Siri’s descendants may soon be able to help you improve customer service at your company.

William Mark, president of information and computing sciences at SRI International (best known for inventing and selling Siri to Apple in 2010), is now creating virtual assistants that can help customers perform complex tasks in banking, shopping, and business.

 

While assistants like Siri know a little about a lot of things, virtual specialists have a narrow main focus, but their knowledge in that subject is very deep, thanks to natural language processing and artificial reasoning capabilities.

“If you say to Siri, ‘Transfer $200 to checking,’ Siri will look up transfer on the web, and probably point you to Wikipedia,” Mark said. “But when you walk into a bank, you expect the person behind the counter to know a lot about banking and help you with your banking needs. That’s the vision for the assistants we’re starting to put together.”

SRI built a platform that makes it easy to build these specialized assistants in different industries. They could be used in any field that requires an interaction, Mark said–including for a retail purchase, a healthcare consultancy, or an insurance query. These assistant bots could live in a company’s app or website, or in a physical space, like a kiosk at a store.

SEE: The key to human-robot relationships is vocalization, says Carnegie Mellon

The goal is to improve interactions and the customer experience, Mark said.

“When you try to do anything complicated in real life, you’re going to be involved in a conversation of some kind,” Mark said. “When we design these computer systems, not only do they need to know about banking, but they also need to know how to have a conversation, where new intents are introduced, but they have to remember that the old intents are not satisfied yet,” Mark said.

For example, a conversation with a bank teller or a specialized banking assistant bot might open with the customer saying, “I want to transfer $200 to my checking account.” The bot would have to ask from which account to extract the money, and be able to answer other questions such as “How much do I have in savings?” while still remembering the original request.

“It’s deeper knowledge, and deeper conversations,” Mark said.

Banking bots

Mark’s smartbot tech is already in use in banking, thanks to SRI spinoff company Kasisto. MyKAI is Kasisto’s smart consumer finance bot that operates through messaging platforms like Facebook Messenger and Slack, and is currently used by Wells Fargo.

MyKAI understands 1,000+ banking terms and definitions, and can answer user questions about finance, said Kasisto CEO and co-founder Zor Gorelov. However, it offers essentially a “read-only” version of a user’s financial information–while you can ask it for the balance of an account, you cannot, say, make a transfer.

“We built a smart bot that acts as your digital personal banker, helping with your banking data, and helping you understand banking in general,” Gorelov said. For example, a user can ask MyKAI, “What do I do if I have too much debt?” and receive advice.

The bot has helped Kasisto understand how consumers interact with banking systems, Gorelov said. “The most important thing to learn is that virtual assistants are not apps–what people do in normal banking apps is not the same as the way they do banking in conversational systems,” he added.

Mobile banking tends to be transactional–a person might log in to check their balance, or transfer money. With the smart bots, people tend to do things they couldn’t do on a regular banking app, using the power of human language to ask questions such as “How much did I spend on Uber last month?” or “What was my largest restaurant transaction this year?”

The company plans to take the KAI bot to other industries as well.

Last week, two new banking bots hit the market: The Kore Smart Bot and the Personetics Savings Bot.

The Kore Smart Bot platform allows people to create, build, test, and deploy natural language processing-enabled bots that can provide employee communication within an enterprise, or customer-facing services, said Lindsay Sanchez, CMO of Kore. The bot can live in any channel, including Facebook messenger, email, SMS, or a company website or app.

Between 40 and 50 companies are working with Kore on bots. The company’s target segment is enterprises with over 2,500 employees.

“Lots of banks are tied down by legacy systems,” Sanchez said. “Since bots create a better engagement experience that is simpler and easier to understand, I think a lot of banks are looking at this as a way to continue to drive revenue and loyalty from what’s becoming the most important channel for them–online, as opposed to the branch.”

Sanchez recommends companies narrow down what they want a bot to do, and what channel they want it to live in, before purchasing one. She also says a bot customized to any company will improve with experience. “The best part about a bot is the more people ask, the more it learns, and better and richer it gets over time,” Sanchez said.

Advice for AI hopefuls

The biggest obstacle for companies building a virtual assistant is the amount of work and expertise it takes, Mark said. “We’re building a piece of software that already has a significant part of the solution,” he added. “Things like speech recognition, natural language understanding, and core reasoning capabilities are all built into the platform.”

Besides at Kasisto, Mark’s technology is currently being used to build other applications that Mark is not able to discuss yet.

“AI is a very fast-moving field,” Mark said. Progress in deep learning and core machine learning algorithms have led to amazing results in areas including vision and speech. The amount of data researchers can access today has also changed the field dramatically, he said, by providing insight for practical solutions.

“Things are changing dramatically, and we are just at the very beginning of AI becoming useful to people, and lots of real world applications,” Mark said. “It’s a great area to go into.”

He predicts that AI will be the hottest area to work in tech in the next decade. “It’s going to mean making systems that are easier for people to interact with, more helpful to people in more situations, and more trusted,” Mark said.

Mark said he expects a couple of trends to fuel this work, the first being the Internet of Things. “The things around us are becoming more and more computational,” Mark said. “It’s been going on for a long time, but I think it will rapidly increase, which means there have to be new ways of interacting with them.”

The other trend is increased autonomy in technology like self-driving cars, but also in devices in your home. “When that happens, we need to find ways to trust those systems more and more, which means we have to design systems so that their behavior stays within a certain range, or that they have the ability to explain themselves if they do something different,” Mark said.

The improvements in AI over the last 10 years will fundamentally change human interaction, Gorelov said. “Conversational UI will become the main UI,” Gorelov said. “People don’t want to buy more apps, they want to get things done and move on with their lives. The conversational approach is the most intuitive one.”

But worry not: The robot revolution will not come in the form of virtual assistants, Mark said. “People like to talk to people,” Mark added. “There’s a different experience when you’re having an interaction with a human being. I don’t see these things replacing that.”

The 3 big takeaways for TechRepublic readers

  1. William Mark, creator of Siri, is now working on virtual specialists, smart bots with lots of knowledge in a specific field such as banking, retail, or healthcare.
  2. The bots are able to mirror human conversation and help users through a process to complete a task, such as transferring money from one bank account to another.
  3. This technology is already being used by SRI International spinoff company Kasisto, with a smart banking bot. Two other banking-specific bots hit the market last week.