ABA Banking Journal - September/October 2017 - 26
SPECIAL REPORT > TECHNOLOGY AND INNOVATION
Regulatory reporting is one area that seems ripe for digital
disruption. Today, filing call reports is a quarterly activity that
requires significant time. It would not be hard to imagine
a software solution that was tied into a bank's back-end
systems and prepopulated all of the key reporting fields.
Moreover, it would be possible for regulators to receive a
steady feed of data from a bank that would give them an
ongoing view into the bank and may reduce the frequency
with which exams are necessary.
Another area that holds great potential is in know-yourcustomer procedures. Today, onboarding and verifying a
customer's identity is a manual, time-consuming task that
relies on physical identification documents such as a driver's
license. In some cases it is even illegal to digitize these
documents. A digital identity could help banks quickly and
accurately get to know new customers and manage risks.
Artificial intelligence has long been in the realm of science
fiction movies, but we are seeing real-world examples of
software that learns and adapts. AI as discussed today
describes a process known as machine learning. Machine
learning allows computers to learn, without specific
programming instructing them how to operate. Unlike
traditional computing-in which devices are given a specific
input and in response take a specific pre-programmed
action-machine learning allows computers to change their
programming and respond differently as new information is
made available. While the underlying technology is complex,
it powers two key use cases that make technology more
Natural language processing. Tech companies have
developed virtual assistants (like Apple's Siri or Amazon's
Alexa) that allow users to interact conversationally with a
computer program by asking questions or giving commands.
Spoken language is not as straightforward as coding
language, as there are often many ways to say something.
AI allows these programs to understand complex voice
commands and translate them into computer code.
Banks have begun building interfaces that work with these
virtual assistants, giving customers access to bank account
information and performing basic tasks such as checking
ABA BANKING JOURNAL | SEPTEMBER/OCTOBER 2017
a balance or paying a bill. While these applications may
seem like a gimmick today, features on the first generation
of smartphones were similarly limited. As voice recognition
technology improves, virtual assistants will become
Big data. Machine learning allows software programs
to analyze large sets of unstructured data. Traditionally,
data analysis required a well-organized and structured
set of data with which a researcher could test specific
hypotheses. Machine learning allows users to draw
valuable insights from much larger sets of data than were
One way this could help bankers is by improving fraud
detection. Traditional fraud monitoring systems rely on
specific non-personal rules (like geography) to detect
fraudulent transactions. Machine learning could be applied
to analyze the transactions of each customer, flagging
transactions that are out of their normal habits.
This improved analytical capability has the potential to
give banks insights that could allow them to develop better
credit models and more accurately identify risks. The
power of big data is, however, highly dependent on the
quality of the data, which is not always easily accessible.
It is still a long time before banks will easily be able to get
big-data insights from their existing data.