knowing where to look
189
00:17:01,680 --> 00:17:08,720
who to trust that will get you I think quite far down the line interesting that would lead
190
00:17:08,720 --> 00:17:15,280
us to another to the second topic of this episode talking about AI I know we are in
191
00:17:15,280 --> 00:17:22,200
the world of AI now as a legal professional should I rely on it should I look into some
192
00:17:22,200 --> 00:17:30,360
for example models that have been trained by legal professionals what's the fine line
193
00:17:30,360 --> 00:17:35,120
between using it or some people I know people they don't rely or don't believe they should
194
00:17:35,120 --> 00:17:41,840
use it I use it a lot I like it I got all of them I've got Claude I got perplexity
195
00:17:41,840 --> 00:17:49,160
ChadGPD Notebook LM I'm constantly sort of testing them and checking I'll put the same
196
00:17:49,160 --> 00:17:53,920
question in all of them to see what kind of responses I get and to get a sense of what
197
00:17:53,920 --> 00:18:01,040
they're good at what they're not so good at and I think I hate looking at a blank screen
198
00:18:01,040 --> 00:18:05,360
when I'm writing a report or anything I'd much rather have something and then change
199
00:18:05,360 --> 00:18:10,360
the whole thing I for me that's just easy and the easier way of working and you can
200
00:18:10,360 --> 00:18:16,520
always find snippets in it I think the key or the risk here is knowing what you can and
201
00:18:16,520 --> 00:18:22,320
what you can't put in there so don't put confidential information don't put personal data I didn't
202
00:18:22,320 --> 00:18:27,720
even put identifiers but I do use it for research quite a lot but I guess the difference here
203
00:18:27,720 --> 00:18:33,720
is when it gives me something that is incorrect I feel like I spot it immediately so some
204
00:18:33,720 --> 00:18:40,720
of the safest ways of using AI is actually by domain level experts because they immediately
205
00:18:40,720 --> 00:18:45,600
see what's wrong even if it sounds plausible it's not going to work because there's other
206
00:18:45,600 --> 00:18:51,240
thing that AI didn't consider so I mean I would use it at Google have it help you with
207
00:18:51,240 --> 00:18:56,560
research have it formulate answers but then check and double check everything I think
208
00:18:56,560 --> 00:19:02,720
that's a sensible thing to do and speak to more experienced colleagues.
209
00:19:02,720 --> 00:19:10,040
I do think it's using AI is similar to using Google it reminds me of early years of when
210
00:19:10,040 --> 00:19:17,920
Google came into place it's the same level the same method however over relying on AI
211
00:19:17,920 --> 00:19:24,240
without putting the efforts into writing into refining putting your writing style is problematic
212
00:19:24,240 --> 00:19:33,000
I like as you said you may use it as an administrative assistant that helps with outlining reframing
213
00:19:33,000 --> 00:19:39,000
putting some feedback revising reviewing although models differ and some of them are better
214
00:19:39,000 --> 00:19:47,720
than others I'm aware that LexisNexis is working on an AI model for the legal profession and
215
00:19:47,720 --> 00:19:53,600
we'll see more models in the future designed specifically for lawyers and we may find something
216
00:19:53,600 --> 00:19:59,840
very powerful and for example they may have access to the database and they can provide
217
00:19:59,840 --> 00:20:04,720
authentic and reliable answers and it would be very helpful for me when I research in
218
00:20:04,720 --> 00:20:12,280
AI I always like to ask for quotes from the file so I go to verify that information myself.
219
00:20:12,280 --> 00:20:17,800
I think people develop different shorthands and ways of working with AI but I would say
220
00:20:17,800 --> 00:20:23,120
it is definitely something to test and try even if you don't rely on the responses even
221
00:20:23,120 --> 00:20:28,600
though you feel like it's not super reliable I think by just having a bit of a play around
222
00:20:28,600 --> 00:20:34,520
with it you get a sense of what's possible and they do get better so the models that
223
00:20:34,520 --> 00:20:40,520
I've been using like comparing them to six months ago or a year ago it's a different
224
00:20:40,520 --> 00:20:46,400
proposition so I think just getting that feel for how you work with them and there's a whole
225
00:20:46,400 --> 00:20:51,040
sort of industry around prompt engineering and how you get it to actually give you quality
226
00:20:51,040 --> 00:20:56,360
responses I don't think it would be wasted time spending a little bit of effort just
227
00:20:56,360 --> 00:21:01,800
understanding what is a good prompt and how do you get the kind of resource you want.
228
00:21:01,800 --> 00:21:05,520
And have a backup plan because sometimes there might be outage or some issues so you need
229
00:21:05,520 --> 00:21:10,720
to have your own precedence and that goes to the point of over relying on it.
230
00:21:10,720 --> 00:21:17,960
I know now people talk about AI and compliance especially from an organization perspective
231
00:21:17,960 --> 00:21:26,080
how do you think organizations should approach compliance using AI or utilizing AI tools?
232
00:21:26,080 --> 00:21:31,920
So I think there's probably two different questions sort of snowballed or rolled into
233
00:21:31,920 --> 00:21:36,520
one there's the question of what you need to do from a compliance perspective to manage
234
00:21:36,520 --> 00:21:41,880
organizational AI so this is AI that's across the business right and then there is the question
235
00:21:41,880 --> 00:21:46,280
of how do you use AI to support a compliance team.
236
00:21:46,280 --> 00:21:53,360
So slightly different but you know they go hand in hand sometimes not always was there
237
00:21:53,360 --> 00:21:58,920
a particular angle you wanted to look at out of those two or?
238
00:21:58,920 --> 00:22:04,480
Yeah these two how to the main uses for AI and also how to utilize it for compliance
239
00:22:04,480 --> 00:22:05,480
purpose.
240
00:22:05,480 --> 00:22:11,040
Yeah so I think it all depends on the kind of organization you're working with is it
241
00:22:11,040 --> 00:22:18,520
early stage is it larger more sophisticated is it quite agile so there's very rarely
242
00:22:18,520 --> 00:22:24,840
a one size fits all but understanding I think where the business is at in terms of technical
243
00:22:24,840 --> 00:22:33,200
capabilities appetite and what they're generally doing so most so I work with AI developers
244
00:22:33,200 --> 00:22:37,200
I work with very sophisticated clients but I also work with clients that are just sort
245
00:22:37,200 --> 00:22:43,680
of testing stuff so I think having the right kind of infrastructure one of the GCs I work
246
00:22:43,680 --> 00:22:49,280
with he always talks about right sizing everything so you don't want to go overboard but you
247
00:22:49,280 --> 00:22:54,440
also don't want to be exposed to too many risks so if you're at a company that's just
248
00:22:54,440 --> 00:23:00,320
sort of exploring what you want to have is infrastructure for safe pilots and innovations
249
00:23:00,320 --> 00:23:05,720
without going over the top right but if you are a more sophisticated place you probably
250
00:23:05,720 --> 00:23:10,640
want a bit of a governance structure around it so that's when we get into the ramus of
251
00:23:10,640 --> 00:23:15,800
responsible AI AI governance it has many many different names and that's more about the
252
00:23:15,800 --> 00:23:21,800
corporate sort of the management system or the corporate structure that supports a more
253
00:23:21,800 --> 00:23:28,400
wide scale use and adoption of AI so we can dig into that for sure but just to touch briefly
254
00:23:28,400 --> 00:23:34,560
on the other use case so I feel like maybe this is my perspective but I feel like compliance
255
00:23:34,560 --> 00:23:39,720
teams and legal teams are just being asked to do more and more and more so I had just
256
00:23:39,720 --> 00:23:46,280
before Christmas 20 different risk assessments for copilot or something like that there's
257
00:23:46,280 --> 00:23:52,280
an increased volume and I think not right now perhaps but over the next year I think
258
00:23:52,280 --> 00:24:00,640
there'll be increasing use cases for exploring how you deploy AI within a compliance setting
259
00:24:00,640 --> 00:24:06,280
and there's some you know pretty low risk lower hanging fruit here you could use it
260
00:24:06,280 --> 00:24:11,520
companies have teams or slack or some kind of internal communication channel one use
261
00:24:11,520 --> 00:24:17,160
case that I know lots of companies have explored and are using is essentially training a chatbot
262
00:24:17,160 --> 00:24:22,000
on your corporate compliance procedures and policies it's usually not super sensitive
263
00:24:22,000 --> 00:24:28,040
it I can't tell you how many times I get pinged just to help people find a policy or explain
264
00:24:28,040 --> 00:24:34,280
a policy so it could take off a bit of the the demand of compliance teams if you had
265
00:24:34,280 --> 00:24:39,560
some kind of automated feature that will answer general questions about where's our policy
266
00:24:39,560 --> 00:24:45,280
what's our limits on the X etc etc and I think I mean I definitely know of different companies
267
00:24:45,280 --> 00:24:52,280
are doing that already and I feel like that's a good way of checking or testing out different
268
00:24:52,280 --> 00:24:58,160
processes but then as you get more sophisticated you know there's much potentially much bigger
269
00:24:58,160 --> 00:25:04,240
tools you can automate aspects of your records of processing activities you can meeting those
270
00:25:04,240 --> 00:25:09,320
you can do you can do all kinds of different things with it really yeah absolutely AI is
271
00:25:09,320 --> 00:25:15,360
on the fields one of the main areas is managing the organization's operations and the works
272
00:25:15,360 --> 00:25:21,720
that organization works and for example I've seen like transcribing meetings and putting
273
00:25:21,720 --> 00:25:29,160
agendas or next actions after the meeting I mean there are many uses for AI and the
274
00:25:29,160 --> 00:25:34,640
compliance field risk assessment although I probably you can't get the verdict or the
275
00:25:34,640 --> 00:25:41,960
overall judgment from AI but figuring out the potential risk in specific area is one
276
00:25:41,960 --> 00:25:50,560
use of AI there are multiple angles in looking at AI but what excites you the most about
277
00:25:50,560 --> 00:25:52,720
AI advancements and compliance
278
00:25:52,720 --> 00:26:03,440
what excites me the most I think it's just it's interesting novel ways of working I think
279
00:26:03,440 --> 00:26:08,720
from a personal perspective because of the work I do which is largely focused around
280
00:26:08,720 --> 00:26:15,600
AI literacy giving the building the infrastructure I feel like I'm learning a lot so with AI
281
00:26:15,600 --> 00:26:20,880
literacy there's aspects of it that's AI strategy based so sort of where's the direction of
282
00:26:20,880 --> 00:26:26,680
the company going there's the general compliance and risk management and legal aspects to it
283
00:26:26,680 --> 00:26:31,840
there's the upskilling and learning I feel like for the first time people in my career
284
00:26:31,840 --> 00:26:36,520
people are coming to me for corporate training materials and normally it's the case that
285
00:26:36,520 --> 00:26:40,440
your legal and compliance we have to chase people to say you have to finish this training
286
00:26:40,440 --> 00:26:46,840
module you know sit down do this one hour versus now it's almost the case that the business
287
00:26:46,840 --> 00:26:52,440
is sort of they want to get access to tools they want to understand how it impacts their
288
00:26:52,440 --> 00:26:57,080
work they want to try things out so increasingly I'm seeing the business coming to legal and
289
00:26:57,080 --> 00:27:01,600
compliance saying can we get training on X can we get training on all of these different
290
00:27:01,600 --> 00:27:08,080
aspects and that's really cool because you have a an opportunity here to upskill people
291
00:27:08,080 --> 00:27:13,320
with real world skills that they need for their job and that they see a personal value
292
00:27:13,320 --> 00:27:19,560
in at the same time as you can weave in some of the threads around the legal risks so helping
293
00:27:19,560 --> 00:27:25,640
them understand contextually what are the risks in the roles that I'm doing or with
294
00:27:25,640 --> 00:27:30,920
the kind of systems I'm working on or with the decisions I'm making so you can sort of
295
00:27:30,920 --> 00:27:35,840
integrate it in a much more natural way which I think is fascinating so I'm doing a lot
296
00:27:35,840 --> 00:27:41,920
of exploring around what AI literacy looks like for different companies and also how
297
00:27:41,920 --> 00:27:48,840
you deliver it because corporate training like the once a year tick box exercise is
298
00:27:48,840 --> 00:27:53,840
fundamentally a defense for the company so we've given the training we've done a bit
299
00:27:53,840 --> 00:27:58,600
okay they didn't follow it that's a problem not really but you guys you know what I mean
300
00:27:58,600 --> 00:28:02,760
right but if you're talking about upskilling and getting people to actually start using
301
00:28:02,760 --> 00:28:08,360
different tools in new ways that's not a theoretical piece of knowledge that's an applied piece
302
00:28:08,360 --> 00:28:14,160
of knowledge so looking at how you change training and learning in the corporate environment
303
00:28:14,160 --> 00:28:20,560
I think is also a very fascinating topic I think even with document management which
304
00:28:20,560 --> 00:28:29,040
is a very important field with drafting policies contracts AI can help with changing the frameworks
305
00:28:29,040 --> 00:28:35,960
and looking at some improvement improvement areas where we can develop or even with drafting
306
00:28:35,960 --> 00:28:40,220
contracts you look for a clause you want to change something you can look at previous
307
00:28:40,220 --> 00:28:47,680
works this is an interesting field that AI could come in place well I tried AI I tested
308
00:28:47,680 --> 00:28:54,840
AI with risk management and I find some of the information provided very useful in figuring
309
00:28:54,840 --> 00:28:59,760
out what are the potential risk and how to address these risks although the final judgment
310
00:28:59,760 --> 00:29:08,000
should be by the risk assessor I know AI is trained by humans so there are many challenges
311
00:29:08,000 --> 00:29:13,660
and at the same time we've seen some of the opportunities as we discussed earlier however
312
00:29:13,660 --> 00:29:20,640
what do you think or what are your thoughts on potential risk of using or introducing
313
00:29:20,640 --> 00:29:28,160
AI to the legal field how should senior legal professionals or lawyers address these risks
314
00:29:28,160 --> 00:29:38,400
by for example working side by side with programmers yeah so we have a concept of guardrails so
315
00:29:38,400 --> 00:29:44,400
when you're doing something with AI you try to limit the potential downside and the risk
316
00:29:44,400 --> 00:29:51,760
aspect of something going wrong so I think as long as you have a again you have to know
317
00:29:51,760 --> 00:29:59,120
AI is all about context right so if the risk of getting something wrong is going to really
318
00:29:59,120 --> 00:30:05,840
impact a person's health their well-being their livelihood that's a significant decision
319
00:30:05,840 --> 00:30:12,520
and that requires quite considered source it requires quite considered testing etc that's
320
00:30:12,520 --> 00:30:16,840
completely different from somebody that wants to use it in a low risk capacity so say you
321
00:30:16,840 --> 00:30:24,320
want to have an AI system for I don't know for predicting stock shortage for your shop
322
00:30:24,320 --> 00:30:29,240
maybe that's the solution right so it automatically automatically places orders for toothpaste
323
00:30:29,240 --> 00:30:35,360
when it on certain market signals and is adaptive and is smart what's the risk going to be like
324
00:30:35,360 --> 00:30:41,480
realistically so perhaps that it goes haywire and orders a million pieces of tubes of toothpaste
325
00:30:41,480 --> 00:30:46,680
when you didn't when you need it so I think understanding where you sit in that spectrum
326
00:30:46,680 --> 00:30:52,120
of risk will help you really inform what kind of what kind of risk you can take for kind
327
00:30:52,120 --> 00:30:58,600
of you know things need safeguards other things don't so understanding the risk profile of
328
00:30:58,600 --> 00:31:04,840
the project would always be my starting point and then you look at things like okay so this
329
00:31:04,840 --> 00:31:08,880
is the nature of the risk profile for the project what about the risk appetite for the
330
00:31:08,880 --> 00:31:14,600
company what about the jurisdiction you're in are the specific requirements just a little
331
00:31:14,600 --> 00:31:20,240
specify something that we need to achieve or not do right so you sort of you take a
332
00:31:20,240 --> 00:31:26,080
layered approach to it in deciding what you can and cannot test so coming full circle
333
00:31:26,080 --> 00:31:30,600
to your question around well what about lawyers wanted to use it if you want to use it for
334
00:31:30,600 --> 00:31:35,920
a bit of research and test out can it help you to after memo without putting client details
335
00:31:35,920 --> 00:31:41,400
confidential information personal data like that's not particularly high risk that's something
336
00:31:41,400 --> 00:31:45,520
people can test out depending on the size of your firm you might want to do it in a
337
00:31:45,520 --> 00:31:50,040
corporate environment and you know there are restrictions typically on the kind of AI you
338
00:31:50,040 --> 00:31:56,320
can use for your company but you know have a bit of common sense I think and as lawyers
339
00:31:56,320 --> 00:32:02,520
if you understand what the risk domains are so confidential information IP data protection
340
00:32:02,520 --> 00:32:06,520
and sort of trying to build frameworks around obviously giving the wrong advice will be
341
00:32:06,520 --> 00:32:10,600
a major one as well but understand what the risk profile is for your professional what
342
00:32:10,600 --> 00:32:16,280
you're trying to do then you can look at what kind of safe as it means yeah that's
343
00:32:16,280 --> 00:32:22,040
an idiot I think the main thing is probably one way of training as a human being we face
344
00:32:22,040 --> 00:32:27,000
a constant challenge of being neutral I'm not biased how to transfer this to the machine
345
00:32:27,000 --> 00:32:32,360
this is actually can be challenging especially when you ask the machine to provide a direct
346
00:32:32,360 --> 00:32:38,840
related answer that's the main challenge yes I mean even with things like that you can
347
00:32:38,840 --> 00:32:43,400
have quite lengthy conversations with people and I realize you're talking about completely
348
00:32:43,400 --> 00:32:49,400
different things right so bias has a meaning that term means something for lawyers and
349
00:32:49,400 --> 00:32:54,480
we're primarily thinking about discrimination when we talk about bias bias being something
350
00:32:54,480 --> 00:32:59,120
different in statistics it means something different in machine learning it means something
351
00:32:59,120 --> 00:33:04,040
different in the context of engineering right so we're not necessarily talking about the
352
00:33:04,040 --> 00:33:08,920
bad kind of discrimination that lawyers are concerned about it can mean different things
353
00:33:08,920 --> 00:33:15,280
so with terms like that it helps to understand what who you're talking to and what they understand
354
00:33:15,280 --> 00:33:20,880
because not all some bias is necessary for machine learning so otherwise you want you
355
00:33:20,880 --> 00:33:27,080
want a machine so say you're trying to sort origins from bananas you wanted to have a
356
00:33:27,080 --> 00:33:31,680
bias towards having round things labeled as orange that's just a crazy example but you
357
00:33:31,680 --> 00:33:37,040
know what I mean right so it's not always the bad thing that lawyers are thinking about
358
00:33:37,040 --> 00:33:44,600
however in the event that it is there's actually quite a lot of guidance out there that will
359
00:33:44,600 --> 00:33:50,960
help so it's one of those areas that regulators across the board have been quite alive too
360
00:33:50,960 --> 00:33:58,000
so you can look at for instance UK ICO has very detailed guidance around fairness they
361
00:33:58,000 --> 00:34:04,680
call it so slightly broader than despise but looking at a fan is I said there's a lot of
362
00:34:04,680 --> 00:34:08,960
reports that have been written there's increasingly a lot of technical standards that will assess
363
00:34:08,960 --> 00:34:15,280
bias so again I think it's probably the case of having doing some reading understanding
364
00:34:15,280 --> 00:34:20,840
the landscape of what bias is what the risks are what the frameworks are and then knowing
365
00:34:20,840 --> 00:34:25,160
where to find the information depending on what you need to understand but for sure you
366
00:34:25,160 --> 00:34:32,720
know what to your point around it's based on human input yes in the sense that it's
367
00:34:32,720 --> 00:34:39,520
trained on data that is probably biased it's trained on historical data so there's a lot
368
00:34:39,520 --> 00:34:44,840
to consider that some excellent books on the topic as well so I think just getting a handle
369
00:34:44,840 --> 00:34:51,360
on what it is how machine learning or AI works how the language models how they work and
370
00:34:51,360 --> 00:34:57,240
what the nature of that bias looks like in different domains I think that's quite helpful
371
00:34:57,240 --> 00:35:02,360
I know the topic is long and we can cover our we can have ours covering the main we
372
00:35:02,360 --> 00:35:09,660
sure can I just wanted to have an introductory episode with you to discuss potential opportunities
373
00:35:09,660 --> 00:35:16,080
for AI and to discuss some concerns that early career professionals or lawyers would have
374
00:35:16,080 --> 00:35:21,840
in the law I would be very happy to have you again in the show thank you so much for all
375
00:35:21,840 --> 00:35:28,720
of your answers and for the detailed guidance for early career professionals you did a great
376
00:35:28,720 --> 00:35:38,040
job and hopefully we have you again in the show fantastic thank you thank you so much
377
00:35:38,040 --> 00:35:43,520
thank you for tuning in to threaded minds long beyond I hope today's episode offered
378
00:35:43,520 --> 00:35:50,480
valuable insights and inspired new ideas stay part of the conversation by following us on
379
00:35:50,480 --> 00:35:57,280
Instagram and sharing your thoughts we always appreciate hearing from you remember to leave
380
00:35:57,280 --> 00:36:04,000
a review and subscribe on your favorite podcast platform whether it's Spotify or Apple podcast
381
00:36:04,000 --> 00:36:09,280
so you never miss an episode join me next time as we continue exploring the threads
382
00:36:09,280 --> 00:36:17,240
that connect law business and life until then this is so high Zada see you on Friday minds
00:00:00,000 --> 00:00:06,280
Welcome to Threaded Minds, Law and Beyond with Sohaib Zada.
2
00:00:06,280 --> 00:00:12,040
This podcast is your space to explore the intersections of law, business, personal growth
3
00:00:12,040 --> 00:00:13,880
and modern life.
4
00:00:13,880 --> 00:00:20,440
I'm your host Sohaib Zada, a legal expert, educator, practitioner and lifelong learner.
5
00:00:20,440 --> 00:00:25,580
Together we will unravel the threads that shape our world.
6
00:00:25,580 --> 00:00:31,600
From legal principles to career strategies, self-development and the challenges of our
7
00:00:31,600 --> 00:00:33,200
interconnected world.
8
00:00:33,200 --> 00:00:38,320
We dive deep to uncover stories and insights that matter.
9
00:00:38,320 --> 00:00:42,680
Let us connect the dots and discover new perspectives.
10
00:00:42,680 --> 00:00:47,960
This is Threaded Minds where every thread leads to a story and every story sparks a
11
00:00:47,960 --> 00:00:52,640
new idea.
12
00:00:52,640 --> 00:00:54,160
Welcome to Threaded Minds.
13
00:00:54,160 --> 00:00:59,120
In today's episode we have a special exceptional guest that I promised you I'm going to have
14
00:00:59,120 --> 00:01:01,080
in this show.
15
00:01:01,080 --> 00:01:09,720
I have Erika Wernemann-Ruth, a legal consultant with expertise in AI, privacy and compliance.
16
00:01:09,720 --> 00:01:10,720
Welcome Erika.
17
00:01:10,720 --> 00:01:11,720
Thank you.
18
00:01:11,720 --> 00:01:13,960
It's a pleasure to be here.
19
00:01:13,960 --> 00:01:16,880
I'm happy to have you here in the show.
20
00:01:16,880 --> 00:01:25,520
Alright so today we are going to explore Erika's career journey and how her work in consulting
21
00:01:25,520 --> 00:01:27,880
has evolved over the years.
22
00:01:27,880 --> 00:01:32,160
There are very interesting key themes that we will cover in this episode.
23
00:01:32,160 --> 00:01:39,800
For example, what early career lawyers ask, what are the evolving challenges that we have
24
00:01:39,800 --> 00:01:43,640
in privacy and compliance.
25
00:01:43,640 --> 00:01:47,800
Before we begin I would like to start with just a simple question.
26
00:01:47,800 --> 00:01:51,300
I have a question about your career journey.
27
00:01:51,300 --> 00:01:57,360
Could you walk us through your career journey and how your work in privacy and for example
28
00:01:57,360 --> 00:02:00,240
in consulting has evolved over the years?
29
00:02:00,240 --> 00:02:02,200
Yes of course.
30
00:02:02,200 --> 00:02:05,760
So I am going way way back.
31
00:02:05,760 --> 00:02:11,000
I actually studied economics and then I did a conversion degree to law.
32
00:02:11,000 --> 00:02:15,800
I trained and qualified with a city law firm in London.
33
00:02:15,800 --> 00:02:22,280
Originally actually a facilitator so I worked in disputes and then in the run up to GDPR
34
00:02:22,280 --> 00:02:29,280
a lot of my clients were technology clients and I just thought the topic was quite fascinating
35
00:02:29,280 --> 00:02:34,840
so I slowly pivoted across rather than sort of an abrupt change.
36
00:02:34,840 --> 00:02:44,080
I then spent some time in the Middle East, I was out in Dubai for about six years and
37
00:02:44,080 --> 00:02:50,200
in that capacity I supported the wider region so I worked across the Middle East, Africa,
38
00:02:50,200 --> 00:02:56,440
sort of bridging or linking the different threads between Asia sometimes and Europe
39
00:02:56,440 --> 00:02:59,120
and UK which is great fun.
40
00:02:59,120 --> 00:03:06,440
I returned to the UK, kept working in data protection primarily but increasingly also
41
00:03:06,440 --> 00:03:12,440
in artificial intelligence so I spent some time on government boards looking at AI ethics
42
00:03:12,440 --> 00:03:18,800
before the EU AI Act came out and generally I always thought emerging tech was an area
43
00:03:18,800 --> 00:03:26,040
that I found fascinating so early on I looked at cryptos, blockchains, various forms of
44
00:03:26,040 --> 00:03:29,520
AI facial recognition and driverless vehicles.
45
00:03:29,520 --> 00:03:34,560
It seemed like if there was something quite techy I tended to try to get involved from
46
00:03:34,560 --> 00:03:40,400
a data protection perspective and then I so I did that I ended up spending a decade in
47
00:03:40,400 --> 00:03:46,440
private practice I spent some time on remote environments to major global players I worked
48
00:03:46,440 --> 00:03:51,760
with some early stage companies and I think it's such a common story isn't it but I loved
49
00:03:51,760 --> 00:03:57,160
my time at the firm I was there for like I said a decade and quite happily on a partner
50
00:03:57,160 --> 00:04:02,920
so that's what I saw in my future and then had my son and realised actually I could do
51
00:04:02,920 --> 00:04:07,920
it with a different pace of life so I was headhunted I was recruited for a start-up
52
00:04:07,920 --> 00:04:14,360
so I did that for a while it sort of dawned on me relatively early on that start-ups perhaps
53
00:04:14,360 --> 00:04:18,520
don't need someone like me full-time they need me to come in and put the infrastructure
54
00:04:18,520 --> 00:04:26,600
in place so after a bit of time with the start-up I took the leap and I set up my own company
55
00:04:26,600 --> 00:04:32,800
so I've been working as a consultant via my own business for about a year and a half.
56
00:04:32,800 --> 00:04:36,940
That's very nice actually this is fascinating when you look at your resume and you see like
57
00:04:36,940 --> 00:04:43,060
for example you study economics and then you moved into the law I think that's what's important
58
00:04:43,060 --> 00:04:51,000
to combine multiple areas multiple majors and try to come up with a niche area where
59
00:04:51,000 --> 00:04:56,800
the clients need your legal services or what you could offer them and it's actually interesting
60
00:04:56,800 --> 00:05:02,120
I noticed that your new resume worked with EverShed Sutherland and then you moved into
61
00:05:02,120 --> 00:05:06,920
your own practice so probably you have good knowledge about the differences between working
62
00:05:06,920 --> 00:05:13,520
for a law firm and starting your own I would say boutique law firm or your own business
63
00:05:13,520 --> 00:05:17,200
providing legal consultancy and services.
64
00:05:17,200 --> 00:05:22,600
It's interesting worldwide in many jurisdictions nowadays you can actually practice law without
65
00:05:22,600 --> 00:05:30,040
a law degree for example in the UK they opened a route SQE for people to practice law you
66
00:05:30,040 --> 00:05:35,920
just need two years of legal practice and then two exams SQE 1 and 2 and you can be
67
00:05:35,920 --> 00:05:40,960
qualified as a solicitor I'm not sure but I heard about California there is another
68
00:05:40,960 --> 00:05:48,700
route for that so you can see there is a shift in the legal education or the competencies
69
00:05:48,700 --> 00:05:57,760
that law societies or regulators expect in the future legal professionals so we'll see
70
00:05:57,760 --> 00:06:04,760
more updates we may see updates with regards to the required qualifications so it doesn't
71
00:06:04,760 --> 00:06:10,000
need for some people to start with a law degree and even if you start with a law degree you
72
00:06:10,000 --> 00:06:16,680
need to prove that you're competent in that specific area by training and having multiple
73
00:06:16,680 --> 00:06:20,960
experiences as you did in different locations it's very interesting.
74
00:06:20,960 --> 00:06:25,600
I think just sort of building on that for a second I'm actually increasingly working
75
00:06:25,600 --> 00:06:32,040
with people that are not in the legal profession so in particular around AI there are I think
76
00:06:32,040 --> 00:06:37,600
multiple paths into the profession so I'm seeing people coming with a philosophy background
77
00:06:37,600 --> 00:06:42,280
they're looking more towards the ethics aspect obviously in compliance you don't have to
78
00:06:42,280 --> 00:06:47,360
be illegal although many people are you don't have to be a lawyer so you can come from a
79
00:06:47,360 --> 00:06:52,840
compliance background a governance background a risk background so I'm sort of interacting
80
00:06:52,840 --> 00:06:56,640
with people from many different backgrounds.
81
00:06:56,640 --> 00:07:02,440
I know especially with AI I'm doing my thesis in crypto trading platforms and the interconnection
82
00:07:02,440 --> 00:07:08,960
with securities compliance and I find out with these new fields we need experience people
83
00:07:08,960 --> 00:07:17,240
from different backgrounds whether it's law ethics social science economics but the problem
84
00:07:17,240 --> 00:07:22,720
the people from public law or human rights or some areas they don't they don't understand
85
00:07:22,720 --> 00:07:28,360
the terms or our specialty so they think we are creating something innovative or something
86
00:07:28,360 --> 00:07:29,720
that doesn't exist.
87
00:07:29,720 --> 00:07:34,160
I find challenging I find it challenging talking about my field sometimes.
88
00:07:34,160 --> 00:07:35,160
Interesting.
89
00:07:35,160 --> 00:07:44,020
All right I know that you often get approached by students or early career lawyers or aspiring
90
00:07:44,020 --> 00:07:52,640
lawyers or solicitors whether it's on LinkedIn on social media so what are the main if you
91
00:07:52,640 --> 00:07:57,880
can summarize for us like the main questions that those people ask with that with regards
92
00:07:57,880 --> 00:08:04,280
to their profession to their practice in the future.
93
00:08:04,280 --> 00:08:10,320
I think that's broadly speaking two different themes there are people are asking about particular
94
00:08:10,320 --> 00:08:18,280
skills or what they can do to stand out certifications generally around upskilling and I guess being
95
00:08:18,280 --> 00:08:24,400
an attractive either employee or somebody who could perhaps make a change into the profession
96
00:08:24,400 --> 00:08:30,480
from a different practice area and I think the other section of queries I tend to get
97
00:08:30,480 --> 00:08:36,000
is more around I don't know like a crystal ball like what do I think about the future
98
00:08:36,000 --> 00:08:41,640
is it worth going into this area or that area what are the risks are the profession going
99
00:08:41,640 --> 00:08:45,680
to is the profession going to change are the roles going to be very different.
100
00:08:45,680 --> 00:08:52,120
So I guess those are the two sort of broad thematic issues people tend to raise out about
101
00:08:52,120 --> 00:08:55,280
and I try to respond as much as I can.
102
00:08:55,280 --> 00:08:59,040
What do you typically respond to them what advice you offer to them.
103
00:08:59,040 --> 00:09:05,880
It's a good question right so I think it depends people when people look at my career and they
104
00:09:05,880 --> 00:09:10,920
a lot of people assume there were some strategic decisions I made along the way I didn't I
105
00:09:10,920 --> 00:09:16,200
was just along for the ride I made different decisions based on what felt right at the
106
00:09:16,200 --> 00:09:21,640
moment there wasn't some overarching plan to my career and if I had to say anything
107
00:09:21,640 --> 00:09:27,000
I think your early career is for testing different things and seeing what you enjoy what you're
108
00:09:27,000 --> 00:09:32,720
good at I think people try to plan 10 15 20 years and advance and that's just not how
109
00:09:32,720 --> 00:09:38,920
careers unfold sometimes I've stayed in position longer positions longer than I intended to
110
00:09:38,920 --> 00:09:44,000
because I enjoy the people sometimes I just put my hand off on randomly got sent off on
111
00:09:44,000 --> 00:09:46,360
an assignment to discover that I really enjoyed it.
112
00:09:46,360 --> 00:09:52,120
So I think you can try you can try to pick up new skills you can certainly be a little
113
00:09:52,120 --> 00:09:59,240
bit sensible about it but I can think you can really plan a 20 30 year career as an
114
00:09:59,240 --> 00:10:04,600
early stage professional to find what you enjoy what you're good at and in terms of
115
00:10:04,600 --> 00:10:08,680
the skills I think that now you mentioned some of the different areas right increasingly
116
00:10:08,680 --> 00:10:14,160
I'm seeing roles open up in legal tech companies I think that would be a fascinating space
117
00:10:14,160 --> 00:10:20,400
to be in I think going straight in house will give you a different kind of skill sets that
118
00:10:20,400 --> 00:10:24,760
is increasingly I think into a bond in private practice as well if you can sort of bridge
119
00:10:24,760 --> 00:10:31,880
the gap between the technology and law in terms of the long term view nobody knows and
120
00:10:31,880 --> 00:10:36,720
if they tell you they know
121
00:10:36,720 --> 00:10:43,880
you see trends and you can sort of you can make some good informed guesses about how
122
00:10:43,880 --> 00:10:49,680
things are going to unfold but if I had to make predictions five years ago I don't think
123
00:10:49,680 --> 00:10:54,960
anything would have been really spot on so I think it's rather than trying to future
124
00:10:54,960 --> 00:10:59,560
proof your career by taking certain steps again think it's better to find out what you're
125
00:10:59,560 --> 00:11:04,720
good at and what you enjoy and sort of try to follow those paths.
126
00:11:04,720 --> 00:11:09,920
I do believe actually this is a great idea I do believe you need to start getting the
127
00:11:09,920 --> 00:11:15,400
training the hand on approach to see well where you find your passion because I remember
128
00:11:15,400 --> 00:11:20,400
myself when I graduated from law school I didn't have that passion for technology law
129
00:11:20,400 --> 00:11:25,600
and securities law and then I found this path so some people may take years may take months
130
00:11:25,600 --> 00:11:31,520
to find out where they can where they can find their passion or there they want to practice
131
00:11:31,520 --> 00:11:39,880
and most importantly I think we won't see as much as in the past with regards to general
132
00:11:39,880 --> 00:11:45,680
practices like in the morning family law in the afternoon criminal law and in the evening
133
00:11:45,680 --> 00:11:50,600
one case in for example commercial law or contracts I do think we'll see more specialized
134
00:11:50,600 --> 00:11:56,000
people more niche areas and I do believe that's what the legal profession requires at the
135
00:11:56,000 --> 00:11:57,000
moment.
136
00:11:57,000 --> 00:12:03,160
Yeah I mean I guess there's different views on that as well I think you're probably right
137
00:12:03,160 --> 00:12:08,840
and sadly from this sort of city perspective that I've always worked in that's always been
138
00:12:08,840 --> 00:12:15,560
the case but in a lot of scenarios it's also understanding having a broad exposure understanding
139
00:12:15,560 --> 00:12:21,360
different kinds of implications and how they might unfold and then speaking to the passion
140
00:12:21,360 --> 00:12:27,480
point I think this is something I don't really see trainees or younger professionals really
141
00:12:27,480 --> 00:12:33,720
think about as I get along in my career and I work with these really impressive people
142
00:12:33,720 --> 00:12:39,720
a lot of the time they are they might work in an area that to an outsider would seem
143
00:12:39,720 --> 00:12:46,360
quite dull but they enjoy it because they're so good at it so I think there's two sides
144
00:12:46,360 --> 00:12:50,840
to the passion that you can pick an area that you're passionate about and develop skills
145
00:12:50,840 --> 00:12:56,680
in it but I think if you stick in an area as you become more proficient and better I
146
00:12:56,680 --> 00:13:01,680
think you also find passion in those areas so it's sort of the depth of knowledge and
147
00:13:01,680 --> 00:13:07,400
the mastery of a topic makes you passionate about it if that makes sense so I don't think
148
00:13:07,400 --> 00:13:11,920
you have to go around chasing your passions necessarily to have a fulfilling career you
149
00:13:11,920 --> 00:13:16,480
can also just build your understanding and your knowledge I think a lot of people then
150
00:13:16,480 --> 00:13:19,400
become passionate as a result.
151
00:13:19,400 --> 00:13:27,160
Okay and let's say as an early career professional or aspiring lawyer I found my path in intellectual
152
00:13:27,160 --> 00:13:33,680
property or AI or compliance do you expect over the years of your experience do you expect
153
00:13:33,680 --> 00:13:38,560
from that lawyer or professional to know everything about that field or what's required what's
154
00:13:38,560 --> 00:13:42,720
the main skill that lawyer should have?
155
00:13:42,720 --> 00:13:49,800
They know everything they really really don't and I think it's one of the I think it causes
156
00:13:49,800 --> 00:13:54,640
a lot of anxiety and not just in early stage professionals people that have been established
157
00:13:54,640 --> 00:14:00,520
in their career because the pace of change it feels like everything is speeding up and
158
00:14:00,520 --> 00:14:06,080
I've seen and heard people struggle a little bit with it they feel like if they can just
159
00:14:06,080 --> 00:14:10,600
read a few more articles attend a few more conferences read up on all of those things
160
00:14:10,600 --> 00:14:17,280
they've saved in their inbox all of the different LinkedIn posts etc then they'll finally arrive
161
00:14:17,280 --> 00:14:24,440
and know what they need to do for their profession and that is just it is wishful thinking because
162
00:14:24,440 --> 00:14:30,280
you're never going to know everything certainly not in the early stage of your career but
163
00:14:30,280 --> 00:14:36,760
also not when as your career progresses there's too much information in the world so I think
164
00:14:36,760 --> 00:14:41,480
the key here is knowing what to prioritize and it's quite liberating when you take that
165
00:14:41,480 --> 00:14:45,400
approach because it was never the case that you can know everything in a domain and I
166
00:14:45,400 --> 00:14:50,000
think increasingly now we're just seeing how unrealistic that is which is in itself sort
167
00:14:50,000 --> 00:14:55,080
of agency producing because you get to pick what you're going to focus on you get to pick
168
00:14:55,080 --> 00:14:59,680
and prioritize and you you're never going to get across like even though I spend all
169
00:14:59,680 --> 00:15:05,560
day every day pretty much reading about AI it's a snippet of all the information that's
170
00:15:05,560 --> 00:15:11,840
out there so I think you need to get good at realizing where you can find information
171
00:15:11,840 --> 00:15:16,240
when you need it and what to prioritize and I think those are really the only decisions
172
00:15:16,240 --> 00:15:23,440
you can make yeah I agree I think also with the fields of AI or some unprecedented areas
173
00:15:23,440 --> 00:15:28,520
we don't have frameworks in some jurisdictions so even if you try to find legal frameworks
174
00:15:28,520 --> 00:15:34,280
you don't have you have to make analysis and try to compare with existing frameworks or
175
00:15:34,280 --> 00:15:40,080
look at some best practices internationally I do believe it's very important for legal
176
00:15:40,080 --> 00:15:44,840
professionals to look into the technology aspect and to get some training in the business
177
00:15:44,840 --> 00:15:49,400
field for example for people who start their own business they need to get training in
178
00:15:49,400 --> 00:15:57,160
the marketing in accounting in how to build clients and all these aspects and that doesn't
179
00:15:57,160 --> 00:16:02,280
happen for people who work under a law firm they just focus on the legal work and meeting
180
00:16:02,280 --> 00:16:08,760
the clients so I do believe it depends on varies from an individual to an individual
181
00:16:08,760 --> 00:16:13,720
what's important is to know where to get the information from and what are the available
182
00:16:13,720 --> 00:16:21,040
resources in your area yes I think that's spot on and you know with the increased prevalence
183
00:16:21,040 --> 00:16:28,120
of people using AI to generate content that sounds plausible I think one of the key skills
184
00:16:28,120 --> 00:16:35,840
will also be to be able to tell quality from AI generated content and that is actually
185
00:16:35,840 --> 00:16:42,360
slightly trickier than it sounds because what platforms like ChadGPD what they produce it
186
00:16:42,360 --> 00:16:48,320
sounds credible and unless the domain you might not spot the errors so I think one of
187
00:16:48,320 --> 00:16:54,400
the skills is coming back to basics what we were taught at university right references
188
00:16:54,400 --> 00:17:01,680
checking quality checking etc I think that's also quite important so knowing where to look
189
00:17:01,680 --> 00:17:08,720
who to trust that will get you I think quite far down the line interesting that would lead
190
00:17:08,720 --> 00:17:15,280
us to another to the second topic of this episode talking about AI I know we are in
191
00:17:15,280 --> 00:17:22,200
the world of AI now as a legal professional should I rely on it should I look into some
192
00:17:22,200 --> 00:17:30,360
for example models that have been trained by legal professionals what's the fine line
193
00:17:30,360 --> 00:17:35,120
between using it or some people I know people they don't rely or don't believe they should
194
00:17:35,120 --> 00:17:41,840
use it I use it a lot I like it I got all of them I've got Claude I got perplexity
195
00:17:41,840 --> 00:17:49,160
ChadGPD Notebook LM I'm constantly sort of testing them and checking I'll put the same
196
00:17:49,160 --> 00:17:53,920
question in all of them to see what kind of responses I get and to get a sense of what
197
00:17:53,920 --> 00:18:01,040
they're good at what they're not so good at and I think I hate looking at a blank screen
198
00:18:01,040 --> 00:18:05,360
when I'm writing a report or anything I'd much rather have something and then change
199
00:18:05,360 --> 00:18:10,360
the whole thing I for me that's just easy and the easier way of working and you can
200
00:18:10,360 --> 00:18:16,520
always find snippets in it I think the key or the risk here is knowing what you can and
201
00:18:16,520 --> 00:18:22,320
what you can't put in there so don't put confidential information don't put personal data I didn't
202
00:18:22,320 --> 00:18:27,720
even put identifiers but I do use it for research quite a lot but I guess the difference here
203
00:18:27,720 --> 00:18:33,720
is when it gives me something that is incorrect I feel like I spot it immediately so some
204
00:18:33,720 --> 00:18:40,720
of the safest ways of using AI is actually by domain level experts because they immediately
205
00:18:40,720 --> 00:18:45,600
see what's wrong even if it sounds plausible it's not going to work because there's other
206
00:18:45,600 --> 00:18:51,240
thing that AI didn't consider so I mean I would use it at Google have it help you with
207
00:18:51,240 --> 00:18:56,560
research have it formulate answers but then check and double check everything I think
208
00:18:56,560 --> 00:19:02,720
that's a sensible thing to do and speak to more experienced colleagues.
209
00:19:02,720 --> 00:19:10,040
I do think it's using AI is similar to using Google it reminds me of early years of when
210
00:19:10,040 --> 00:19:17,920
Google came into place it's the same level the same method however over relying on AI
211
00:19:17,920 --> 00:19:24,240
without putting the efforts into writing into refining putting your writing style is problematic
212
00:19:24,240 --> 00:19:33,000
I like as you said you may use it as an administrative assistant that helps with outlining reframing
213
00:19:33,000 --> 00:19:39,000
putting some feedback revising reviewing although models differ and some of them are better
214
00:19:39,000 --> 00:19:47,720
than others I'm aware that LexisNexis is working on an AI model for the legal profession and
215
00:19:47,720 --> 00:19:53,600
we'll see more models in the future designed specifically for lawyers and we may find something
216
00:19:53,600 --> 00:19:59,840
very powerful and for example they may have access to the database and they can provide
217
00:19:59,840 --> 00:20:04,720
authentic and reliable answers and it would be very helpful for me when I research in
218
00:20:04,720 --> 00:20:12,280
AI I always like to ask for quotes from the file so I go to verify that information myself.
219
00:20:12,280 --> 00:20:17,800
I think people develop different shorthands and ways of working with AI but I would say
220
00:20:17,800 --> 00:20:23,120
it is definitely something to test and try even if you don't rely on the responses even
221
00:20:23,120 --> 00:20:28,600
though you feel like it's not super reliable I think by just having a bit of a play around
222
00:20:28,600 --> 00:20:34,520
with it you get a sense of what's possible and they do get better so the models that
223
00:20:34,520 --> 00:20:40,520
I've been using like comparing them to six months ago or a year ago it's a different
224
00:20:40,520 --> 00:20:46,400
proposition so I think just getting that feel for how you work with them and there's a whole
225
00:20:46,400 --> 00:20:51,040
sort of industry around prompt engineering and how you get it to actually give you quality
226
00:20:51,040 --> 00:20:56,360
responses I don't think it would be wasted time spending a little bit of effort just
227
00:20:56,360 --> 00:21:01,800
understanding what is a good prompt and how do you get the kind of resource you want.
228
00:21:01,800 --> 00:21:05,520
And have a backup plan because sometimes there might be outage or some issues so you need
229
00:21:05,520 --> 00:21:10,720
to have your own precedence and that goes to the point of over relying on it.
230
00:21:10,720 --> 00:21:17,960
I know now people talk about AI and compliance especially from an organization perspective
231
00:21:17,960 --> 00:21:26,080
how do you think organizations should approach compliance using AI or utilizing AI tools?
232
00:21:26,080 --> 00:21:31,920
So I think there's probably two different questions sort of snowballed or rolled into
233
00:21:31,920 --> 00:21:36,520
one there's the question of what you need to do from a compliance perspective to manage
234
00:21:36,520 --> 00:21:41,880
organizational AI so this is AI that's across the business right and then there is the question
235
00:21:41,880 --> 00:21:46,280
of how do you use AI to support a compliance team.
236
00:21:46,280 --> 00:21:53,360
So slightly different but you know they go hand in hand sometimes not always was there
237
00:21:53,360 --> 00:21:58,920
a particular angle you wanted to look at out of those two or?
238
00:21:58,920 --> 00:22:04,480
Yeah these two how to the main uses for AI and also how to utilize it for compliance
239
00:22:04,480 --> 00:22:05,480
purpose.
240
00:22:05,480 --> 00:22:11,040
Yeah so I think it all depends on the kind of organization you're working with is it
241
00:22:11,040 --> 00:22:18,520
early stage is it larger more sophisticated is it quite agile so there's very rarely
242
00:22:18,520 --> 00:22:24,840
a one size fits all but understanding I think where the business is at in terms of technical
243
00:22:24,840 --> 00:22:33,200
capabilities appetite and what they're generally doing so most so I work with AI developers
244
00:22:33,200 --> 00:22:37,200
I work with very sophisticated clients but I also work with clients that are just sort
245
00:22:37,200 --> 00:22:43,680
of testing stuff so I think having the right kind of infrastructure one of the GCs I work
246
00:22:43,680 --> 00:22:49,280
with he always talks about right sizing everything so you don't want to go overboard but you
247
00:22:49,280 --> 00:22:54,440
also don't want to be exposed to too many risks so if you're at a company that's just
248
00:22:54,440 --> 00:23:00,320
sort of exploring what you want to have is infrastructure for safe pilots and innovations
249
00:23:00,320 --> 00:23:05,720
without going over the top right but if you are a more sophisticated place you probably
250
00:23:05,720 --> 00:23:10,640
want a bit of a governance structure around it so that's when we get into the ramus of
251
00:23:10,640 --> 00:23:15,800
responsible AI AI governance it has many many different names and that's more about the
252
00:23:15,800 --> 00:23:21,800
corporate sort of the management system or the corporate structure that supports a more
253
00:23:21,800 --> 00:23:28,400
wide scale use and adoption of AI so we can dig into that for sure but just to touch briefly
254
00:23:28,400 --> 00:23:34,560
on the other use case so I feel like maybe this is my perspective but I feel like compliance
255
00:23:34,560 --> 00:23:39,720
teams and legal teams are just being asked to do more and more and more so I had just
256
00:23:39,720 --> 00:23:46,280
before Christmas 20 different risk assessments for copilot or something like that there's
257
00:23:46,280 --> 00:23:52,280
an increased volume and I think not right now perhaps but over the next year I think
258
00:23:52,280 --> 00:24:00,640
there'll be increasing use cases for exploring how you deploy AI within a compliance setting
259
00:24:00,640 --> 00:24:06,280
and there's some you know pretty low risk lower hanging fruit here you could use it
260
00:24:06,280 --> 00:24:11,520
companies have teams or slack or some kind of internal communication channel one use
261
00:24:11,520 --> 00:24:17,160
case that I know lots of companies have explored and are using is essentially training a chatbot
262
00:24:17,160 --> 00:24:22,000
on your corporate compliance procedures and policies it's usually not super sensitive
263
00:24:22,000 --> 00:24:28,040
it I can't tell you how many times I get pinged just to help people find a policy or explain
264
00:24:28,040 --> 00:24:34,280
a policy so it could take off a bit of the the demand of compliance teams if you had
265
00:24:34,280 --> 00:24:39,560
some kind of automated feature that will answer general questions about where's our policy
266
00:24:39,560 --> 00:24:45,280
what's our limits on the X etc etc and I think I mean I definitely know of different companies
267
00:24:45,280 --> 00:24:52,280
are doing that already and I feel like that's a good way of checking or testing out different
268
00:24:52,280 --> 00:24:58,160
processes but then as you get more sophisticated you know there's much potentially much bigger
269
00:24:58,160 --> 00:25:04,240
tools you can automate aspects of your records of processing activities you can meeting those
270
00:25:04,240 --> 00:25:09,320
you can do you can do all kinds of different things with it really yeah absolutely AI is
271
00:25:09,320 --> 00:25:15,360
on the fields one of the main areas is managing the organization's operations and the works
272
00:25:15,360 --> 00:25:21,720
that organization works and for example I've seen like transcribing meetings and putting
273
00:25:21,720 --> 00:25:29,160
agendas or next actions after the meeting I mean there are many uses for AI and the
274
00:25:29,160 --> 00:25:34,640
compliance field risk assessment although I probably you can't get the verdict or the
275
00:25:34,640 --> 00:25:41,960
overall judgment from AI but figuring out the potential risk in specific area is one
276
00:25:41,960 --> 00:25:50,560
use of AI there are multiple angles in looking at AI but what excites you the most about
277
00:25:50,560 --> 00:25:52,720
AI advancements and compliance
278
00:25:52,720 --> 00:26:03,440
what excites me the most I think it's just it's interesting novel ways of working I think
279
00:26:03,440 --> 00:26:08,720
from a personal perspective because of the work I do which is largely focused around
280
00:26:08,720 --> 00:26:15,600
AI literacy giving the building the infrastructure I feel like I'm learning a lot so with AI
281
00:26:15,600 --> 00:26:20,880
literacy there's aspects of it that's AI strategy based so sort of where's the direction of
282
00:26:20,880 --> 00:26:26,680
the company going there's the general compliance and risk management and legal aspects to it
283
00:26:26,680 --> 00:26:31,840
there's the upskilling and learning I feel like for the first time people in my career
284
00:26:31,840 --> 00:26:36,520
people are coming to me for corporate training materials and normally it's the case that
285
00:26:36,520 --> 00:26:40,440
your legal and compliance we have to chase people to say you have to finish this training
286
00:26:40,440 --> 00:26:46,840
module you know sit down do this one hour versus now it's almost the case that the business
287
00:26:46,840 --> 00:26:52,440
is sort of they want to get access to tools they want to understand how it impacts their
288
00:26:52,440 --> 00:26:57,080
work they want to try things out so increasingly I'm seeing the business coming to legal and
289
00:26:57,080 --> 00:27:01,600
compliance saying can we get training on X can we get training on all of these different
290
00:27:01,600 --> 00:27:08,080
aspects and that's really cool because you have a an opportunity here to upskill people
291
00:27:08,080 --> 00:27:13,320
with real world skills that they need for their job and that they see a personal value
292
00:27:13,320 --> 00:27:19,560
in at the same time as you can weave in some of the threads around the legal risks so helping
293
00:27:19,560 --> 00:27:25,640
them understand contextually what are the risks in the roles that I'm doing or with
294
00:27:25,640 --> 00:27:30,920
the kind of systems I'm working on or with the decisions I'm making so you can sort of
295
00:27:30,920 --> 00:27:35,840
integrate it in a much more natural way which I think is fascinating so I'm doing a lot
296
00:27:35,840 --> 00:27:41,920
of exploring around what AI literacy looks like for different companies and also how
297
00:27:41,920 --> 00:27:48,840
you deliver it because corporate training like the once a year tick box exercise is
298
00:27:48,840 --> 00:27:53,840
fundamentally a defense for the company so we've given the training we've done a bit
299
00:27:53,840 --> 00:27:58,600
okay they didn't follow it that's a problem not really but you guys you know what I mean
300
00:27:58,600 --> 00:28:02,760
right but if you're talking about upskilling and getting people to actually start using
301
00:28:02,760 --> 00:28:08,360
different tools in new ways that's not a theoretical piece of knowledge that's an applied piece
302
00:28:08,360 --> 00:28:14,160
of knowledge so looking at how you change training and learning in the corporate environment
303
00:28:14,160 --> 00:28:20,560
I think is also a very fascinating topic I think even with document management which
304
00:28:20,560 --> 00:28:29,040
is a very important field with drafting policies contracts AI can help with changing the frameworks
305
00:28:29,040 --> 00:28:35,960
and looking at some improvement improvement areas where we can develop or even with drafting
306
00:28:35,960 --> 00:28:40,220
contracts you look for a clause you want to change something you can look at previous
307
00:28:40,220 --> 00:28:47,680
works this is an interesting field that AI could come in place well I tried AI I tested
308
00:28:47,680 --> 00:28:54,840
AI with risk management and I find some of the information provided very useful in figuring
309
00:28:54,840 --> 00:28:59,760
out what are the potential risk and how to address these risks although the final judgment
310
00:28:59,760 --> 00:29:08,000
should be by the risk assessor I know AI is trained by humans so there are many challenges
311
00:29:08,000 --> 00:29:13,660
and at the same time we've seen some of the opportunities as we discussed earlier however
312
00:29:13,660 --> 00:29:20,640
what do you think or what are your thoughts on potential risk of using or introducing
313
00:29:20,640 --> 00:29:28,160
AI to the legal field how should senior legal professionals or lawyers address these risks
314
00:29:28,160 --> 00:29:38,400
by for example working side by side with programmers yeah so we have a concept of guardrails so
315
00:29:38,400 --> 00:29:44,400
when you're doing something with AI you try to limit the potential downside and the risk
316
00:29:44,400 --> 00:29:51,760
aspect of something going wrong so I think as long as you have a again you have to know
317
00:29:51,760 --> 00:29:59,120
AI is all about context right so if the risk of getting something wrong is going to really
318
00:29:59,120 --> 00:30:05,840
impact a person's health their well-being their livelihood that's a significant decision
319
00:30:05,840 --> 00:30:12,520
and that requires quite considered source it requires quite considered testing etc that's
320
00:30:12,520 --> 00:30:16,840
completely different from somebody that wants to use it in a low risk capacity so say you
321
00:30:16,840 --> 00:30:24,320
want to have an AI system for I don't know for predicting stock shortage for your shop
322
00:30:24,320 --> 00:30:29,240
maybe that's the solution right so it automatically automatically places orders for toothpaste
323
00:30:29,240 --> 00:30:35,360
when it on certain market signals and is adaptive and is smart what's the risk going to be like
324
00:30:35,360 --> 00:30:41,480
realistically so perhaps that it goes haywire and orders a million pieces of tubes of toothpaste
325
00:30:41,480 --> 00:30:46,680
when you didn't when you need it so I think understanding where you sit in that spectrum
326
00:30:46,680 --> 00:30:52,120
of risk will help you really inform what kind of what kind of risk you can take for kind
327
00:30:52,120 --> 00:30:58,600
of you know things need safeguards other things don't so understanding the risk profile of
328
00:30:58,600 --> 00:31:04,840
the project would always be my starting point and then you look at things like okay so this
329
00:31:04,840 --> 00:31:08,880
is the nature of the risk profile for the project what about the risk appetite for the
330
00:31:08,880 --> 00:31:14,600
company what about the jurisdiction you're in are the specific requirements just a little
331
00:31:14,600 --> 00:31:20,240
specify something that we need to achieve or not do right so you sort of you take a
332
00:31:20,240 --> 00:31:26,080
layered approach to it in deciding what you can and cannot test so coming full circle
333
00:31:26,080 --> 00:31:30,600
to your question around well what about lawyers wanted to use it if you want to use it for
334
00:31:30,600 --> 00:31:35,920
a bit of research and test out can it help you to after memo without putting client details
335
00:31:35,920 --> 00:31:41,400
confidential information personal data like that's not particularly high risk that's something
336
00:31:41,400 --> 00:31:45,520
people can test out depending on the size of your firm you might want to do it in a
337
00:31:45,520 --> 00:31:50,040
corporate environment and you know there are restrictions typically on the kind of AI you
338
00:31:50,040 --> 00:31:56,320
can use for your company but you know have a bit of common sense I think and as lawyers
339
00:31:56,320 --> 00:32:02,520
if you understand what the risk domains are so confidential information IP data protection
340
00:32:02,520 --> 00:32:06,520
and sort of trying to build frameworks around obviously giving the wrong advice will be
341
00:32:06,520 --> 00:32:10,600
a major one as well but understand what the risk profile is for your professional what
342
00:32:10,600 --> 00:32:16,280
you're trying to do then you can look at what kind of safe as it means yeah that's
343
00:32:16,280 --> 00:32:22,040
an idiot I think the main thing is probably one way of training as a human being we face
344
00:32:22,040 --> 00:32:27,000
a constant challenge of being neutral I'm not biased how to transfer this to the machine
345
00:32:27,000 --> 00:32:32,360
this is actually can be challenging especially when you ask the machine to provide a direct
346
00:32:32,360 --> 00:32:38,840
related answer that's the main challenge yes I mean even with things like that you can
347
00:32:38,840 --> 00:32:43,400
have quite lengthy conversations with people and I realize you're talking about completely
348
00:32:43,400 --> 00:32:49,400
different things right so bias has a meaning that term means something for lawyers and
349
00:32:49,400 --> 00:32:54,480
we're primarily thinking about discrimination when we talk about bias bias being something
350
00:32:54,480 --> 00:32:59,120
different in statistics it means something different in machine learning it means something
351
00:32:59,120 --> 00:33:04,040
different in the context of engineering right so we're not necessarily talking about the
352
00:33:04,040 --> 00:33:08,920
bad kind of discrimination that lawyers are concerned about it can mean different things
353
00:33:08,920 --> 00:33:15,280
so with terms like that it helps to understand what who you're talking to and what they understand
354
00:33:15,280 --> 00:33:20,880
because not all some bias is necessary for machine learning so otherwise you want you
355
00:33:20,880 --> 00:33:27,080
want a machine so say you're trying to sort origins from bananas you wanted to have a
356
00:33:27,080 --> 00:33:31,680
bias towards having round things labeled as orange that's just a crazy example but you
357
00:33:31,680 --> 00:33:37,040
know what I mean right so it's not always the bad thing that lawyers are thinking about
358
00:33:37,040 --> 00:33:44,600
however in the event that it is there's actually quite a lot of guidance out there that will
359
00:33:44,600 --> 00:33:50,960
help so it's one of those areas that regulators across the board have been quite alive too
360
00:33:50,960 --> 00:33:58,000
so you can look at for instance UK ICO has very detailed guidance around fairness they
361
00:33:58,000 --> 00:34:04,680
call it so slightly broader than despise but looking at a fan is I said there's a lot of
362
00:34:04,680 --> 00:34:08,960
reports that have been written there's increasingly a lot of technical standards that will assess
363
00:34:08,960 --> 00:34:15,280
bias so again I think it's probably the case of having doing some reading understanding
364
00:34:15,280 --> 00:34:20,840
the landscape of what bias is what the risks are what the frameworks are and then knowing
365
00:34:20,840 --> 00:34:25,160
where to find the information depending on what you need to understand but for sure you
366
00:34:25,160 --> 00:34:32,720
know what to your point around it's based on human input yes in the sense that it's
367
00:34:32,720 --> 00:34:39,520
trained on data that is probably biased it's trained on historical data so there's a lot
368
00:34:39,520 --> 00:34:44,840
to consider that some excellent books on the topic as well so I think just getting a handle
369
00:34:44,840 --> 00:34:51,360
on what it is how machine learning or AI works how the language models how they work and
370
00:34:51,360 --> 00:34:57,240
what the nature of that bias looks like in different domains I think that's quite helpful
371
00:34:57,240 --> 00:35:02,360
I know the topic is long and we can cover our we can have ours covering the main we
372
00:35:02,360 --> 00:35:09,660
sure can I just wanted to have an introductory episode with you to discuss potential opportunities
373
00:35:09,660 --> 00:35:16,080
for AI and to discuss some concerns that early career professionals or lawyers would have
374
00:35:16,080 --> 00:35:21,840
in the law I would be very happy to have you again in the show thank you so much for all
375
00:35:21,840 --> 00:35:28,720
of your answers and for the detailed guidance for early career professionals you did a great
376
00:35:28,720 --> 00:35:38,040
job and hopefully we have you again in the show fantastic thank you thank you so much
377
00:35:38,040 --> 00:35:43,520
thank you for tuning in to threaded minds long beyond I hope today's episode offered
378
00:35:43,520 --> 00:35:50,480
valuable insights and inspired new ideas stay part of the conversation by following us on
379
00:35:50,480 --> 00:35:57,280
Instagram and sharing your thoughts we always appreciate hearing from you remember to leave
380
00:35:57,280 --> 00:36:04,000
a review and subscribe on your favorite podcast platform whether it's Spotify or Apple podcast
381
00:36:04,000 --> 00:36:09,280
so you never miss an episode join me next time as we continue exploring the threads
382
00:36:09,280 --> 00:36:17,240
that connect law business and life until then this is so high Zada see you on Friday minds
