SANTA CLARA, Calif. – Largely triggered by Google’s demo showing Google Assistant imitating human voice to book a hair appointment (see here; it’s pretty impressive), the chatter on Day 3 of FinovateSpring here was all about conversational banking.

Many banks are adopting conversational chatbots, but there is still no equivalent of Amazon’s Alexa or Google’s Assistant in the banking world — yet. Banks like Bank of America with its Erica chatbot and Royal Bank of Canada with Nomi are making early efforts in this arena, but they remain efforts.

None of the voice chatbots have reached the sophistication required to make them reliable, frictionless banking channels. One major problem is that the conversational banking is not inclusive, Theodora Lau of at Unconventional Ventures, said on a panel.

“If it (the chatbot) cannot understand my accent, it becomes a burden,” Lau said.

No customer is going to want to repeat a sentence numerous times, only to have the bot tell them, sorry, it cannot go through with the requested transaction, because it cannot understand what the user is saying. Sure, there are creators of bank-focused chatbots that are able to understand different languages like New York-based Kasisto or Michigan-based Clinc, but whether they can detect accents is another matter.

Security Concerns

Then there is the issue of unnecessary authentication. As Windsor Tanner, vice president of voice biometrics for realtime authentication at the compliance and security firm Nice Ltd., asked, “Why am I being asked to provide the last four digits of my Social Security?”

Tanner, who was also on the voice banking panel with Lau, added, “Why am I being asked my date of birth or mother’s maiden name?… That’s information that’s available on Facebook or my social media sites anyway. It’s not that difficult for a fraudster to find that out and authenticate as me.”

Security should not be a burden to the customer, Tanner added.

Speaking of security, there are new ways a fraudster can hack a user’s accounts using voice. Yesterday, The New York Times reported that researchers discovered Apple’s Siri, Amazon’s Alexa and Google’s Assistant can receive “hidden commands,” that aren’t audible to the human ear. The article stated that researchers were able to:

[S]ecretly activate the artificial intelligence systems on smartphones and smart speakers, making them dial phone numbers or open websites. In the wrong hands, the technology could be used to unlock doors, wire money or buy stuff online — simply with music playing over the radio.

“But why use your phone carrier to do voice banking in the first place,” CRMNext CEO Joe Salesky told Bank Innovation. “Why isn’t the voice banking feature built into a bank’s mobile app?”

This, according to Salesky, will allow a two-step authentication without burdening the user: first authentication is when the user logs into her smartphone and the second is when she logs into her banking app to get to voice feature.

“Not a bad idea,” one banking executive told Bank Innovation, but banks have not yet fully figured out how to use that technology.

Banks Treading Carefully

“There’s a reason why Amazon and Google are bragging about Alexa and Google Assistant,” Salesky said. “It’s because that’s something to brag about, everyone else is still trying to figure out how to do it, and they already are doing it.”

Ultimately, conversational banking must predict a customer’s need and fulfill that need without making a fuss about it. Lau refers to it as ambient banking or background banking.

“I don’t want to be thinking about my bank,” she said. “I don’t want to be thinking how and when do I need to pay my bills.”

What she means is that people want their banks to be able to resolve their specific needs. For instance, Tanner said, when a customer asks, “What’s my balance?” the voice banking feature needs to be intelligent enough to realize that the customer doesn’t just want to know her balance, but is also asking what is her safe-to-spend amount based on that balance and bills that will be due soon.

In order to achieve this level of advancement, the artificial intelligence used by the application needs to access and understand specific data. Conversational banking or even other AI-driven styles of banking can only be as good as the data available. And then that data needs to be structured and cleaned into a format that can be fed and understood by the AI engine and translated into an action — all in realtime, Carolyn Criscitiello, head of digital payments at HSBC, explained to Bank Innovation.

When it comes to data, banks still remain wary of what data to use and where to use, in fear of any of legal implications. The last thing a bank wants is to be involved in a privacy brouhaha like Facebook or Equifax. So while products like Alexa and Google Assistant might be trailblazing voice robotics to help their customers, don’t hold your breath for your bank to do the same soon — although be prepared to hear many banks talk about it.