Name: Code & Capital - AI in action
Uploaded: 2024-06-12T13:13:10.956Z
Duration: 490.52
Description: Code & Capital explores how smart players are harnessing Ai to solve real pain points for their businesses internally and for their customers at Finovate Spring with Vivian Yeung Executive Vice President, Chief Digital & Technology Officer at Fremont Bank.

Code & Capital - AI in Action.mp4

Code & Capital - AI in Action.vtt

ENGLISH

{{00:00:01}}
One of the use cases that I see
throughout the industry is using

{{00:00:20}}
artificial intelligence for
elevating the client experience.

{{00:00:24}}
We are actually doing it ourselves.

{{00:00:26}}
This elevation just really
enhances the user experience.

{{00:00:30}}
We understand the customer's behaviors.

{{00:00:33}}
The preferences and through
artificial intelligence, it's

{{00:00:36}}
just going to be remarkable on
the power and utilization of it.

{{00:00:40}}
And I think in the future, we'll be
looking at how do we really balance

{{00:00:45}}
that out with some of the risks
and some of, just the authenticity

{{00:00:50}}
of the, data and information.

{{00:00:59}}
In personalized banking, especially,
you think about it, it's really unique

{{00:01:03}}
to every individual and every client.

{{00:01:06}}
So with that, with personalized banking
is data, and where does the data reside?

{{00:01:11}}
How do you really take in
all of the information?

{{00:01:14}}
Use the artificial intelligence
tools to integrate, grab the data,

{{00:01:20}}
have the right level of data and
information, because data is essential

{{00:01:24}}
in order to be able to bring together
all of the personalized banking

{{00:01:28}}
experiences for an individual client.

{{00:01:32}}
Data's key when it comes to
artificial intelligence, especially

{{00:01:35}}
when we're talking about larger
language models and data sets.

{{00:01:39}}
I think that'll be one of the powerful
implementations, but it's also a struggle,

{{00:01:44}}
too, because I think with financial
services, especially, we have different

{{00:01:48}}
disparate systems, and then, really, how
do we get at all of the data, and then

{{00:01:53}}
make sure that the data is actually in
a way that's organized and presentable

{{00:01:59}}
for the artificial intelligence
models to be able to leverage and use.

{{00:02:09}}
Great.

{{00:02:10}}
I think in lending, especially,
there's some really cogent use cases

{{00:02:15}}
currently in today's environment.

{{00:02:17}}
For example, a document and viewability
of the document using artificial

{{00:02:21}}
intelligence to look at the documents,
identify the documents, sometimes

{{00:02:25}}
unstructured, sometimes structured, and
then taking the field elements and then

{{00:02:29}}
actually putting them into their systems
of record or originations platforms.

{{00:02:34}}
It makes it just great.

{{00:02:36}}
Automated versus doing a
manual process as an example.

{{00:02:40}}
I think there are going to be
future use cases of lending around.

{{00:02:46}}
Just what is the approval as an example?

{{00:02:50}}
It's pretty, far in the future, but
I think this is where we have to

{{00:02:54}}
actually do a balance of what is
the artificial intelligence actually

{{00:02:59}}
saying in terms of approved or
disapproved, and what is the actual

{{00:03:03}}
thought process that goes into that.

{{00:03:05}}
So there definitely is going to be
a checks and balance, I think, in

{{00:03:08}}
the interim with humans, and then
also what's generating from the

{{00:03:12}}
artificial intelligence aspect.

{{00:03:14}}
And then on the fraud prevention
side, I think this particular

{{00:03:17}}
area is a big focal point.

{{00:03:19}}
Not just for Fremont Bank as
an example, but I think for

{{00:03:22}}
financial institutions in totality.

{{00:03:24}}
It's how do you prevent fraud
using artificial intelligence.

{{00:03:28}}
And we have some rudimentary ways of
doing that today with, just anomaly

{{00:03:36}}
detection, behavioral aspects, or even if
somebody's going in and making payments,

{{00:03:42}}
how often do they normally make payments?

{{00:03:45}}
And has that actually changed?

{{00:03:47}}
And the artificial intelligence tool
can look at it and say you know what?

{{00:03:50}}
It actually has changed.

{{00:03:52}}
Let's make a flag.

{{00:03:53}}
Let's alert and then have
somebody take a look at it.

{{00:03:56}}
That's an example of a fraud detection.

{{00:04:05}}
I think one of the things that
really excites me is when you

{{00:04:08}}
think about chat GPT and you
think about multimodal artificial

{{00:04:13}}
intelligence it's, so powerful, right?

{{00:04:16}}
You have, text, you have voice, you
have facial recognition, emotions.

{{00:04:22}}
And when those come into play, all in
aggregate, think about a client calling

{{00:04:27}}
in and being able to determine if they're
disgruntled or if they're happy if

{{00:04:33}}
they're really pleased with the company.

{{00:04:36}}
and being able to use artificial
intelligence for let's say the agent

{{00:04:40}}
or the representative that's taking
the call to have at least a good idea

{{00:04:45}}
or understanding of what that person
is feeling emotionally and then that's

{{00:04:50}}
how best to structure the conversation
with the person that's calling in.

{{00:04:55}}
I think that's just such a phenomenal
user experience for the client and

{{00:05:00}}
also for the associate to feel a
little bit better about how to take

{{00:05:04}}
the time to service the client and
have that right level of conversation.

{{00:05:09}}
That's one example.

{{00:05:10}}
Sentiment analysis is another one.

{{00:05:13}}
Being able to take the phone call
it would be text to speech or speech

{{00:05:19}}
to text, and then taking the text
in and really analyzing what is the

{{00:05:23}}
call about, and then figuring out,
is this call, Are this person likely

{{00:05:29}}
to continue with the company because
of the interaction and the scripting

{{00:05:34}}
that actually was inside of the call?

{{00:05:37}}
I think sentiment analysis is going to
be another emerging kind of technology

{{00:05:43}}
for artificial intelligence and the
way that we use it within industry.

{{00:05:55}}
I think for all of us, We all want to
have a customer interaction with a human.

{{00:06:02}}
I think a human touch for us as a
financial institution is hugely important.

{{00:06:08}}
We have that direct interaction with our
clients, especially as a community bank.

{{00:06:14}}
So this particular area of
focus is very critical for us.

{{00:06:18}}
We wouldn't have artificial
intelligence tools that interface with

{{00:06:22}}
our clients directly at this point.

{{00:06:25}}
We want to make sure that we
understand what the artificial

{{00:06:28}}
intelligence response is.

{{00:06:31}}
So how we leverage and utilize it is
being able to make the associate or the

{{00:06:36}}
representative that's taking in a call or
taking in a branch visit and understanding

{{00:06:42}}
What the result is of the artificial
intelligence and then looking at it and

{{00:06:47}}
say, yep, that's the right response.

{{00:06:48}}
As an example, I'll
put that into the chat.

{{00:06:51}}
So it makes it faster to be able to answer
a client's question, but it's still coming

{{00:06:56}}
from the associate who's looking at those.

{{00:06:59}}
Responses first.

{{00:07:01}}
I think when we talk about authenticity
and we talk about the credibility

{{00:07:06}}
utilizing artificial intelligence,
I do think that at some point we are

{{00:07:10}}
going to have some disclaimers around
artificial intelligence that's been

{{00:07:13}}
generated by AI and the utilization of it.

{{00:07:17}}
And then the other piece in the
future will be about the authenticity.

{{00:07:21}}
So I think as we move
along, is it generated?

{{00:07:26}}
Is it a deepfake?

{{00:07:28}}
How do I know that it is?

{{00:07:29}}
These particular pieces, I think there's
going to be validation technology.

{{00:07:35}}
You take technology and you
combat it with technology.

{{00:07:38}}
And I think in the future, you'll have
a lot of services that have immutable

{{00:07:42}}
video immutable, maybe data store
conversations, and actually having

{{00:07:49}}
an authenticated or verified by.

{{00:07:51}}
Similar to like when you look
at a website, there's a little

{{00:07:53}}
padlock, you know that it's secure.

{{00:07:55}}
I think similarly in the future
we'll have some technologies that

{{00:07:59}}
tells us that this is artificially
intelligence generated and that it's

{{00:08:04}}
verified and authentic versus not.

One of the use cases that I see
throughout the industry is using

artificial intelligence for
elevating the client experience.

This elevation just really
enhances the user experience.

The preferences and through
artificial intelligence, it's

just going to be remarkable on
the power and utilization of it.

And I think in the future, we'll be
looking at how do we really balance

that out with some of the risks
and some of, just the authenticity

In personalized banking, especially,
you think about it, it's really unique

So with that, with personalized banking
is data, and where does the data reside?

How do you really take in
all of the information?

Use the artificial intelligence
tools to integrate, grab the data,

have the right level of data and
information, because data is essential

in order to be able to bring together
all of the personalized banking

Data's key when it comes to
artificial intelligence, especially

when we're talking about larger
language models and data sets.

I think that'll be one of the powerful
implementations, but it's also a struggle,

too, because I think with financial
services, especially, we have different

disparate systems, and then, really, how
do we get at all of the data, and then

make sure that the data is actually in
a way that's organized and presentable

for the artificial intelligence
models to be able to leverage and use.

I think in lending, especially,
there's some really cogent use cases

For example, a document and viewability
of the document using artificial

intelligence to look at the documents,
identify the documents, sometimes

unstructured, sometimes structured, and
then taking the field elements and then

actually putting them into their systems
of record or originations platforms.

Automated versus doing a
manual process as an example.

I think there are going to be
future use cases of lending around.

It's pretty, far in the future, but
I think this is where we have to

actually do a balance of what is
the artificial intelligence actually

saying in terms of approved or
disapproved, and what is the actual

So there definitely is going to be
a checks and balance, I think, in

the interim with humans, and then
also what's generating from the

And then on the fraud prevention
side, I think this particular

Not just for Fremont Bank as
an example, but I think for

It's how do you prevent fraud
using artificial intelligence.

And we have some rudimentary ways of
doing that today with, just anomaly

detection, behavioral aspects, or even if
somebody's going in and making payments,

how often do they normally make payments?

And the artificial intelligence tool
can look at it and say you know what?

Let's alert and then have
somebody take a look at it.

I think one of the things that
really excites me is when you

think about chat GPT and you
think about multimodal artificial

You have, text, you have voice, you
have facial recognition, emotions.

And when those come into play, all in
aggregate, think about a client calling

in and being able to determine if they're
disgruntled or if they're happy if

and being able to use artificial
intelligence for let's say the agent

or the representative that's taking
the call to have at least a good idea

or understanding of what that person
is feeling emotionally and then that's

how best to structure the conversation
with the person that's calling in.

I think that's just such a phenomenal
user experience for the client and

also for the associate to feel a
little bit better about how to take

the time to service the client and
have that right level of conversation.

Being able to take the phone call
it would be text to speech or speech

to text, and then taking the text
in and really analyzing what is the

call about, and then figuring out,
is this call, Are this person likely

to continue with the company because
of the interaction and the scripting

I think sentiment analysis is going to
be another emerging kind of technology

for artificial intelligence and the
way that we use it within industry.

I think for all of us, We all want to
have a customer interaction with a human.

I think a human touch for us as a
financial institution is hugely important.

We have that direct interaction with our
clients, especially as a community bank.

So this particular area of
focus is very critical for us.

We wouldn't have artificial
intelligence tools that interface with

We want to make sure that we
understand what the artificial

So how we leverage and utilize it is
being able to make the associate or the

representative that's taking in a call or
taking in a branch visit and understanding

What the result is of the artificial
intelligence and then looking at it and

As an example, I'll
put that into the chat.

So it makes it faster to be able to answer
a client's question, but it's still coming

from the associate who's looking at those.

I think when we talk about authenticity
and we talk about the credibility

utilizing artificial intelligence,
I do think that at some point we are

going to have some disclaimers around
artificial intelligence that's been

generated by AI and the utilization of it.

FinovateSpring

Finovate

Code & Capital

Expert Insight

Code & Capital explores how smart players are harnessing Ai to solve real pain points for their businesses internally and for their customers at Finovate Spring with Vivian Yeung Executive Vice President, Chief Digital & Technology Officer at Fremont Bank.