JPMorgan to Invest in AI, Tackle Privacy Concerns

“We are not satisfied with the level of fundamental research in AI that is focused on financial markets,” says Sanoke Viswanathan, chief administrative officer of the corporate and investment bank at JPMorgan. “In discussing with clients, on the types of issues we are dealing with, there isn’t enough core research going on in areas such as market simulations, time-series predictions and things like that. So we want to set up a research capability that is focused on that.”

He says there is not only a shortage of understanding and development in AI in the sector, but that there is a lack of trust in the technology centered on its use of data. Viswanathan was speaking during a panel at the TradeTech Europe conference, held on April 25 in Paris, France.

The bank has been a big investor in AI in the past. In its asset-management business, for instance, it has developed powerful capabilities using machine learning to support alternative investment strategies, and employed a battery of data scientists in the process. Likewise, it is known to be exploring the applications of the technology across its investment banking and retail arms, each with different objectives.

Speaking on the sidelines of the conference, he explained to WatersTechnology that companies and individuals are concerned with sharing their information with the likes of digital assistants, such as Siri and Alexa, and providing personalized information to third-party tech firms. This lack of confidence in AI has created a “trust barrier” and that a level of “explainability” or transparency is required to inform clients on how their data is being used.

“There are millions of companies who are distributing content through these types of digital assistants who are going to be concerned about giving up their data to tech companies,” he says. “And similarly, if you are accessing your bank account [through a digital assistant] and you want to know what your balance is, that tech company might now have access to your balances.”

He highlights that firms have a responsibility to protect client interests and any AI systems utilize their data correctly and to their knowledge. One of the challenges to overcome in this area is when explaining the use of more sophisticated and complex versions of the emerging technology.

“The big concerns or issues to work through I think is when we get into more sophisticated AI, like deeper neural networks; this explainability issue,” he says. “How do you explain how the model is working and is it optimizing and coming up with the right explanations that are comfortable for clients, regulators and ourselves?”

Investment institutions and firms also have to consider the challenges involved in scaling the technology, in addition to the costs of cleaning up data, its curation and its governance. Given the upcoming  General Data Protection Regulation (GDPR) the issue of trusting AI technologies is proving more relevant than ever. During the panel discussion at the TradeTech event, Viswanathan outlined the types of controls taken to manage the technology and the data use.

“80 percent of the effort is in going into how do you have the right data governance, the right curation and ensuring that the right sort of data is used by the right sort of people and ultimately we need to be able to trace the lineage of where the data is being used in trading data sets,” he says.

Although the technology can prove challenging and is largely unexplored, Viswanathan explains that there are many benefits to reap from using AI. These include analytics optimization, sentiment analysis, language processing and anomaly detection for targeting patterns of suspicious behavior.