Club @ Sibos
D
espite the “hoopla” about artificial in-
telligence (AI) in the financial services
industry, for the foreseeable future its
use will be mostly limited to tedious and repet-
itive clerical work, says Professor of Industrial
Engineering and Operations Research at
Columbia University, David Yao. The technolo-
gy is “fairly mature” in areas such as customer
data processing and form filling, while there
is promise in portfolio optimisation and robo
advisory although Yao cautions how successful
such applications will be is uncertain.
Two other areas of promise include security
and sentiment and news analysis. Separating
hype from reality is always difficult when it
comes to new technologies; Yao says those re-
sponsible at financial institutions should make
sure they keep up with new technologies and
concepts. “Trying to guess the future is always
difficult, but people should stay open-mind-
ed and ensure they have a workforce that is
dynamic and open to new knowledge.”
At the European Payment Summit in the
Netherlands earlier this year, Vivek Bajaj,
global vice-president of Watson Financial
Services (IBM’s cognitive services division
for the banking industry) said AI was in itself
of “no value”. Rather, it is all about context –
firms shouldn’t think about ‘AI products’, they
should think about business products that
solve specific business problems.
Finding contextual information among the
huge amount of data and information that is
flooding the world is the main challenge for
financial institutions, he added. Institutions
must decide for themselves which use cases
deliver the most value. Bajaj highlighted sever-
al areas that banks should consider as suitable
for AI deployment, including customer insight
and engagement, regulatory compliance and
payments (particularly instant payments).
Increasingly, financial institutions are being
forced to engage with their customers (both
retail and corporate) on the platforms of the
customers’ choice. They are required to de-
liver insights, rather than the digital experi-
ence itself. This is referred to as augmenting
intelligence, providing insights on which
humans, rather than machines, can act. Bajaj
says the organisations with the most data
won’t necessarily be winners; the winners
will be those that can deliver useful insights
to customers. Examples of AI deployment
in this area include Brazil’s Bradesco, which
uses an augmented intelligence AI system. Its
virtual agent, Bradesco Inteligência Artificial,
assists employees and customers in answering
questions about bank products. Since imple-
mentation, Bradesco has recorded customer
satisfaction levels of higher than 85%, with
94% of queries being handled by the virtual
agent.
In regulatory compliance, AI can be applied
successfully in areas such as know your cus-
tomer (KYC), anti-money laundering (AML)
and surveillance. For example, Promontory
ARTIFICIAL INTELLIGENCE
Financial Group, along with IBM Watson, ap-
plies AI to regulatory compliance. The system
can identify potential misconduct through
application of voice surveillance technology
to discern trader conversations, as opposed to
simply investigating events after the fact. The
system parses millions of pages of regulatory
documents and alerts banks of the actions they
need to take.
In instant payments, AI will be crucial to en-
abling immediate settlement while performing
the required checks and balances for regulato-
ry compliance, including fraud analysis.
Where should financial institutions start
with AI? Bajaj advised that an understanding
of the use case should be well and truly es-
tablished. As a first step, financial institutions
should consider applying AI in the context of
existing business products, using existing data;
“nothing should happen without a business
case”.
In determining a business case, financial
institutions must ensure they are “ambitious
enough” says Jean Devambez, global head
of digital and acceleration at BNP Paribas
Securities Services. “We are convinced that
AI can bring value to us and to our clients. But
financial institutions need to bear in mind that
AI requires reengineering of workflows; it is
not just a new technology but is a new way of
working. You must be prepared to review the
way you work.” Devambez says rather than se-
lecting small use cases to test AI technologies,
financial institutions should target sufficiently
Jean Devambez, BNP Paribas Securities Services
would have posed significant compliance and
risk issues for the institution. “It used to be
difficult to work with smaller companies be-
cause many audits – IT, security, compliance,
etc were required. But if the business case is
strong enough it makes sense to persevere,”
he says.
Innova tackles two challenges for financial
services companies and institutional inves-
tors – growing regulatory requirements and
mounting volumes of data. “By taking a stake
in Fortia, we are not only helping a start-up
flourish but also ensuring our clients have
access to the latest technologies to grow and
“We are convinced that AI can bring value to us and to our clients.
But financial institutions need to bear in mind that AI requires
reengineering of workflows; it is not just a new technology but is
a new way of working. You must be prepared to review the way
you work”
JEAN DEVAMBEZ, BNP PARIBAS SECURITIES SERVICES
broad processes to ensure they get the most
out of the technology. “This will take more
time to implement but the scope to use and
to leverage the benefits of AI are very broad.
AI is relevant across the spectrum of digital
transformation, enabling financial institutions
to improve their own processes along with
those of their customers while also developing
new business models.”
In January 2017 BNP Paribas Securities
Services took a minority stake in financial
technology start-up Fortia Financial Solu-
tions. The company’s investment compliance
platform, Innova, enables asset managers and
owners to monitor the compliance of their
funds with local and international regulatory
requirements.
When BNP began working with Fortia about
three years ago, the start-up had only seven
employees, says Devambez. In the past, this
8 | CLUB@SIBOS | Bringing the world to Sibos and Sibos to the world
develop their business, meet compliance re-
quirements and enhance operational efficien-
cy,” says Devambez.
In early October, the World Economic Forum
(WEF) released a report into the opportuni-
ties, challenges and broader societal impli-
cations of AI in financial services. It says AI
investments by financial institutions are being
made across a broad spectrum, from relatively
conservative efforts to improve existing pro-
cesses to “bold bets” on new capabilities and
business models. This spectrum begins with
the use of AI automation to improve the effi-
ciency of business-as-usual processes (Yao’s
foreseeable future applications) and ends with
building new products, services and business
models that use AI at the core.
Says the report: “AI resolves the traditional
trade-off between cost and customisation,
allowing institutions to offer tailored products