How FIs Can Train Human And AI Employees To Work Together

Whether in banking, security, or retail, there is a clear trend regarding artificial intelligence (AI) and machine learning: they’re on the rise, and their proponents are fiercely in favor.

Indeed, there is a lot of good that can come out of AI. In the area of fraud detection, for instance, it is much faster and more thorough than humans when weeding out fraud attempts from millions, even billions, of daily transactions that live employees would never have time to sort by hand, handing over only the most suspicious activities for manual review.

To remain competitive, many eCommerce players are hustling to implement AI solutions to protect themselves (and their legitimate customers) from the consequences of fraud, which include both financial losses and customer losses as increased authentication processes can lead to increase friction at the point of sale, creating delays and driving customers away.

Banks, as well, are being pressured to move faster and faster, yet provide ever-greater consumer protections at the same time — two gaps that AI can go a long way toward filling.

But the degree of efficiency these organizations envision is a goal — in some cases, an end without means. Along the way, training employees to work with new systems is not always easy — especially if they have to learn multiple new systems in a short span of time as their employer overhauls the organization’s functions and roles.

Tectonic Shifts in Environment

This is a challenge Rephael Sweary, co-founder and president of WalkMe, knows all too well. Sweary says there are four major shifts taking place which, in isolation, may be navigable for organizations and their employees — but with all four combined, the tectonic shift is knocking the feet out from under many financial institutions and online retailers.

First, the new generation of employees is not the same as the last generation. They grew up with technology at their fingertips, so they know how to access information, but not always how to remember it, Sweary said. Therefore, requiring memorization from employees is not going to create efficiency in the organization.

Second, although similar — the new workforce is a whole new beast. Employees are used to changing positions often. They are highly distracted, with a phone or iPad open beside their computer and social media running in the background. Their digital IQ is high, said Sweary, which can be an asset if used correctly, but their expectations are different, and employers much recognize that in order to milk the most efficiency out of their digital workforce.

“Years ago,” said Sweary, “you went into an office and only saw computers in the accounting area. Today, if you walk in and don’t see a computer on the desk, you think it’s not an office. ATM, CRM, ERP: it’s all digital, and that’s true more than anywhere in financial services.”

The third shift Sweary mentioned is decentralization. In the past, he said, decisions were made in the C-level suite and other departments followed suit afterward. Today, those same decisions are being made in the line of business manager. Furthermore, it is common to see a few different vendors working together in one company, although it is difficult to bring in too many due to increasing regulations and security concerns.

Fourth and finally, Sweary said migration to the cloud has been a critical shift — one which many financial institutions delayed and which they are now pursuing aggressively, he said, like a slingshot that has been pulled back over the past several years only to shoot them forward at breakneck pace.

“There isn’t just one digital transformation initiative,” Sweary said. “There are multiple, run by different departments. So one employee could get a change in the customer relationship management (CRM) system, then one in human resources, then one in how expense reporting is done, and then a new method of video conferencing, and then a change in how staffing is done – all within a couple of months. It’s rocky water — like the middle of the ocean, and the employee is sitting there in a small boat just trying to ride the next wave.”

Training People and AI to Work Together

To create a digital experience that guarantees adoption, said Sweary, a different approach is needed. Training was once about teaching employees to use software, but with AI, organizations can now teach the software to adapt to employees, offering contextual training and guidance rather than hours or days of training and testing.

Such an approach, just like cramming for the big test in high school, leads to information draining out of employees’ heads just as fast as it’s learned, Sweary said. It is much better to conduct training on the go, right in the work environment where the employee can apply the new skill immediately, hands-on.

According to Sweary, modern training must be two things in order to be successful.

First, it must be personalized. What is this specific employee’s role? What tasks and processes does he need to complete on a regular basis, and how might he be using the system slightly differently than a colleague in a different role? Artificially intelligent systems can be trained to adapt to the user, rather than forcing many different users to adapt to the same software, and organizations who wish to drive efficiency in their workforce should leverage that, Sweary said.

Digital personalization is not a stretch today, said Sweary, giving the example of the Amazon home page. Someone who is female, uses a PC, and lives on the East Coast has a completely different experience than someone who is male, uses a Macbook, and lives on the West Coast, he said. Training that is built on machine learning could be just as personalized.

Second, said Sweary, training must come “just in time,” not a week ahead of time. A week is more than long enough for the employee to forget what he’s been shown, whereas an anticipatory AI can make a good guess of what the employee may want to do next and offer a tip exactly when she needs it.

Sweary gave this example. While off the clock, one of his co-founders got a phone call from a colleague who needed help making a wire transfer — a process that was, indeed, part of her role, but not one she needed to remember often. Luckily, the WalkMe co-founder was able to walk her through the process over the phone. But a week went by, and she was calling him off the clock again with the same question.

An anticipatory AI could have answered that question for her before she even needed to ask. That’s where the idea for WalkMe was born: The co-founder began to envision a system to offer process walk-throughs to whoever needs them, exactly at the moment of confusion.

What Resistance?

Initially, selling product teams on such an idea proved challenging, said Sweary. The general sentiment was that, if users needed just-in-time guidance, then the product team was not doing its job correctly. Everything was supposed to be intuitive.

However, Sweary said that such resistance is waning. The teams who build AI systems are beginning to see that, no matter how well they do their job, some users are going to need extra help with certain difficult areas, and it is better to identify those and offer the extra help employees need than to pretend the system is completely navigable without any additional training or guidance.

AI enthusiasts often tout the time and cost savings that machine learning can generate for banks, retailers, and other organizations, and those savings are real and achievable, said Sweary. But there is no shame in setting up a guided walkthrough to help employees help the company reach that goal.