Building artificial intelligence technology into a financial system is tricky—mainly because artificial intelligence technology is still in its nascent stage.
Moving it out of that stage, according to experts, means innovating a very specific sect of the technology, known as natural language processing (NLP).
“NLP is the way machines can mimic the understanding of language, figure out the intent of language,” Felix Candelario, global financial services solutions architect for Amazon Web Services, said during a panel at the SIFMA fintech conference yesterday. “There is a lot of back-end focus [in NLP] on things such as chatbots, fraud detection.”
In other words, NLP is the part of AI technology that allows ‘bots and virtual assistants, such as Siri and Alexa, to pull the appropriate data when a user says or texts: “Alexa, check my bank account.”
The main problem with integrating the technology, especially into complex legacy systems like financial services, is that language itself is a complex system.
“There’s a multitude of possible responses to a question, so we have to deal with that ambiguity,” said Marc Andrews, vice president for IBM Watson financial services solutions, during the panel. “We have to focus on coming up with the best answer as opposed to the correct answer.”
IBM, the world’s oldest startup, is beginning to employ Watson’s talents in more complex fields, such as financial compliance—an area where AI’s benefits have drawn particular interest among financial institutions.
“We’ve started feeding all the regulation into Watson, to work on managing that regulatory pain,” said Andrews. “So, Watson will be able to tell you what is obligatory in this area where FIs constantly need to adhere, and constantly need to deal with change.”