Revolutionising finance -The blueprint for implementing AI and automation in investment strategies.mp4

Revolutionising finance -The blueprint for implementing AI and automation in investment strategies.vtt

ENGLISH

{{00:00:01}}
www.

{{00:00:03}}
technomaista.

{{00:00:06}}
com www.

{{00:00:11}}
technomasta.

{{00:00:13}}
com www.

{{00:00:18}}
umai.

{{00:00:21}}
edu

{{00:00:34}}
There are different aspects, of course.

{{00:00:36}}
You can use this for generating
trading strategies, going from

{{00:00:40}}
very basic algorithms, or we can
use it for onboarding customers.

{{00:00:53}}
If we take, for instance, the advent
of large language models and the

{{00:00:57}}
ability to use instructor chat models
to onboard your customers, guide them

{{00:01:03}}
through different processes, this kind
of automation helps you in Speed up

{{00:01:09}}
the processes within your company.

{{00:01:19}}
I think it's more transparent
and it's more non emotional.

{{00:01:26}}
You, you don't have a biased
guidance from a human.

{{00:01:31}}
So you might not.

{{00:01:32}}
think that you get tricked in any way
because the AI or the chatbot you're

{{00:01:37}}
interacting with is non emotional with
you and this is a clear instruction.

{{00:01:49}}
There are different things.

{{00:01:51}}
For instance, you can use differential
privacy, which is just a method of

{{00:01:56}}
using the gradient descent and add
some noise and clip the gradients.

{{00:02:01}}
to obscure private data or there are
other measures you can do using a custom

{{00:02:08}}
lung chain which is named bad actor
chain and it's a technique how you can

{{00:02:13}}
see how the data or the large language
model for instance could be exploited.

{{00:02:27}}
I mean there are pretty capable
of extracting different patterns.

{{00:02:32}}
So we have a lot of data available
and with that, we have the need

{{00:02:38}}
of having more advanced algorithm
helping us to process the data.

{{00:02:43}}
And so I think in general, more
advanced algorithm, more data

{{00:02:47}}
processing will be set in place.

{{00:02:50}}
I think AI and automation definitely has
a massive role to play in the future.

{{00:02:54}}
There's no escaping it.

{{00:02:55}}
I think that is inevitable.

{{00:02:58}}
Most organizations are already obviously
on, on, a big journey of automation.

{{00:03:03}}
But I think we are going past the
level of automation into AI, i.

{{00:03:06}}
e.

{{00:03:07}}
building intelligence into
our processes and building

{{00:03:10}}
intelligence into our products.

{{00:03:12}}
So there's definitely a
huge, room for AI to play.

{{00:03:15}}
I was on a panel yesterday here
where we deliberate a little

{{00:03:19}}
bit more in, on this topic.

{{00:03:22}}
But there are specific areas where AI
will definitely have a much bigger impact.

{{00:03:27}}
The first being anything that
is repetitive that, we do,

{{00:03:32}}
reporting being a big piece of
it massive, wins using automation

{{00:03:36}}
and AI on that, ensuring quality.

{{00:03:39}}
Second, where you're using huge swaths
of data huge volumes of data to make

{{00:03:43}}
decisions, again, AI would be a far
superior partner to have in play

{{00:03:48}}
than having a lot of people around
the room making those decisions.

{{00:03:51}}
And the last one is where you're
starting to make predictive proactive

{{00:03:56}}
strategic steers for your business.

{{00:03:58}}
Again, AI can do a lot of the legwork
to help drive strategic decisions

{{00:04:03}}
for the organization as well.

{{00:04:12}}
For instance, you can actually look
into risk from another perspective.

{{00:04:18}}
So if you are having the ability to
analyze more data, so for instance Bidden

{{00:04:25}}
Capital uses cryptocurrency, so we use on
chain data, we analyze the data patterns

{{00:04:31}}
and the fact we can mitigate risk.

{{00:04:33}}
In terms of cost saving and efficiencies
in the organization, I think there's

{{00:04:38}}
obviously been a significant improvement.

{{00:04:41}}
Where in the past, if you were building
risk models, or you were running risk

{{00:04:45}}
reports, or looking to build monitoring
dashboards you would have 10 individuals

{{00:04:51}}
to a topic or to a to a problem as opposed
to now you have a single model or a

{{00:04:58}}
single platform that can cover 50 to 60
different things that you're worried about

{{00:05:04}}
and it also has inbuilt intelligence to
be able to highlight to you early on what

{{00:05:08}}
are the specific areas where risks are
deteriorating or where you need to spend

{{00:05:12}}
time and energy or focus of the business.

{{00:05:15}}
So it's definitely tremendously
not just improved efficiency

{{00:05:18}}
in terms of the capture.

{{00:05:20}}
But it's also improved
efficiency in terms of focus and

{{00:05:23}}
prioritization of the organization.

{{00:05:25}}
And it's definitely improved human
error reduced the reliance on

{{00:05:30}}
human and qualitative judgment and,
brought much more systematic thought

{{00:05:34}}
process into, our risk thinking.

{{00:05:45}}
If you use like chatbots for onboarding
customers, you can teach them to interact

{{00:05:51}}
more humane and make even some jokes.

{{00:05:53}}
So you can add.

{{00:05:55}}
on top of that human level of experience?

{{00:05:59}}
That's a really great question.

{{00:06:01}}
Does AI fully replace human expertise?

{{00:06:05}}
Absolutely not.

{{00:06:06}}
There are things that AI can do very well.

{{00:06:09}}
Things like absorbing and consuming
vast volumes of data and coming

{{00:06:13}}
up with some sensible correlations
and predictive views on, things.

{{00:06:18}}
What it doesn't necessarily always get,
or at least as current capabilities.

{{00:06:23}}
is sensitivity to context of of markets
and also the sensitivity to customers

{{00:06:30}}
and sensitivity to relationship with
the clients or the client outcomes.

{{00:06:34}}
And so you definitely need to have,
especially for critical processes

{{00:06:37}}
or critical decisions, there is
a human layer that you definitely

{{00:06:40}}
need to have and that human layer is
more and more becoming specialized

{{00:06:45}}
and very focused on the domains.

{{00:06:47}}
But we do advise, we're very focused
on a great customer experience and

{{00:06:51}}
all of that is founded on very strong
data analytics that helps us drive

{{00:06:56}}
decisions, but ultimately we do have.

{{00:06:58}}
A human overlay in ensuring that we're
not fully or blindly reliant on data,

{{00:07:04}}
but there are, there is an ability
to check and challenge even what the

{{00:07:08}}
output of the models are, and which,
I think is critical as we develop

{{00:07:12}}
and start adopting AI more broadly.

You can use this for generating
trading strategies, going from

very basic algorithms, or we can
use it for onboarding customers.

If we take, for instance, the advent
of large language models and the

ability to use instructor chat models
to onboard your customers, guide them

through different processes, this kind
of automation helps you in Speed up

I think it's more transparent
and it's more non emotional.

You, you don't have a biased
guidance from a human.

think that you get tricked in any way
because the AI or the chatbot you're

interacting with is non emotional with
you and this is a clear instruction.

For instance, you can use differential
privacy, which is just a method of

using the gradient descent and add
some noise and clip the gradients.

to obscure private data or there are
other measures you can do using a custom

lung chain which is named bad actor
chain and it's a technique how you can

see how the data or the large language
model for instance could be exploited.

I mean there are pretty capable
of extracting different patterns.

So we have a lot of data available
and with that, we have the need

of having more advanced algorithm
helping us to process the data.

And so I think in general, more
advanced algorithm, more data

I think AI and automation definitely has
a massive role to play in the future.

Most organizations are already obviously
on, on, a big journey of automation.

But I think we are going past the
level of automation into AI, i.

building intelligence into
our processes and building

So there's definitely a
huge, room for AI to play.

I was on a panel yesterday here
where we deliberate a little

But there are specific areas where AI
will definitely have a much bigger impact.

The first being anything that
is repetitive that, we do,

reporting being a big piece of
it massive, wins using automation

Second, where you're using huge swaths
of data huge volumes of data to make

decisions, again, AI would be a far
superior partner to have in play

than having a lot of people around
the room making those decisions.

And the last one is where you're
starting to make predictive proactive

Again, AI can do a lot of the legwork
to help drive strategic decisions

For instance, you can actually look
into risk from another perspective.

So if you are having the ability to
analyze more data, so for instance Bidden

Capital uses cryptocurrency, so we use on
chain data, we analyze the data patterns

In terms of cost saving and efficiencies
in the organization, I think there's

obviously been a significant improvement.

Where in the past, if you were building
risk models, or you were running risk

reports, or looking to build monitoring
dashboards you would have 10 individuals

to a topic or to a to a problem as opposed
to now you have a single model or a

single platform that can cover 50 to 60
different things that you're worried about

and it also has inbuilt intelligence to
be able to highlight to you early on what

are the specific areas where risks are
deteriorating or where you need to spend

time and energy or focus of the business.

So it's definitely tremendously
not just improved efficiency

But it's also improved
efficiency in terms of focus and

And it's definitely improved human
error reduced the reliance on

human and qualitative judgment and,
brought much more systematic thought

If you use like chatbots for onboarding
customers, you can teach them to interact

on top of that human level of experience?

There are things that AI can do very well.

Things like absorbing and consuming
vast volumes of data and coming

up with some sensible correlations
and predictive views on, things.

What it doesn't necessarily always get,
or at least as current capabilities.

is sensitivity to context of of markets
and also the sensitivity to customers

and sensitivity to relationship with
the clients or the client outcomes.

And so you definitely need to have,
especially for critical processes

or critical decisions, there is
a human layer that you definitely

need to have and that human layer is
more and more becoming specialized

But we do advise, we're very focused
on a great customer experience and

all of that is founded on very strong
data analytics that helps us drive

A human overlay in ensuring that we're
not fully or blindly reliant on data,

but there are, there is an ability
to check and challenge even what the

output of the models are, and which,
I think is critical as we develop

At the QuantMinds International conference 2023, we get an exclusive insight into implementing AI and automation in investment strategies. Discussing this are Nicole Königstein,Chief Data Scientist & Head of AI & Quant Research at Wyden Capital and Chandi Bhan, Chief Risk Officer at Wise.