Will regulatory compliance drive artificial intelligence adoption?
With their vast warehouses of structured and unstructured data, and the competitive pressures they are facing from nimble technology companies, financial firms may seem like prime candidates for artificial intelligence (AI) technology. But firms could be attracted by another driver to AI and cognitive technologies: compliance.
"One incredibly important aspect of a digital agent, this is the thing that is drawing insurance companies to us, drawing the financial services companies to us, and the healthcare and medical services providers, is regulatory compliance obligations," said Jonathan Crane, chief commercial officer for IPsoft.
Research firm IDC predicts that "cognitive software," which includes artificial intelligence and other tools that offer intelligent advice and assistance, will grow at a compound annual growth rate approaching 35 percent over the four-year period from 2015 to 2019.
In financial services, and other heavily-regulated industries, firms that are struggling to train large, dispersed workforces to comply with continually evolving regulations are starting to see the value of cognitive technologies to address compliance obligations, Crane said.
"Have a digital compliance officer input that change in the regulations that just occurred, passed the legislature, went through Parliament, went through the US Congress, whatever it is. You input that into the process ontology for 'Amelia' and she cannot make a mistake," Crane said, referring to the name of IPsoft's artificial intelligence-based digital agent solution.
Artificial intelligence-based digital agents, like IPsoft's Amelia, incorporate semantic learning, which includes the rules and knowledge that is programmed into them, but they also enhance that semantic knowledge base with on-the-job or "episodic" learning. If Amelia is working with an insurance customer over the phone, for example, and the customer has unusual circumstances, Amelia would conference a live agent into the call, Crane explains. However, if a second customer would call in with similar unusual circumstances, Amelia would recall the types of questions the live agent asked. As a result of this episodic learning, Amelia would be able to handle more of the call, and not have to call a live agent to assist until later in the process.
"She learns from interfacing with the public as to what are their concerns, what are their likely questions, how do I solve those, where do I get that information?" Crane said.
As Amelia learns where to find answers to a wider array of questions, those answers are extracted and become part of her database, he explains.
Amelia was first launched late 2014, and after several trials, version 2 was launched this year. The new version includes enhanced episodic learning capabilities as well as an avatar to give customers a life-like visual image to interact with.
Embedding compliance regulations in Amelia was a concept that came from customers during trials of the product. In financial services customer service is another use case, both as a front end or as an additional advisor alongside an agent, particularly from banks looking to become more digital.
"We are going to see a lot of changes come about now from the digital company concept, and digital companies require automation to function," Crane said. "The new concept of digital labor, or software-defined labor, is a new thing for people to get their heads around. We are just beginning in this new phase of using technology like this to change the way we do business."