CUs and AI
Artificial intelligence is not some Hollywood fiction – it is happening right now, according to Rojin Nair, general manager of fintech for Calgary, Canada-based Celero Solutions.
“Today, we have intelligent automation. The next step is autonomous systems,” he explained during a breakout session on AI. “Today we have artificial general intelligence. The next step is artificial super intelligence.”
The AI panel at the 2018 NACUSO annual conference included, from left, Rojin Nair, Celero Solutions, John Best, Best Innovation Group, Scott Daukas, TwinStar CU, Srinivas Njay, Payjo. Standing behind Brian Lauer, partner with Messick, Lauer & Smith, moderator
Artificial general intelligence goes past machine learning to deep learning, Nair continued. He said most of what a loan officer does at a credit union today can be repeated.
“Consumers want to get financing where they buy. They do not want to be at a dealership ready to by a car, then have to go back to the credit union to apply for loan. They want financing at dealership.”
Nair predicted machine intelligence will surpass human intelligence by 2025. He said with heavily regulated and controlled intelligent systems, human beings and machines can co-exist.
John Best, CEO of Best Innovation Group, Tampa, Fla., said contrary to popular opinion, AI is not exploding due to Moore’s Law – the prediction that computing power would double every two years.
“It is because of data,” he declared. “We have so much data.”
According to Best, AI at a credit union eventually will run through a master AI engine that sees all data. Based on past behavior, this AI will make intuitions. “Autonomous banking will be just like autonomous cars,” he said. “The first step of AI is mastering analytics. Step one is to be analytical, then predictive, then finally prescriptive – can you affect consumer behavior.”
Scott Daukas, chief risk officer at $1.2 billion TwinStar Credit Union, Lacey, Wash., said he is paid to worry about everything that affects the credit union. “So I worry about the risks connected to AI. Many of these risks can be mitigated.”
Daukas said the big question is: with massive amounts of data, who owns the data? He said the problem with “black box” decision making – meaning autonomous decisions made by AI is – will decisions be made that humans cannot audit?
“AI might make decisions that inadvertently discriminate by race or gender,” he said. “There also are systemic risks: what if the AI company fails? Then there are credit union-specific challenges: Is your credit union ready, from a governance perspective, for AI? We have to be ready for the cultural changes AI will bring.”
Srinivas Njay, founder and CEO of Payjo, Inc., San Mateo, Calif., said his company has created an AI-powered banking assistant.
“It is a platform to help employees make decisions to help consumers,” he said. “The best platforms will be language and channel agnostic, supporting multiple languages by voice and text.”