Today, we’re explore the fascinating world of generative AI and its potential applications in the business landscape.
Joined by Katie Simmonds, Data and Technology Lawyer at Womble Bond Dickinson, we’ll be delving into the essential do’s and don’ts for successfully implementing generative AI technologies.
Discover the dos: From creating innovative marketing campaigns to enhancing product design and customer experiences, we unveil the myriad opportunities that generative AI brings to businesses.
Navigate the don’ts: Amidst the vast possibilities, pitfalls also exist. Tune in as Katie shares her experiences and insights on the challenges and ethical considerations surrounding generative AI.
hello everyone thank you for tuning in
as always I am Kaye bitman the content
director at she can code and today we
are discussing do’s and don’ts for
deploying generative AI in business
today we’re going to be talking about
the fascinating world of AI and its
potential applications in the business
landscape thankfully I’m joined by the
amazing Katie Simmons data and
Technology lawyer at wble Bond Dickinson
and we’re going to delve into the
essential dos and don’ts for
successfully implementing generative AI
Technologies welcome Katie thank you so
much for joining us thank you for having
me we’re going to kick off we got so
much to talk about but we’re GNA kick
off with a bit of background about
yourself if that’s okay just to set the
scene of course um say you’ve mentioned
I’m a lawyer at wble BN Dickinson I’ve
always been a tech geek I think since I
can remember I’ve always been kind of
fiddling with Technologies um and trying
to understand how they work so I I
specialize in Tech and data work and
have a particular focus on emerging
Technologies including
AI amazing when you say fiddling were
you that young child that kind of took
things apart and your parents were kind
of like what happened to
this sadly I was always good at taking
things apart I’ve never quite extended
to the physical ability to mend things
or put them back together so I think
that’s probably how I’ve ended up in
more of an advisory role as opposed to a
kind of doing role in the tech space
amazing and were were you encouraged
into that role by somebody when would
did were you interested in Tech at
school was that a teacher or a parent
what kind of LED you off in that
direction I think it’s the ability to
create something kind of that
desperation to try and create something
new um I’ve always wanted to do that so
actually kind of innovation and
Technology seem like a good place to
begin and since being at one world one
Dickinson I’ve only really been
encouraged um so you know we’ve recently
set up an emerging Technologies Group
for example and that kind of has the
wholehearted support of Partners around
me so yeah lots of encouragement yeah
amazing yes we always tell on these
podcast it’s so important to fall or not
fall deliberately end up at the right
company that is super encouraging um to
to the people in its business uh and
knows how to retain that Talent um so
that is good to hear and we have so much
to talk about today um we are going to
to kick off with can I ask you to
summarize what in Earth is generative AI
it’s a branch of AI That’s able to
produce novel realistic and seemingly
original content in response to requests
and prompts from users so it includes
text artwork code or audio all of these
things can be produced using generative
AI Technologies now Research into
generative AI has been ongoing for a
number of years but the recent release
of publicly available tools like chat
GPT have really accelerated the
development and Adoption of these
Technologies yeah it’s something that um
we we keep hear it’s one of those
buzzword isn’t it we keep hearing it at
the minute I don’t think a lot of us
understand um what it is what it means
um how it’ll be used in business so I
was pleased um when this podcast topic
come up um because you’re all right it
is it’s it’s everywhere at the moment um
a lot of a sudden understand what it is
and something else that is every aware
at the minute on everybody’s lips is
chat
GPT which a lot of us say wrong um but a
lot of us have heard of it or attempting
to use it at work um how does chat GP
work so it allows users to ask questions
and then it provides responses back um
at a basic level I like to simplify what
it does into three core steps firstly
it’s essentially a web scraper that
reviews vast volumes of data that are
available in the internet secondly it’s
capable of Performing very sophisticated
mathematics using its large language
model now this model essentially
analyzes how statistically probable it
is that certain sentences and words
should go together now finally chat GPT
then uses its Knowledge from steps one
and two to produce almost instant anous
responses to queries now these responses
mimic human speech and writing patterns
and in my opinion there’s no doubt that
chat GPT could pass the original touring
test um the concept of which for those
who are slightly less geeky than me um
being to test whether a human conceives
a response given by a computer to be
another human so so while we are a long
way from artificial general intelligence
which invis visages AI systems solving
tasks in ways that aren’t limited to how
they’re trained this combination of
accuracy coupled with the ability to
mirror human speech patterns is
unique yes it’s so interesting you
you’re absolutely right um there was a
funny enough there was a conversation
this morning with a colleague um and I
and we were talking about um using uh
chat uh GPT at work and um I said well I
I I put in when I put in a question
question I tend to say please please can
I have this and I and I said I don’t
know why I do that but it’s almost like
I I feel rude if I don’t ask please that
you know I’m I’m asking you to do
something and to generate a response for
me so I ask please and we did have that
discussion of what would happen if we
weren’t polite you know would would it
come back and and say well that’s rude
or you know would it not like me if I if
I wasn’t poed in the way that I asked
you know and you are right the way that
speaks back to you does obviously sound
um like a human response um and uh yeah
is it’s um incredibly clever but yeah I
only happen to come up this morning do
do I do I speak to it nicely or can I be
rude will it know I’m being rude because
it’s mirroring human speech patterns as
you said being more General um what can
generative AI
do so the opportunities are seemingly
endless we’re seeing business is using
generative AI to write code um during a
recent conversation with a head of
operations of a startup I was told that
they hadn’t had to hire as many coders
as they’d originally planned as
generative AI tools had been so
effective for them another use case is
drug Discovery companies using
generative AI to create new drugs
optimize drug properties and assess the
toxicity in drugs um that for me is a
really really good example of kind of
tech doing good here and as a law firm
we’re looking at a number of use cases
such as summarizing judgments for
seminars staff queries about policies um
on this one the plan is that we’ll have
an area of our kind of internal Hub that
we essentially be able to ask a question
to such as how much adoption leave am I
entitled to or am I entitled to a
sabatical the aim is that this tool will
then scan all of our policies and
provide employees with an answer and
specific reference so that’s quite a
basic um functionality but when you
think about it the amount of kind of
coms that you could get through asking
HR for particular questions the aim is
that generative AI will kind of minimize
that and make it more
efficient um more generally we’re also
looking at areas we can incorporate Tech
into our client delivery so the ability
to provide insights on large volumes of
contracts generative AI would be very
helpful here the drivers for us as a
firm are all aimed at improving
efficiencies and speed of delivery for
us and our clients um one point that I
wanted to highlight is that generative
AI is just one tool so it’s important
not to have tunnel vision and think that
it will always be the answer generative
AI is part of a much wider Suite of
tools that are available including
Automation and rule or template based
systems now these are powerful tools and
also have a wide application I I love
the fact that you pointed out you know
just just going for all of those
different policies for instance in in
your own firm um because that it’s it’s
very open to human error if somebody’s
having to go through and do that um you
know even by eye or if they are using a
piece of software and just doing a quick
find or whatever it may be um they still
a lot for for one person um to take on
so you’re right there there are um you
know the ways that we can use AI at the
moment um uh hopefully are freeing up
people’s time and and and really helping
them there there is a lot of negative
press around it as well and we’re all
trying to navigate what that looks like
and and where we’re going to go but
how do you think it is for for
businesses at the moment I mean you
mentioned quite a few case studies there
that would really help businesses um you
know how beneficial do you think it is
yeah well benefits include accelerated
creation of content so machine learning
Technologies can analyze and learn from
vast data sets producing outputs much
quicker than humans can um this is key
in areas like the pharmaceutical
industry where drug Discovery can take
upwards of a decade another benefit is
the ability to personalize customer
experiences chatbot can can be more
humanlike and more responsive to
customer
queries um a final and obvious benefit
is is money taking the coding example we
know that coders are already using
generative AI to write a lot of their
code this has efficiency advantages as
well as additional technological
capabilities we also know that the UK
government’s backing Ai and has a big
10-year plan to make the UK an AI
superpower
I I love that that um example that you
picked out there about personalizing
customer experiences that’s I mean
businesses you know try and do that
every single day but for for AI just to
be able to point that out and to really
tailor something to to a customer um
that’s going to be invaluable um yeah
production marketing content you know it
can be produced really quickly for
example um content that would take teams
kind of several hours they’re finding
that chat GPT can have a really good
first stab I’m not condoning using chat
GT in this way um but it’s something
that businesses are just starting to do
that I’m seeing a lot of our clients
kind of experimenting with and then now
thinking well what what what are the
risks what are the legal kind of
challenges around using it in this way
yeah and you saying now having a first
step as well even if you chopck
something and that’s what we’ve been
doing we’ve been using it for ideas to
generate ideas and then you know
obviously we take that back tweak
whatever we need build out from that but
even just to have a first stab at
something and think you know what we we
wouldn’t have thought of that that’s a
great idea um so yeah is almost kind of
using it to start with and then a human
stepping in at the moment and as you
said all those things are running
through your head as well um you know
using something like that um as part of
your business because it as we mentioned
it does have its benefits um but also
surely it must also come with its risks
right
yeah I think it’s important to look at
the limitations and risks around using
these Technologies which makes them more
controversial and definitely adds to the
hype um key challenges that can arise
from using generative AI tools include
verifying the accuracy of the underlying
data as ultimately algorithms are only
as good as the data sets they’re trained
on now taking chat GPT as an example
it’s an excellent web scraper but it
cannot challenge the accuracy of the
underlying sources that it scrapes the
data from it can be particularly
challenging to identify these kind of
Errors where chat GPT has Blended fact
fiction and inaccurate data all together
now this dovetails into the risk of
discrimination it it will be more
difficult for organizations to check
whether generative AI is perpetuating
historic biases and discrimination into
its
responses a classic example of
discrimination involving AI was an
advertising algorithm that was showing
more AI jobs to men and secretarial jobs
to women now the difficulty here was
that the algorithm was actually working
exactly as it should be as more women
were clicking on the secretarial jobs
and vice versa and this is a really good
example of how it’s not easy to identify
discrimin a and issues particularly
where for all intents and purposes the
AI Tech is working exactly as it should
be so there’s really a danger of
becoming overly reliant on these
Technologies yeah and where this happens
businesses fail to be able to understand
how they’re using people’s personal data
or verify that a response is right I’m
repeatedly asked well how do I check it
how do I test it how do I rely on it and
and the answer is you can’t you know
straightforward you can’t just rely on a
black and white response it takes you as
a business to look at it in a wider
context so as well as causing privacy
and discrimination law issues It
ultimately could cause broader
reputational problems as well as
challenges with ascertaining the best
ways to maximize how you’re using these
Technologies um you can’t work out
strategy for kind of mid to long term if
you can’t get the short term right and
now is probably a good time to highlight
that even open ai’s own website states
that chat GPT sometimes writes plausible
sounding but incorrect or nonsensical
answers so this really emphasizes that
where an organization is using these
Technologies it’s vital that someone has
oversight of how it works um and there’s
some some sort of verification of the
output being
produced yes definitely and yes you
mentioned their reputational issues and
and whether or not is accurate because
um as a as a journalist it can be
alarming the types of things that it
spits out because you’re trained to
obviously um you know have credible
sources only pick things from places
where you know um that that they have
that is a credible source and if you’re
not sure you go and track things down
yourself you always have to be able to
to work things back to a source that you
can rely on um and you you can put
things in in uh uh chat uh gbt and ask
it to generate you like a listicle or
something like that and I remember doing
it and thinking how do I know if that is
even accurate and how do I trace that
back to the source because you can’t it
just goes against everything that any
credible publication would have any
writer create so it got me quite
alarming um and as you said it’s kind of
it’s at the minute I think for a lot of
people it works great as an idea Factory
or something that can you know kind of
kick off something um spark an idea and
then you know you go down your usual
route of um making sure that you’re
using credible sources um but you are
right because it can have such uh
problems with company’s reputation you
know if you put something out and uh and
it’s wrong um what on Earth what on
Earth do you do at that point um and on
that should companies be worried um
about uh regulatory issues or security
concerns you haven’t even covered
security concerns at the
minute yes so taking the general data
protection regulation or gdpr as an
example here the gdpr places an
obligation on data controllers to be
transparent and to be able to explain
how an AI system operates
AI is now very much in the spotlight
with new regulation and laws being
developed globally so we expect as this
focus on AI continues we’ll see an
increasing number of individuals looking
to organizations to better explain how
AI is being used to make decisions about
them um this could be challenging in the
context of generative AI
Technologies generative AI also raises
complex questions regarding the
ownership authorship and accountability
of the generated code and content so for
instance who owns the IP rights of the
coach who’s responsible for any errors
or damages caused by any generated code
or content how can developers ensure
that they’re not infringing on any
existing copyrights or patents now these
are all questions that aren’t
particularly new to generative AI
they’ve been around for as long as
people have been using open- Source
software but it’s something that I think
is very much on a lot of organizations
what Radars who want to kind of maximize
the potential here to use these
Technologies now in terms of security
it’s really a double-edged sh on on the
one hand it can be used as a tool to
bolster cyber security practices
including writing new code quicker to
patch potential
vulnerabilities on the other hand threat
actors or attackers can use these tools
too to improve their hacking techniques
and identify
vulnerabilities so we’re seeing an
increasing number of organizations
creating kind of new senior roles to
deal with the risks and benefits of
using these Technologies a role that
we’ve kind of heard a lot of
organizations talk about is Chief AI
officer um this trend is likely to
continue as we see the new AI regulation
coming into force in the EU alongside
the developments of the UK’s regulatory
position it’s it’s always interesting in
the tech industry how um new jobs come
about and how fast they appear and like
you just said that is a new job um that
you know a few years ago we wouldn’t
have thought of that Chief AI officer
and and all of a sudden that is is a
role another role in technology that um
that we need to fill um but yeah it’s
always interesting how fast the industry
moves and how we have to keep up and um
as you said as well I think every time
somebody talks about hackers I always
think you know how companies tend to say
um they’re always one step ahead of the
curve and they they’re there they’ve
already leared you know how to tap into
the vulnerabilities of um uh Ai and and
what businesses are doing so you have to
um always try and be one step ahead and
I’m sure they’ve already figured it out
um and on that are there as a side note
to that are there ethical considerations
surrounding um generative AI that that
we need to consider as
well yes so we’ve already touched
briefly on discrimination but AI has the
potential to make bias decisions based
on flawed input data or programming um
businesses could also be acquiring
pseudo ethical AI products to satisfy
their Environmental and social
governments targets only to later
discover that the claims made around the
qualities of the design are actually
false it’s a real question of kind of
how can you test what you’re buying here
and how can you verify that the solution
meets all of its descriptions that um
solution providers claim it will do so
in addition companies could be selling
AI for good initiatives while also
selling surveillance technology to
corrupt governments and questionable
corporate customers um this is known as
AI ethics washing that’s a new one AI
ethics washing I hav know that it just
the mind boggles um of the amount of uh
terms that that continue to to be um
invented in the tech industry um it’s
quite comparable to greenwashing when
you think about it so people saying
something is clean tech and and it
perhaps isn’t behind the scenes so um
allowing companies to kind of offload
that that Cen Tech onto their suppliers
and saying that that they’re they their
kind of carbon Footprints lower but
actually it overall the carbon
Footprints exactly the same it’s a bit
like that with ethics washing in terms
of kind of false claims um that could be
made yeah yeah um also there have been
many arguments that AI will lead to job
losses is one of those negative pressing
about I that we keep reading do you
think that’s going to be the case or is
that another one of those in the tech
industry there seems to be every few
years something comes along and then we
say everybody’s going to lose their job
and actually those people just go off
into different directions and then
there’s something new and we’ll say
everyone’s going to lose their job is
that kind of the same as what’s
happening with AI or do you think people
will lose their
jobs you’ve hit the nail and the head
really there so so potentially we could
see job losses but focusing on the
manufacturing sector as an example where
manufacturers are adopting new tech that
means less employees are needed now is a
really good time to ident to upskill
existing employees is there’s a lack of
Engineers and workers in the
manufacturing sector for all of the new
things that are being introduced
including generative AI so so while
there may be less employees needed in
one area of the business there should in
theory be other opportunities for those
employees subject to
upskilling more generally it’s crucial
that where any AI Technologies are being
used there’s some sort of verification
of the output before it’s blindly used
we’ve already touched upon this but that
can create a new subset of roles for
those who can be part of this
verification process the optimis me
would also say that a that generative AI
allows us to do more with the same
Workforce as opposed to staying in a
steady state with a smaller
Workforce we’re still a long way from
Mass job losses due to AI it it’s still
currently only working with existing
data that it can access either on a
database or the internet there’s still
major breakthroughs that have to happen
before we reach anything that resembles
human level AI which is likely if that
if that is um successful that could
result in bigger job losses but but I
don’t think so I’d like to say that the
optimism he wins here and I I like that
that that you said it’s just going to
enable businesses to do more um with the
same Workforce and I think that that is
what we need to come down on on that
side of the bence on that um because you
you’re all right there is always going
to be that negative um uh side to things
and it is down to companies to you know
if they want to retain good talent then
it’s finding other things for them to do
um or just pitching it that we’re doing
more um with the same Workforce uh so I
definitely agree with that um we are
almost out of time but I’ve got one last
question for you we have spoke a little
bit about how fast the industry moves um
particularly how fast even AI um is
moving um so what do you think the
future holds for generative AI a recent
research suggests that chat gbt ‘s
performance is declining over time and
all AI models can suffer from drift
which is essentially the performance of
a model degrading over time whether this
performance will improve be corrected is
yet to be seen and for me these flaws
well this flaw coupled with the other
flaws and the pr privacy and
discrimination risk that we’ve touched
upon today have really emphasize the Gap
in regulation in this space well
globally we’re seeing the Reg
landscape develop and we’re expecting
the UK to host the first Global AI
safety briefing in Autumn this year it’s
always the case that technology is one
step ahead so it’s really with
businesses right now to make sure that
they’re developing these Technologies in
the right way to avoid falling foul of
regulation in future my view is that
we’ll still be talking about generative
AI for years to come even at the end of
the government’s 10-year plan how
positively we’re able to talk about it
will in be will in part be dependent on
the approach taken to regulation
globally would there needing to be a
balanced approach to ensure individuals
rights and freedoms are protected while
simultaneously allowing innovators to
develop and use AI to its full
potential yes and I love that just
focusing on you know make making sure
that things are built in the right way
now um instead of you know something
that we have to fix in the future um and
you’re absolutely right there is so much
more to come uh in this area and time
will only tell whether or not it’s all
positive talk or negative um but uh as
you said I am also an optimist and I
think that um it’s it’s going to be uh
hopefully more positive news um uh with
generative AI um but Katie thank you so
much it is a topic that has been on
everybody’s lips um and is so good to
share what it is and how it’s going to
be beneficial to businesses so thank you
so much for taking the time out to chat
with me today it’s been an absolute
pleasure having you on here not at all
thanks for having me and to everybody
listening as always thank you so much
for joining us and we hope to see you
again next
time
