Sophia Warwick, a solution architect at process transformation company Sutherland explained that in a regulated environment, processes must be controlled allowing no room for ‘creativity’ from the digital entity.
“Robots learn from the human experts and there is just no reason to harness the risks associated with artificial intelligence. The learning process takes too long from a standing start, where a process needs to perfect to deliver the right information in exactly the right way,” she added.
She added artificial intelligence could be used for less risky, high volume transactions ‘with a long tail’ like travel agency advice where it’s ‘not financial suicide if someone picks the wrong hotel.’
The ideal use for AI is in a middle space as an academic library for a human expert, like a doctor, she said, enabling them to filter the latest academic studies on a particular ailment, but the human still makes the decision.
“Technology has to earn the right to be valid,” added Warwick.
Understanding the terms
As mortgage technology evolves and the industry stretches to apply digitisation to traditional patterns of work, confusion can reign over new terms and their exact meanings.
Dynamo’s chief operating officer, Jenny Watts said artificial intelligence and machine learning are closely-related but different.
“AI is the science behind making computers smart and machine learning is built on the algorithms used to encourage machines to think like humans.
“Whether anyone is using true AI or machine learning in the mortgage industry is up for debate,” she said.
“But the opportunity it presents our industry is exciting – in time it could be used for all types of automation of customer communications, voice recognition, fraud prevention and to match the preferences of our customers with our mortgage advisers,” she added.
Some may be in a hurry and want to get straight to business or have time on their hands and want to have a chat and ask lots of questions, said Watts.
Will AI be useful in the future if not right now for mortgage customers?
Watts said the technology is already here and it will be useful at some point to remember customer data from a previous application or from an earlier stage in the mortgage journey saving consumers time and hassle.
Chatbots are also able to offer more realistic and naturalistic exchanges thanks to natural language processing so true robo advice is further away but is already technically possible, added Watts.
“We can still use AI techniques in a regulated industry, as long as we understand and abide by our obligations to the consumer in doing so – for example, GDPR is very clear that any automated decision-making must be transparent, and it must always be possible for a human to override it or a data subject to challenge it,” said Watts.
Dynamo is already assessing the practical application of AI to its mortgage advice business and its advisers. Dynamo is the first online mortgage adviser to be accepted by the FCA’s Advice Unit which provides regulatory feedback to firms developing automated models to deliver lower cost advice and guidance to consumers.
Human training, monitoring and intervention remain critical to the advice process but how long can that continue?
“Technology should be used to take away all the number crunching and data analysis and leave the person monitoring the mortgage advice risk to interpret the data and frankly look for instances that just don’t stack up or pass the sniff test.
“At the moment, so much of this is still managed by spreadsheets which means there is huge scope for technology to support the compliance and risk professionals who work really hard to ensure good customer outcomes, said Watts.