From financial crime to supply chains, JPMorgan’s head of AI research outlined 7 areas the bank seeks to apply technology

Established financial institutions often have remarkably slow, clumsy technology – but that is what tempts Manuela Veloso, head of machine learning at Carnegie Mellon University, at JPMorgan Chase three years ago.

“JPMorgan has been around for a long time. Unlike some tech companies like Amazon or Google, JPMorgan was not born digital and that was one of the challenges that caught my eye: How do you bring AI to these non-digital services?” Veloso told Insider.

Veloso is the head of artificial intelligence research at JPMorgan, and is currently leaving her role at the university. Her work, unlike many technology teams on Wall Street, is not driven by business needs. Instead, Veloso remembers infinite ti-if.

What if Chase gained 2 million online users tomorrow? What if

Federal Reserve
has it raised interest rates aggressively this week? And what are the ripple effects of these if-ifs?

Exploration is the key to Veloso’s research. It is one of the goals of JPM’s lead AI researcher to “make people embrace, to grow their hearts, to use AI”, even if it does not necessarily lead to technological development or a business tool.

Veloso is at the helm of a team of nearly 60 academics whose specialization covers math, cryptography, electrical engineering and machine learning, to name a few.

Describe the seven main challenges her team is trying to solve with AI.


Financial crime

JPMorgan is researching how artificial intelligence can help eradicate financial crime, which is one of the most frustrating problems Veloso has faced since joining the bank.

“I did not expect financial crime to be such an important focus for my team,” he said.

Identity theft will cost US businesses $ 721 billion in 2021, a 42% increase from the $ 502 billion loss in 2019, according to a March report by Aite Group, a financial research and consulting firm. In this time frame, almost half (47%) of American respondents reported being the victim of some form of identity theft, such as account acquisition or application fraud.

One way to fight financial crime is to explore how data and artificial intelligence can improve fraud and detection, he said. The bank is working to extend its detection beyond individual incidents and instead train AI to detect behavioral patterns or sequences of events that are indicative of financial crime. Graphs are used to model these behavioral networks, he added.


Large financial systems

A crane in a yard lifting a shipping container

The companies relied on simple supply chains until the pandemic stopped.

Justin Sullivan / Getty Images


The Veloso team uses AI to predict the impact of interactions between different players on large financial systems, often referred to as multi-agent systems. Examples include supply chains or when multiple parties are involved in a trade or exchange.

A key part of this – and an undervalued application that humans have not yet exploited – is the use of AI technology simulations to understand and test the ripples of these interactions, Veloso said. But to understand that, researchers need to go beyond the historical data, he added.

Real data, while valuable, does not necessarily allow the user to expand to different capabilities. Veloso likened AI development to historical data on a copier. “Whatever happens can be reproduced,” but it does not let you go beyond the circle. The simulations allow you to explore a much larger space than the actual data tells you, “he said.


Data management

Despite the limitations that historical data may have when developing AI, data is still embedded within JPMorgan. The Veloso team is exploring how to use artificial intelligence to manage data, such as securely sharing data between multiple departments in a bank or externally, he said.

The bank uses AI to obtain public information and help with checks on who has access and sharing rights, Veloso said. There is a lot of emphasis on infrastructure, especially for making this data available in the cloud, he added.

The team is also deepening its understanding of how cryptography and synthetic data can help protect privacy and data protection, he added.


Empowerment of employees

Another key goal of the research team is to understand how AI can improve the way employees work.

This includes using internal chatbots to help call center agents access through banking resources, use AI methods to classify emails, and automate report generation and PowerPoints, Veloso added.


Customer experience

chase atm


Justin Sullivan / Getty


Veloso’s work also focuses on the customer experience. This includes how quickly employees access customer data and use data to streamline the customer experience.

For example, the information required for your customer knowledge regulations already exists in JPMorgan documents or publicly available. However, mapping existing information in a given data field is a manual process, Veloso said.

Now, JPMorgan automates this with natural language editing and presents the customer with a form filled out by the AI ​​that they need to check is correct, he said.


Regulation and compliance

As a global bank, JPMorgan has thousands of obligations and rules to understand and know which regulations apply to which jurisdictions and companies.

If a banker talks on the phone with a customer from Singapore, or writes a contract with a customer in Italy, understanding the laws that govern what they do is crucial.

It is a complex navigation system, Veloso said. The bank analyzes how natural language processing can help and spends time understanding how human-artificial interactions evolve, he said.


Corporate and industrial values

All this work must be done in a way that strengthens the values ​​of explanation, trust, justice and the social good of the bank and the industry, while maintaining a realm of strong technology.

Much of this work is concentrated in the bank’s Explanatory Center of Excellence. Created in July 2020 and led by the AI ​​research team, the team shares techniques and tools that support understanding how technology works and ensuring its proper use.

Related Posts

Leave a Reply

Your email address will not be published. Required fields are marked *