As financial institutions buy into AI, these factors will be key to successful transformations

When the Canadian government unveiled a world-first national artificial intelligence strategy, it was a signal for homegrown companies to start paying to this emerging technology, no matter the industry — and financial institutions heard the call. 

Financial institutions are particularly suited to benefit from AI, with massive amounts of data that can be used to automate processes and the resources to make solutions more secure. But what does success look like? That's the question MNP thought leaders and other panelists explored during a session on how AI is transforming interactions with financial institutions at FFCON's digital summit. 

The scope of AI transformations is enormous

“Financial services providers are adopting AI for everything from mundane task automation, to creating consistent customer services and experiences, diving deep into behavior analysis, as well as delivering efficient fraud prevention and detection,” says Massimo Iamello, an assurance and accounting partner at MNP. “They also need to manage the risks and concerns that arise along the journey of transformation.”

The rise in cloud technology has allowed AI applications to flourish, and financial institutions are taking advantage as they slowly migrate from on-premise legacy systems. Dev Mishra, MNP’s solution lead for data engineering, AI and machine learning, says the cloud has enabled businesses to start up AI initiatives more quickly — meaning more customer data can inform AI applications as it’s generated. In the past, businesses had to rely on manual surveys, which aren’t always accurate. 

Financial institutions are also making acquisitions for tech talent who can immediately put valuable customer data to use. MNP recently acquired T4G, a 25-year-old Canadian company that helps businesses collect, organize, and use their data for business insights and applications. 

“MNP has the base, the experience and employees with potential backing, while T4G is on the cutting edge with very senior and experienced consultants who’ve worked exclusively in the data space for decades,” says Mishra. “They were just lacking the scale and the firepower.”

Freeing humans for more complex tasks

Mishra summarizes the use of AI in financial services in three ways: adding new revenue streams, reducing operational costs and improving the customer experience. 

The growing accuracy of AI voice recognition means bots can now verify your personal information. Customers can speak naturally to bots about problems like password resets, which are easier to automate, leaving room for human customer service agents to answer more complex questions.

“The days of IVR [interactive voice recognition], where you have to press one or two, are gone; now you just talk naturally like you’d speak to a human being,” says Mishra. 

While AI can automate simple tasks, it can also be used to enhance security. As technology evolves, so does the ability of fraudsters, and Mishra says AI has strong use cases in fraud detection. MNP worked with a Canadian government body to automate auditing of employee insurance and travel claims, which reached up to 70,000 claims a year. The AI MNP developed could track anomalies like duplicate claims, saving the client up to $3 million a year and reducing time spent reconciling the claims for weeks to hours. 

Biased algorithms a growing concern

Mishra notes the importance of ensuring enterprises create unbiased AI applications. Machine learning, which uses data and algorithms to make predictions, requires datasets to train itself. If those datasets already have bias present, it could build an application that is itself biased. 

For this reason, many large companies have responsible AI frameworks to ensure they use AI in an ethical and responsible way. Mishra provides an example of predicting students’ SAT scores where gender is included as a variable for prediction. 

“Imagine you predicted a woman’s score is dependent on her gender, but did not factor in the same correlation for men. You’ve introduced a bias without knowing,” he says. “However, if you're predicting whether you have COVID, that is a medical use case. It is maybe more prone to males, so gender becomes a critical variable.”

Many banks may start to use facial recognition to verify the age of potential customers. If some groups of people aren’t adequately represented in a dataset, the AI could categorize people from some ethnicities are younger or older than they are. “It's very important for data scientists building these models to understand the models and remove bias,” Mishra says. 

Keeping security front of mind

For financial institutions thinking of exploring AI, Mishra says security, privacy and compliance with policies is critical. While major cloud providers like Microsoft invest billions into cyber security, financial institutions must also have the right governance processes in place. 

“Cloud as a platform offers security, but it’s the governance around cloud that will decide whether the application is secure,” says Mishra.

For more information on how you can leverage AI in your organization, contact Dev Mishra MBA, MBAN, BEng, at 416.596.1711 or [email protected]