Page 33 - Banking Finance November 2019

ARTICLE

Avoiding fraud and money laundering is a challenge for many account lifestyle, appetite for risk, expected returns on
financial organizations. AI has the potential to help banks investment, etc.
become more efficient in the process of detecting fraud and
money laundering. To quickly identify potential fraud, AI ATMs: Image/face recognition using real-time camera
engineers have developed tools and systems that images and advanced AI techniques such as deep learning
automatically conduct and compress data that normally can be used at ATMs to detect and prevent frauds/crimes.
requires many hours of labor in just matter of minutes.
AI is not without challenges
Larger institutions are more inclined to update their legacy A wide implementation of a high-end technology like AI in
systems due to the rising number of fintech companies that India is not going to be without challenges. From the lack
are adopting AI. One of the Banking Giants, Citibank, is of a credible and quality data to India's diverse language
already using machine learning and Bigdata to prevent set, experts believe a number of challenges exist for the
criminal activities and monitor potential threats to Indian banking sector using AI.
customers in commerce. The company has adopted a new
anti-money laundering structure and has invested over $11 A key challenge is the availability of the right data. Data is
million to launch a new personal finance app that the lifeblood of AI, and any vulnerability arising from
encourages customers to participate in third party services. unverified information is a serious concern for businesses.
Imagine for example, the risks that could arise from KYC
Process Automation compliance AI systems if the data sources are incorrect. Or
Process Automation is one of the key drivers of automation consider the efficacy of a fraud detection AI system without
in banks and financial institutions, but it's also evolving into the right kind of data. Structured mechanisms for collecting,
cognitive process automation, where AI systems are able validating, standardizing, correlating, archiving and
to perform more complex automation JP Morgan Chase distributing AI relevant data is crucial.
recently invested in a new technology called COiN that
reviews the documents and extracts data in much less time India has 150+ languages with sizable spoken population.
than it would take a human. This tool reviews about 12,000 Applications which use speech to text or text to speech rely
documents (which without automation would require more on natural language processing (NLP) libraries and
than 360,000 hours of work) in just seconds. techniques. Banks can use the existing technologies to start
with to support some major Indian languages, but in order
Customer Support and Helpdesk: Humanoid Chatbot to effectively reach out to wider population in India, much
interfaces can be used to increase efficiency and reduce cost more progress is required on NLP front.
for customer interactions. (e.g. SIA of SBI)
Data access and data privacy is a central aspect of any AI
Risk Management: Tailored products can be offered to work banks do. These aspects will be of paramount
clients by looking at historical data, doing risk analysis, and importance with introduction of regulations in Europe such
eliminating human errors from hand-crafted models. as GDPR (General Data Protection Regulation). GDPR

Security: Suspicious behavior, logs analysis, and spurious
emails can be tracked down to prevent and possibly predict
security breaches.

Digitization and automation in back-office processing:
Capturing documents data using OCR and then using
machine learning/AI to generate insights from the text data
can greatly cut down back-office processing times.

Wealth management for masses: Personalized portfolios
can be managed by Bot Advisors for clients by taking into

BANKING FINANCE | NOVEMBER | 2019 | 33