FEU.INT.001.mp4

Back

Or filter by your interest

Browse all our content

Finovate

Unlock all the content with only {{price}}

Nina Schick.vtt

English

{{00:00:03}}
FEU.INT.001: Hi, I'm David Penn,
research analyst with Finovate.

{{00:00:06}}
I'm here with Nina Schick.

{{00:00:07}}
She is the author, generative AI
expert, and founder of Tamang Ventures.

{{00:00:11}}
Nice to see you.

{{00:00:11}}
Great to be here.

{{00:00:13}}
I enjoyed your keynote earlier today.

{{00:00:14}}
I know you just finished
a Q&A few minutes ago.

{{00:00:16}}
How did that go?

{{00:00:17}}
Great.

{{00:00:17}}
And always when you're talking
about AI, we're only starting

{{00:00:21}}
to scratch the surface.

{{00:00:22}}
So when we go into the Q&A, the
Q&A is actually my favorite part.

{{00:00:27}}
There were so many good questions and I
felt Time limited us, but we could have

{{00:00:32}}
gone on all day, all week, all month.

{{00:00:34}}
I think one of the things when
you mentioned in your keynote, the

{{00:00:36}}
idea that the opinions on AI ranged
from the dangerous and devastating

{{00:00:40}}
to the stupid and who cares?

{{00:00:42}}
You can imagine the range of
questions that might pop up and

{{00:00:44}}
given those sort of extremes.

{{00:00:45}}
So very, interesting.

{{00:00:47}}
Let's start talking about AI and
particular in financial services.

{{00:00:50}}
One of the questions I'm thinking that a
lot of people are curious about are what

{{00:00:54}}
are some of the things that financial
institutions need to know when it comes

{{00:00:57}}
to implementing things like generative AI?

{{00:01:00}}
So I'd say that the most important
component of implementing is making

{{00:01:05}}
sure your data is ready, structured
well, in an ecosystem where the

{{00:01:10}}
security is high, because there's
There is no AI implementation

{{00:01:15}}
or application without the data.

{{00:01:16}}
And in fact, for any knowledge industry,
like financial services or like banking,

{{00:01:21}}
a lot of the value you're going to get in
terms of actual AI applications for your

{{00:01:26}}
specific use case, whether it's to launch
a customer services chatbot or actually

{{00:01:32}}
as a kind of research tool, it's going to
come from that institutional knowledge.

{{00:01:36}}
All that data structured and
unstructured within your organization.

{{00:01:40}}
So making sure that data is ready
to be optimized is going to be key.

{{00:01:46}}
Sure.

{{00:01:47}}
I'm curious about the idea
of de risking adoption.

{{00:01:49}}
I imagine some of what you're talking
about now has to do with that.

{{00:01:52}}
What are some of the pain points
that financial institutions tend

{{00:01:54}}
to encounter when they're making
this implementation progress?

{{00:01:57}}
So obviously, again, one of the
most important ones is making

{{00:02:00}}
sure that your data is secure.

{{00:02:02}}
And there were lots of questions around,
especially when some of these foundational

{{00:02:06}}
models first started to emerge.

{{00:02:08}}
And certain organizations were just
throwing in data into platforms like

{{00:02:14}}
chatGPT and finding out that data wasn't
safe before you start experimenting

{{00:02:19}}
with any models or any partners to
ensure the integrity of your data is

{{00:02:23}}
going to be absolutely fundamental.

{{00:02:26}}
But also, I would say
more broadly when you see.

{{00:02:28}}
The new applications that AI allows,
including on cyber or kind of biometrics

{{00:02:36}}
and identity hacking, you're going to have
to upgrade your systems as well, right?

{{00:02:40}}
Your cyber systems, you're going to have
to ensure that they are AI resilient.

{{00:02:44}}
And part of that might, again, be
working with a cyber partner who is

{{00:02:48}}
developing AI as part of the defense
strategy more broadly to mean that your

{{00:02:56}}
risks are going to go, your surface
area of attack is going to be smaller.

{{00:02:59}}
It's interesting that you
mentioned partnerships.

{{00:03:00}}
One of the things that's interesting to
me as I think about this is even before

{{00:03:04}}
generative AI started to come on the
scene, a lot of financial institutions

{{00:03:06}}
were understanding the power of their
data for personalization, a lot of things.

{{00:03:09}}
And it just feels like with AI, that
whole challenge and opportunity has

{{00:03:13}}
just been turned up to 11, so to speak.

{{00:03:14}}
And I'm just wondering, in terms of
financial institutions being able to It

{{00:03:19}}
seems like a lot of work that they might
not necessarily have the pedigree for.

{{00:03:23}}
So I'm wondering about
that particular challenge.

{{00:03:25}}
Is this almost certainly a
partnership type of thing?

{{00:03:27}}
They're going to have to get help?

{{00:03:28}}
Absolutely.

{{00:03:29}}
And I'd say more broadly,
that's just the line.

{{00:03:33}}
It's the direction in which everything
is moving as we're in this kind of

{{00:03:37}}
digital ecosystem where you need
to have so many different layers of

{{00:03:42}}
expertise, not only domain specific
expertise, but now you're talking about

{{00:03:46}}
AI applications and AI implementations.

{{00:03:49}}
So you need to have that data science
background or machine learning background.

{{00:03:53}}
And what we are starting to see
in the ecosystem is that lots of

{{00:03:57}}
big players or even small players
are going to partner with the tech

{{00:04:01}}
companies or with AI companies.

{{00:04:04}}
I would just say, choose
your partners wisely.

{{00:04:07}}
But in part, the reason why this is
folding unfolding and a partnership kind

{{00:04:14}}
of model is because it simply doesn't
make sense for a single institution to try

{{00:04:21}}
and build up that level of AI capability,
which is ultimately going to be available

{{00:04:27}}
to you as a software update, right?

{{00:04:29}}
Maybe it makes sense if you're Bloomberg.

{{00:04:30}}
So that's what Bloomberg tried to do.

{{00:04:32}}
They were one of the first financial
services institutions that looks at

{{00:04:37}}
building their own foundational models.

{{00:04:39}}
So they build Bloomberg GPT, which
is a large language model, which is

{{00:04:42}}
optimized for financial services based
on all of their historical data, right?

{{00:04:49}}
But even now the coin is still out as to
whether what Bloomberg built can actually

{{00:04:54}}
be better than some of the best in class
frontier models developed by AI specific

{{00:05:00}}
companies like Open AI, like a GPT-4.

{{00:05:03}}
But regardless of that, most organizations
will, it will be that I think what's

{{00:05:09}}
going to happen is that you have these...

{{00:05:11}}
this platform layer, which is being
developed by the AI and tech companies,

{{00:05:15}}
and then other institutions plug
in and optimize that platform for

{{00:05:20}}
their use case with their data.

{{00:05:21}}
But they're still going to
need a lot of services around.

{{00:05:24}}
How do you optimize the data?

{{00:05:26}}
How do we build this?

{{00:05:28}}
So I don't think AI is
going to take away any jobs.

{{00:05:30}}
It's going to cause a lot of new jobs.

{{00:05:32}}
Take place as you implement
this AI, the AI applications

{{00:05:36}}
That makes a lot of sense.

{{00:05:36}}
And it's also interesting.

{{00:05:37}}
You mentioned in Bloomberg, they've been
such a pioneer with data and financial

{{00:05:40}}
services that you could almost imagine
of all of the entities that would

{{00:05:43}}
give it a try to do it on their own.

{{00:05:44}}
Bloomberg would be one of the
ones that would give it a shot.

{{00:05:47}}
And it's interesting to see that
we're still trying to find out

{{00:05:48}}
if they've pulled it off or not.

{{00:05:50}}
Yeah.

{{00:05:51}}
And I, don't use because I don't
work in financial services.

{{00:05:54}}
I don't use their portal, but again,
this is speaks to the larger point I was

{{00:05:59}}
making about these unfolding capabilities.

{{00:06:03}}
They're already here.

{{00:06:05}}
It doesn't matter whether or
not you decide to have an AI

{{00:06:09}}
strategy in your institution.

{{00:06:10}}
And by the way, I think you should,
because of the way that it is so

{{00:06:15}}
quickly transforming all knowledge work.

{{00:06:17}}
So you really need to have a strategy and
look at what your competitors are doing.

{{00:06:22}}
But even if you don't, these capabilities
are going to be wrapped up in the

{{00:06:26}}
hardware and the software, the platforms,
the devices that we all already use.

{{00:06:31}}
And that's happening right now from
your very basic Microsoft Office

{{00:06:34}}
being updated with these co pilot
capabilities to the other end where,

{{00:06:38}}
you know, Bloomberg building their
own large language model for finance.

{{00:06:42}}
They're going to continue to do
things like that because of their

{{00:06:45}}
immense proprietary data set,
which is going to deliver them.

{{00:06:49}}
So much value in terms of delivering
better services to their customers.

{{00:06:54}}
So it is coming, it's here as
a hardware or software update,

{{00:06:58}}
regardless of whether you like it or
not, or are engaging with it or not.

{{00:07:02}}
I like the point that you made in your
keynote about it being a meta technology.

{{00:07:05}}
It's not to think of it as a
screwdriver, as a hammer, but as tools.

{{00:07:08}}
This is a much broader category.

{{00:07:10}}
Let's take a look outside of
financial services just a little bit.

{{00:07:13}}
In terms of some of the things that you
might be seeing in other industries,

{{00:07:15}}
maybe a bit ahead of financial
services, perhaps, what are some

{{00:07:17}}
of the applications of either AI in
general or generative AI that you're

{{00:07:21}}
seeing again outside of financial
service that you think are interesting?

{{00:07:24}}
So again, I think this comes back to this
point about AI being this meta technology.

{{00:07:30}}
And I'm so often asked in what
industry do you see the biggest

{{00:07:34}}
impact or what are the applications?

{{00:07:37}}
And really the answer to that is
how long is a piece of string.

{{00:07:40}}
So rather than one industry
in particular standing out, I

{{00:07:44}}
think the mindset matters more.

{{00:07:45}}
So innovative thinkers have
already been deploying AI in

{{00:07:51}}
really interesting use cases.

{{00:07:53}}
And.

{{00:07:53}}
Some of those include scientific research,
medical research, and some of the most

{{00:07:59}}
interesting thing about the application
of AI in medicine has been, and it comes

{{00:08:03}}
back to this philosophical question about
machine intelligence and is AI going to

{{00:08:08}}
augment us or automate us, but what we've
been seeing in medical diagnostics is that

{{00:08:13}}
actually when an AI system has all the
right data, it can make better informed

{{00:08:20}}
decisions than a clinician on diagnostics.

{{00:08:23}}
So Google had a large language model,
which is just trained on medical data

{{00:08:29}}
to help clinicians make, diagnoses and
what their study found it's called AMIE,

{{00:08:35}}
their LLM, their medical LLM is that the
best performing entity on diagnostics

{{00:08:43}}
was AI diagnosing on its own, and the
second best was the clinician assisted

{{00:08:49}}
by AI but the worst performing was
the clinician acting alone without AI.

{{00:08:55}}
That pretty much tells you.

{{00:08:57}}
Very interesting.

{{00:08:57}}
I know that a lot of folks are curious
about the idea that you mentioned.

{{00:09:00}}
The replacement, the
augmentation and so forth.

{{00:09:02}}
So it's interesting to see that the
augmentation end of it, it still holds up

{{00:09:06}}
as a valued way of using the technology.

{{00:09:09}}
I think the answer to the augment or
automate is going to be both because I,

{{00:09:16}}
anyone who says that there isn't going
to be disruption in the labor market,

{{00:09:22}}
I think is telling you a fib because so
clearly there's going to be disruption.

{{00:09:27}}
It doesn't mean that humans will never
have work and everything is going

{{00:09:31}}
to be done by AI, but the way that
we work is going to be different.

{{00:09:34}}
And those of us who know how to work with
AI systems and also have domain expertise,

{{00:09:40}}
we're going to do the best, right?

{{00:09:41}}
Because similar to the analogy of the
clinician who acted without AI and perform

{{00:09:48}}
so much worse than well, actually, AI was
the best performing, acting even better

{{00:09:53}}
than the clinician was unassisted by a was
a clinician who was assisted by AI right?

{{00:09:58}}
So if we assume that we're still
going to want to have human doctors

{{00:10:03}}
then You optimize outcomes for
yourself by being able to work with

{{00:10:07}}
these systems while still having
deep domain knowledge and expertise.

{{00:10:11}}
Interesting.

{{00:10:12}}
I'm curious maybe by, by finishing
up the looking beyond generative AI

{{00:10:17}}
perhaps into some of either further AI
applications or even other technologies

{{00:10:20}}
that are adjacent that we may not be
knowing of, may not be knowledgeable

{{00:10:23}}
of that you might be aware of.

{{00:10:24}}
I know that you talked in your keynote
about the idea of, Artificial General

{{00:10:28}}
Intelligence, but there being a
preceding step before that you think

{{00:10:31}}
we've already been embarking on.

{{00:10:33}}
Is that sort of where you think the focus
for the next few years might be before

{{00:10:37}}
we move on to some of these other areas?

{{00:10:38}}
Yes.

{{00:10:40}}
I think that before we get to AGI and
we spend so much time discussing and

{{00:10:45}}
debating AGI is what we're already
starting to see happening, right?

{{00:10:50}}
This kind of by update,
software and hardware.

{{00:10:56}}
And before you know it in the next.

{{00:10:57}}
Three to five years, certainly within
the next decade, where you have these

{{00:11:00}}
increasingly capable machine systems that
are integrated into the fabric of life.

{{00:11:05}}
But again, speaking about AI as a
meta technology, there's so many

{{00:11:09}}
adjacent technologies like quantum
computing, for instance, and, the

{{00:11:14}}
other field that I think is so I
talk about AI as representing this

{{00:11:19}}
tipping point in human evolution.

{{00:11:22}}
And I think the other technology
that has a similar as profound impact

{{00:11:27}}
that's unfolding now is biotechnology
where we're really being able to

{{00:11:32}}
look at the components of life itself
through looking at the DNA, being

{{00:11:38}}
able to manufacture ways to fight
human disease, where I think these two

{{00:11:44}}
technologies, artificial intelligence
and biotechnology are going to have a

{{00:11:47}}
really significant impact on the condition
of humanity over the 21st century.

{{00:11:54}}
Very interesting.

{{00:11:54}}
I just wanted to ask you one last
question, an overview question to wrap up.

{{00:11:58}}
I find myself sometimes in
conversations with people.

{{00:12:00}}
They know them in technology to some
degree, and they want to talk about AI.

{{00:12:03}}
And I'm just curious for yourself over
the  recent years how do you feel about

{{00:12:07}}
the mainstream conversation about AI?

{{00:12:09}}
Obviously a lot of things that maybe
are on the extremes, but do you

{{00:12:12}}
feel increasingly comfortable that
people are becoming better able to

{{00:12:15}}
talk about it in the mainstream?

{{00:12:17}}
Or do you still think there's
just a lot of hurdles and

{{00:12:19}}
misconceptions and things like that.

{{00:12:20}}
It's still making it very difficult for
people to even think straight about AI

{{00:12:23}}
I think there are hurdles and
misconceptions, which is why I always

{{00:12:27}}
start with first principles, you know
So often we talk about AI for instance

{{00:12:31}}
even in a business context and we
launch immediately into these are the

{{00:12:36}}
15 financial service applications that
I've been building with Gen AI and this

{{00:12:41}}
is how much it's optimized my team's
productivity but it's really important

{{00:12:46}}
in my view to start with the foundational
principles about what is AI, why did

{{00:12:52}}
we get here and what does it mean?

{{00:12:54}}
And I think just having that
conceptual wrapping is really

{{00:12:57}}
important as we navigate the age
of AI, but the kind of macro view.

{{00:13:01}}
But on the other hand, of course,
it's going to be controversial.

{{00:13:05}}
You're talking about something which is
so deeply philosophical, a technology

{{00:13:09}}
which early data suggests is being not
only capabilities are developing at an

{{00:13:15}}
exponential rate, but adoption as well.

{{00:13:18}}
If it's going to change the fabric
of society, if it's going to change

{{00:13:21}}
economic models, it's so important
that we have these debates.

{{00:13:25}}
More broadly, I'm just very
happy to see that this is a

{{00:13:30}}
debate that people are having.

{{00:13:32}}
We're going to have many, opinions, but
having that debate is really important.

{{00:13:37}}
Absolutely.

{{00:13:37}}
It's important to have guides
like yourself to help us as we

{{00:13:39}}
navigate this debate as well.

{{00:13:41}}
So thank you very much for joining us
and being a part of FinovateEurope.

{{00:13:43}}
And thank you for joining us as well.

FEU.INT.001: Hi, I'm David Penn,
research analyst with Finovate.

She is the author, generative AI
expert, and founder of Tamang Ventures.

I know you just finished
a Q&A few minutes ago.

And always when you're talking
about AI, we're only starting

So when we go into the Q&A, the
Q&A is actually my favorite part.

There were so many good questions and I
felt Time limited us, but we could have

I think one of the things when
you mentioned in your keynote, the

idea that the opinions on AI ranged
from the dangerous and devastating

You can imagine the range of
questions that might pop up and

Let's start talking about AI and
particular in financial services.

One of the questions I'm thinking that a
lot of people are curious about are what

are some of the things that financial
institutions need to know when it comes

to implementing things like generative AI?

So I'd say that the most important
component of implementing is making

sure your data is ready, structured
well, in an ecosystem where the

security is high, because there's
There is no AI implementation

And in fact, for any knowledge industry,
like financial services or like banking,

a lot of the value you're going to get in
terms of actual AI applications for your

specific use case, whether it's to launch
a customer services chatbot or actually

as a kind of research tool, it's going to
come from that institutional knowledge.

All that data structured and
unstructured within your organization.

So making sure that data is ready
to be optimized is going to be key.

I'm curious about the idea
of de risking adoption.

I imagine some of what you're talking
about now has to do with that.

What are some of the pain points
that financial institutions tend

to encounter when they're making
this implementation progress?

So obviously, again, one of the
most important ones is making

And there were lots of questions around,
especially when some of these foundational

And certain organizations were just
throwing in data into platforms like

chatGPT and finding out that data wasn't
safe before you start experimenting

with any models or any partners to
ensure the integrity of your data is

But also, I would say
more broadly when you see.

The new applications that AI allows,
including on cyber or kind of biometrics

and identity hacking, you're going to have
to upgrade your systems as well, right?

Your cyber systems, you're going to have
to ensure that they are AI resilient.

And part of that might, again, be
working with a cyber partner who is

developing AI as part of the defense
strategy more broadly to mean that your

risks are going to go, your surface
area of attack is going to be smaller.

It's interesting that you
mentioned partnerships.

One of the things that's interesting to
me as I think about this is even before

generative AI started to come on the
scene, a lot of financial institutions

were understanding the power of their
data for personalization, a lot of things.

And it just feels like with AI, that
whole challenge and opportunity has

And I'm just wondering, in terms of
financial institutions being able to It

seems like a lot of work that they might
not necessarily have the pedigree for.

So I'm wondering about
that particular challenge.

Is this almost certainly a
partnership type of thing?

And I'd say more broadly,
that's just the line.

It's the direction in which everything
is moving as we're in this kind of

digital ecosystem where you need
to have so many different layers of

expertise, not only domain specific
expertise, but now you're talking about

So you need to have that data science
background or machine learning background.

And what we are starting to see
in the ecosystem is that lots of

big players or even small players
are going to partner with the tech

I would just say, choose
your partners wisely.

But in part, the reason why this is
folding unfolding and a partnership kind

of model is because it simply doesn't
make sense for a single institution to try

and build up that level of AI capability,
which is ultimately going to be available

Maybe it makes sense if you're Bloomberg.

They were one of the first financial
services institutions that looks at

So they build Bloomberg GPT, which
is a large language model, which is

optimized for financial services based
on all of their historical data, right?

But even now the coin is still out as to
whether what Bloomberg built can actually

be better than some of the best in class
frontier models developed by AI specific

But regardless of that, most organizations
will, it will be that I think what's

going to happen is that you have these...

this platform layer, which is being
developed by the AI and tech companies,

and then other institutions plug
in and optimize that platform for

But they're still going to
need a lot of services around.

So I don't think AI is
going to take away any jobs.

Take place as you implement
this AI, the AI applications

Nina Schick, Author, Generative AI Expert, and Founder of Tamang Ventures, joins the Finovate team at FinovateEurope to talk about the realities of adopting generative AI and what we can learn from early adopters across other industries.