From robo-advisers to chat bots, new technology is transforming financial services?

Recently a point made by many in the conversation sparked by the article — is that Financial services firms are hoovering up tech talent to work on their own disruptive technologies. Two of those emergent technologies are robo-financial advisers and intelligent virtual assistants. To discuss the evolution of both, below are excerpts from the conversations with  Cynthia Loh of Charles Schwab and Ken Dodelin of Capital One. Cynthia Loh, VP of digital advice and innovation, Charles Schwab You recently published a report entitled 'The Rise of Robo.' What did the findings reveal to you about different segments of the investor universe? First, while it's a conception that millennials are today's robo power users, boomers actually see a lot of appeal. So we found that nearly half of boomers using a robo-adviser say that robo-advice is perfect for their life stage. Second, depending on what type of demographic you are, you see a different benefit for robo-advice overall. So people are using robos for their specific needs, depending on their demographic. Will robos replace humans? One of the really telling findings of the report is that over 70 percent of people want a robo-adviser that also has access to human advice. A big misconception about robo-advisers is that one, it's just a computer and you never can talk to anyone; and two, that when people are engaging with robo-advisers they don't want that human advice. That's actually the opposite of what we saw in this survey. You can use robo-advice to automate and leverage technology for the things that technology is really good at and really efficient at, but you can also access financial advisers who are CFPs [certified financial planners] to provide personalized advice. Where are we in the evolution of robo-advisers? I think we're pretty early in the evolution. Our report found that 60 percent of Americans see themselves using a robo by the year 2025. Does the market cycle change the demand for robo-advice? It will be really interesting to see what happens if there's a bigger market dip. You haven't really seen a significant downturn since the advent of robo-advisers 10 years ago. But I think there's a lot of folks using robo-advice because they like that it takes the emotion out of investing. So when you see volatile markets, it's actually a great time to leverage technology because there isn't that emotion involved. You oversee a group of tech-minded professionals at a decades-old financial services company. How do you attract them? I was in financial services for a long time, then spent five years at startups, then now have come back to the more traditional sector. The benchmark isn't other financial services providers anymore; it's really any consumer-facing business. It's: How do consumers behave across all industries — that's what our benchmark is. We as Charles Schwab can't just say we're better than X/Y/Z financial services firm in New York; we have to say we're just as good as an Amazon or an Uber. Ken Dodelin, VP of conversational AI products, Capital One In the evolution of conversational AI, what stage are we in? We're in the very early stages — the toddler stage at best. We launched a chat bot and didn't even know it. We were involved in conversational AI — at least our customers were — and we weren't. It went like this: We had an SMS-based fraud alert. So we send out a text message to folks when our machine-learning algorithms detect some suspicious activity, some abnormal transactions, on an account. And it's a very rigid message, it's like, here's the transaction and text back CONFIRM or DENY. What we found was that 15 percent of the responses were not 'confirm' or 'deny'; they were 'yes,' 'yeah that was me,' thumbs-up emoji, 'that was the baseball equipment I bought for my grandson.' And we didn't understand any of it. So here we had a conversational interface in play; we just weren't party to it. So we launched Eno in SMS as a reactive thing. Now with Eno, the customer gets the same alert and then if they respond 'yes,' 'yeah that was me,' or they just fat-finger it like all of us do all the time, Eno understands it. We've improved our understanding from 85 percent to north of 99 percent. That's important for a couple reasons. One is from a customer experience standpoint, you can imagine, you're in an elevated emotional state when you get one of these alerts, and your stress level gets compounded when the machine doesn't understand you. So there's a great customer experience win here. There's also a shared benefit for the customer and for us, which is that minutes matter when it comes to taking action on a compromised account, and the sooner we can correctly identify a fraudulent activity and take action, the better. So this has some pretty good business impact for us. Instead of teaching humans to behave like machines — "send us back exactly these characters," which doesn't work — we need to train the machines to understand humans. Sometimes we, banks, get trapped in this "we have something to tell you, now we're going to engage in a digital experience about that and nothing else." But human conversations don't go in a linear fashion. The idea that we might highlight to someone, "Hey, this transaction looked funny. Is it yours?" If they say, no, that's not mine, there's probably more going on in their head. Like, "I wonder if there were other fraudulent transactions on my account"; "Can I see my most recent transactions?"; etc. With Eno, they can just ask that. Is your chat bot replacing human employees? The dirty little secret about AI is that it requires a lot of humans. It requires humans to build and also to get data in the right format so that we can use these machine-learning algorithms. And so we, like a lot of companies that are focused on AI, have increased both the number of software engineers but also folks in the data science field and then supporting those in the data science field, to get our data into a place where it can be best utilized. What are some of the challenges facing the evolution of chat bots? Chat bots struggle. It's a huge change in customer behavior that certainly hasn't been a big bang of customer adoption. But conversational AI to me, and the way our team thinks about it, is more about using AI to bring things to customers, not waiting for them to come to us, and allowing them to take recommended actions in the moment in channel — not ask you to go log in somewhere and navigate through a site. One woman we brought in during our research, we introduced a payment reminder to her, and her reaction was less about the act of now being able to make the payment; the value to her was all about the mental release. It was, "This is great because now I don't have to worry about whether I've forgotten the day, I just know that if Eno hasn't texted me then I'm fine." That's not the product you build, but that was the value that was communicated. And I think we're going to have a lot more use cases that fit that mold. What's been the business impact of Eno on Capital One? We haven't published the impact, but I'll put it to you this way: The folks who manage fraud for us, when they say minutes matter, they mean it. There's a reason they prefer to text people than to email them, because the response rate is faster. The sooner we can take corrective actions on a positively identified compromised account, when you scale that up across 40-plus million customers, you're getting into sizable numbers of savings. How did you name Eno? There is an unfortunate stereotype that has emerged, which is that all AIs have to be female. And so we're accidentally falling into this pattern of behavior where the thing in your house that you bark out orders to is always a female. That does not reinforce the right behavior. So we were very intentional about the naming of our intelligent assistant. Ultimately we decided that a gender-neutral name was the right way to go. How else have you tackled bias? We use humans to do reinforcement learning — for our natural language processing, for example — and if you don't have a diverse group of folks doing that, you wind up with an intelligent assistant that understands the world as a 20-something college-educated white female sees it, as an early member of our team who fit that description said. And so we've been intentional about bringing a diverse group of folks in to participate in the training of the AI. How have you attracted technology-minded talent to a bank like Capital One? There is a fierce battle for talent, no doubt. At Capital One, we've had a lot of success, though. Great engineers want big, hairy problems to work on — problems that are interesting, that are cutting edge, and that result in products that a scaled population is going to see and benefit from. And they've found those things here.   Data source: LinkedIn Professionals