Banking And AI: Why We Also Need The Human Touch
Despite investing enormous amounts in people resources to prevent money laundering and terrorist financing and comply with regulations, banks have paid approximately $320 billion in fines over the last ten years alone. Cue artificial intelligence (AI) and machine learning, the latest technologies promising financial institutions a way to outflank criminals in the world of digital finance.
I listened to some interesting fintech scenarios during the launch of the SAP Next-Gen Innovation Community for Financial Services at the SAP Leonardo Center in New York City, and one of the most impressive was from Surendra Reddy, Founder and CEO of Quantiply. The California-based startup is infusing AI into its software to help banks address financial crime, risk and compliance.
“Working with SAP, can we bring machine learning and AI to augment investigators so they can proactively stop activities before anything happens. Unlike humans, machines can see patterns in minutes not hours, and they never sleep,” said Reddy.
Find the best people and algorithms for the job
Reddy is quick to point out that algorithms don’t eliminate people. Technologies like AI and machine learning accelerate and scale what were tedious processes, freeing up people for what they do best – applying judgment that ferrets the good from the bad.
“Rule-based systems may not catch certain signals, wrongly flagging good customers while bad actors pass right through. Mistakes like this burden customers, hurt business relationships and cost the bank fines.” said Reddy. “People can examine false positives, false negatives and anomalies for smarter decisions.”
Transparency and speed are the watchwords
AI solves one of the biggest challenges banks have – pulling together and analyzing scads of information from terrorist watch lists, sanctioned countries, layers of shell companies and beneficiaries, and comparing it to millions of transactions worldwide. It’s time-consuming, mind-numbing work that people hate and machines love.
“Having humans perform these processes is a cost-ineffective process for banks that are dealing with spiraling costs, spikes in cybercrime and increasing regulations,” said Reddy. “Using AI they can simplify compliance to regulations like the GDPR in the European Union. Banks need complete control and transparency of customer data.”
20x performance improvements with SAP HANA
A member of the SAP Startup Focus program, Quantiply has been working with SAP since the inception of SAP HANA. The Quantiply Sensemaker suite of cognitive applications uses the SAP HANA in-memory database and predictive analytics to deliver real-time data that addresses some of the most important risk and compliance issues banks face, including Know Your Customer (KYC). Reddy said that running machine learning algorithms on SAP HANA speeds up performance by a factor of 20.
“Our solution deploys multiple cognitive agents to search signals that catch possible fraud early with real-time updates,” said Reddy. “The agents are continuously learning from human feedback as well.”
Banks must think long-term
While people versus machines arguments grab headlines, the most successful banks will think about machine learning beyond short-term automation benefits.
“Knowledge workers carry a lot of information in their heads that walks out the door when they leave the company. Machine learning institutionalizes the memories of the experts, making that information available to any employee for lower retraining costs, especially the next generation of workers,” said Reddy. “Even when certain tasks are completely automated, new, higher-level responsibilities for humans are opening up.”
Follow me @smgaler