The application of artificial intelligence and machine learning in finance has expanded far beyond the clerical aspects of banking and the cybersecurity required to safeguard institutions and customers. A.I. is being used by many organizations and their clients to make decisions and suggestions with remarkable speed and attention to peoples’ needs, using individualized data points and market trends.

Given the complexity of these algorithms and the necessity for clients to be comfortable with what is happening with their money, financial institutions must be transparent. “We’re not just providing you movies.” Morgan Stanley’s chief analytics and data officer Jeff McMillan said at the Fortune Brainstorm A.I. conference in Boston, “we’re giving you investment advice.” He emphasized the importance of “explainability” in keeping customers pleased and businesses honest.
“You can go get these algorithms off of AWS and Azure. They’re not secret. And creating recommendations, it’s not super hard on a relative basis,” he added. “But explaining the recommendation to someone who has zero expertise in data science is really hard. Explaining why you are saying, ‘This idea is a good idea,’ in very layman’s terms is very, very difficult. We spent as much time trying to communicate the complex in simple terms as we did making the thing super complex.”
Jeff McMillan, Managing Director, Chief Analytics & Data Officer for Morgan Stanley Wealth Management
He claims that focusing on this can assist a company better understand a client’s needs, which in turn helps increase the efficacy of their A.I. technologies. “You don’t fail because your machine-learning algorithm isn’t ‘good enough,’” he explained. “In my experience, you fail because you can’t get that last mile where the machine learning intersects with human behavior. It’s that intersection where you’re trying to sort of cross that chasm, we have to say, ‘Listen, we’re not here to replace you job. We’re here to make you job actually better and more efficient,’ where I think organizations really struggle. This is an A.I. conference, but it’s really about the A.I. human element where I think organizations like all of us really need to focus our time and attention.”
Carol Juel, executive vice president and chief technology and operations officer of Synchrony, agreed and added that the pandemic had a significant impact on digital transactions. She noted her company’s acquisition of Pets Best, a pet health insurance company, and how the surge in people getting animals during the era of isolation provided the company with a wealth of new information about its clients.
“In financial services, in general, we have a significant volume of data,” she said. “On every individual customer, we have thousands of attributes that tell us about them and then we have behavioral attributes. We combine that with proprietary partner data and that’s the key piece because those then can drive the opportunities for personalization. Understanding someone’s digital engagement, knowing how much they spend, knowing what their behavior patterns are digitally, [then] you can drive that experience. And as the world moves massively digitally over the last 20 months, how we then improve authentication to reduce fraud, how we know more about those customers, so that experience can be frictionless. You can ultimately allow for those high-risk track transactions to flow through digitally, as more and more is happening through those channels. We measure everything from volume of customer satisfaction to reduction in call center volume. If you have that underlying data to build those models that you can rely on to reduce fraud and improve experience, that’s sort of the holy grail of digital transformation in financial services.”
Both panelists stressed the need for ethics, responsibility, and transparency in the way they employ A.I. and machine learning to make judgements, which was a recurring subject throughout the conference. According to McMillan, it is one of the most significant things a company can do.
“You have to define, ‘What does it mean to be ethical?’ How are you defining that, because in my business it’s going to be different for pharma than it’s going to be for manufacturing. So, you have to define what it means for the model to operate within the bounds of conformity. Organizationally, you have to have those conversations, you have to bring the right people to the table to define that. Then you have to have an independent group of people to monitor the efficacy of what you do and hold you – hold me as the guy that does this for a living – accountable and create transparency. If I go out of bounds, you have to call me to task and there has to be those forums to say, ‘Jeff is operating outside of the defined boundaries of the standards that we’ve defined for this, and we need to have a conversation.”
Finally, Juel provided some words of wisdom, as well as caution, to all enterprises that reply on artificial intelligence. “There’s a lot of, oftentimes, hype about the technology. The technology is amazing, I love tech, but I think you have to take a step back from that and really bring people along the journey. A.I. is and can be transformational for your organization. But it’s not the technology, it’s how does it apply to the business that you are in? What are the challenges? How fast do you want to move? Will your culture allow you to move at the speed you want to? What are the things you need to adapt in order to take advantage of the opportunities there? Because A.I. is not going to solve the problems of any organization if you think that’s the solution. There’s a whole host of culture, previous investment, how it’s viewed, how important it is. If something’s not working, [you need to say] ‘Hey, we can’t do that. Let’s move on.’ And that’s not a failure, that’s learning. A.I. has given us the opportunity to say, ‘That data didn’t meet what we intended it to do, that model didn’t work, let’s move on.’”
Information from Fortune Daily
Read Full Article Here: A.I.’s role in financial transactions needs the human element of ‘explainability’