This September, financial institutions, venture capitalists, well established businesses, and startups alike joined together in New York City for 4 days of demos, panels, keynotes and roundtable discussions navigating the financial technology landscape.
Of the 70 companies that demoed, AI dominated the discussion; with over 15% of companies insisting that “AI” is the driving force behind their tech. As the demos continued, it became increasingly clear that the disparities between those AIs are immense, and the extent to which their functionalities vary should not be overlooked. While Finovate 2016 outlined banks’ need to implement AI into their platforms, Finovate 2017 identified several key factors that matter most when choosing a virtual banking assistant.
Days 1 & 2 – Demos: The AI Differentiators
In financial services, an industry overwhelmingly saturated with competition, customer service is the one true differentiator. Over the 2 days of demos, 11 AI companies presented their products, showcasing their use cases in the fintech. From AI powered marketing software Optimove, to digital adoption platform WalkMe, the presenting companies had one shared commonality: an emphasis on delivering a positive user experience. And it’s not easy to understand why: according to a study conducted this year by FIS:
Only 23% of consumers are satisfied with their banking experience.
It became abundantly clear during the two days of demos that the necessity for excellent customer service is the guiding principle behind banks’ interest in implementing AI into their platforms. However, a bank can only provide a best-in-class customer experience through AI if the AI itself is superior. This leads to the next differentiator: superior tech.
While the AI behind the demos of investment services, data security and adoption platforms seemed to provide meaningful results, conversational AI is a much harder technological space to navigate. Of the two “conversational” bank assistants that demoed, neither used natural language processing to understand their user’s queries, and that’s a tremendous problem. According to an HSBC study conducted this year:
Twice as many people (14%) would trust a robot to perform heart surgery than open a savings account (7%)
A bank will only enjoy the benefits of implementing a conversational AI (increased customer engagement across channels, reduction of call volume to customer care centers, and increased customer satisfaction) if they choose an AI that is capable of understanding unbounded, natural language. Essentially, the virtual assistant a bank chooses needs to be more than just a chat bot. Voice recognition, and rule based approaches to AI won’t increase the current state of low consumer trust values with bots or provide the essential customer experience banks are looking for. In fact it will do the opposite, disenfranchising customers and leading to frustration with the banking institution that deployed the AI in the first place.
Seamless Omni-channel Integration:
According to V12 Data, on average:
Companies with omnichannel customer engagement strategies retain 89% of their customers, compared to 33% for companies with weak omnichannel customer engagement.
Why send your customers to an external app when you can deliver the experience of a virtual assistant right from your mobile app, Alexa Skill or Google Home? Banks don’t just desire omni-channel integration, they require it, and the AI vendors that demod who were incapable of doingso were quickly dismissed at this year’s Finovate.
Days 3 and 4: Panels, Keynotes and Roundtables
AI Thought Leadership
Artificial Intelligence didn’t just make its way into the Finovate demos, several panels, a keynote presentation and a showcase all centered around AI. Here are the highlights:
Panel: AI, Banks and Chatbots: The ABCs of Intelligent Financial Assistants
Keynote: Tim Urban: The Road to Artificial SuperIntelligence
AI Showcase: The Future of AI
AI, Banks and Chatbots: On Day 3, Jason Mars, CEO of Clinc, Hari Gopalkrishnan, Director of Client-Facing Platforms Technology at Bank of America and Darrius Jones, AVP of Enterprise Innovation at USAA sat down to discuss the role of intelligent virtual assistants for financial institutions. These leaders established three essential problems facing banks in an increasingly technological and simultaneously financially illiterate landscape.
Identifying the Problem
A Demand for Independence: The contemporary consumer seeks control of their finances without walking into a branch or contacting a call center. Banks responded to this need with mobile and web banking, but as Hari Gopalkrishnan noted during the panel:
“We are well past the point where a 5 inch screen can actually make it easy for them (the consumer) to navigate the experience of all the great things we have to offer them”
A (superior) virtual banking assistant should offer the user the independence they seek without the complication of navigating an app or website. The right conversational AI eliminates the frustration of 27 clicks and a phone call to the care center, with a simple voice query and helpful, human-like response.
Creating the Natural Language Interface: The number of consumers who enjoy their experiences with traditional chatbots is alarmingly low, if they’ve had the experience at all. According to HSBC:
69% of consumers either haven’t heard of or couldn’t explain the role of robo-advisors in banking.
Why is this? The people who do understand the applications for AI in banking have had bad experiences and don’t promote the technology. The challenge when talking to a chatbot or virtual assistant is that the interaction feels constrained, as if there is a correct and singular way the user must ask a question (and there often is).
Engaging in these unnatural interactions removes the trust, and the motive to use these interfaces in the first place. Identifying this problem, Clinc has made it their mission to take the best contributions from academia in machine learning to build a solution that represents the next generation of virtual assistants like Siri, and Google Assist and then putting it in the hands of banks to enable an experience that users trust and enjoy.
Communicating without Context: The goal for banks when implementing AI is unanimous: creating a digital platform that provides an experience that is better than the user would have with a human. At its core, this means, as Darrius Jones articulated during the panel:
“Creating an AI that can decipher meaning, comprehending 98% of what a user means while only recognizing 60% of what that user actually says”
How does a machine conduct a dialogue to solve your problem in a way that feels authentic and human rather than artificial? This is the problem Clinc seeks to answer, and one that other chat bots often miss. It isn’t just about saving time; it’s rethinking the digital experience and creating a human interaction where a human interaction doesn’t exist.
Keynote Presentation: The Road to Artificial SuperIntelligence
Tim Urban, Harvard graduate and the creator of the blog Wait but Why gave an illuminating presentation on the current status of artificial intelligence, where its headed and where it may end up. While the predictions of far-future use cases for AI were certainly exciting, more practical was his informative explanation of AI’s current abilities and its limitations. A transcript of the keynote can be found on Urban’s blog.
The Future of AI
AI Showcase: On the final day of the conference Tommy Marshall of Accenture, Audrey Solomon of WalkMe, Pini Yakul of Optimove and Jason Mars of Clinc showcased their uses of AI in banking, outling a roadmap for the future. Tommy Marshall explored the “Five Levers to Create Value from AI” arguing that “Intelligent Automation, Enhanced Judgement, Enhanced Interaction, Intelligent Products, and Enhanced Trust” are they keys to delivering tangible benefits to enterprise through AI technology. He isn’t wrong: according to a study from Time etc:
51% of Americans surveyed said they would feel uncomfortable sharing personal data with with an AI system.
Fostering public trust in AI platforms is a key component to leading the AI revolution, but it isn’t the only piece missing.
Jason Mars, CEO of Clinc, took an academic approach to his presentation, educating the audience on the latest advancements in conversational AI, and how they’re being used at Clinc to create banking experiences that feel authentically human, without the need for an actual human in the room.
The key takeaway from Mars’ presentation is that advancements in artificial intelligence are extremely new, and very few of those advancements are actually deployable. Given that fact, most “conversational” AI companies use rule-based approaches in their software.
What does this mean? They teach the AI how a human sentence is structured (a subject, a noun, a verb) and give it the vocabulary to recognize speech that is formulaic, and grammatically correct. But as humans, we don’t speak formulaically and we certainly aren’t always grammatically correct. This is why traditional voice assistants are so frustrating to interact with, because the majority of the time, they don’t understand our usage of language.
The only way to deliver an experience through a virtual assistant that feels authentically human is to teach the AI to understand context, not just to recognize words, but to understand their meaning. This is a much more difficult task, and the main reason chatbots don’t do it is that they don’t know how.
The AI behind Clinc’s virtual assistant Finie, uses artificial neural networks to learn the way a human brain does. Through natural language processing and deep learning, the AI not only understands natural language queries, it also gets better over time. It is this kind of conversational technology that is the future of AI, and Clinc is pushing the boundaries of what that future looks like, every day.