AI is a hot topic in financial services. And its easy to see why. Increasing margins on transactions, decimated by compliance costs and low interest rates, reaching new market segments, and automating routine tasks, makes AI innovation attractive. And in one sense, FinServ has always been about algorithmic innovation. There is no higher potential ROI than beating the market. Advanced analytics for program trading have been banging away at this goal for decades, with a rich base of advances.
But AI in FinServ is struggling. We looked at the seven most important use cases, from AI-enhanced customer engagement, digital money management, decision management, to robo-administration. We found a low rate of adoption overall and found that different use cases are advancing at different rates because of maturity of AI technology, effect on the customer, and investment appetite.
And here's why. For one, few vendors and enterprises had an approach to chatbots that can really follow a conversation. Conversational intelligence that builds on IVR, voice, and web interactions that surround a financial transaction or an advice session, are poorly understood and not pre integrated into the platform components. Cloud based solutions from Amazon, Apple, and Google NLP platforms are not ready. They have no domain focus. Finnancial firms fear that using Siri or Cortana will open up their customers' data to the general web or worse that someday these tech giants will provide a better customer experience and disrupt their business. All are reasonable concerns.
Further broader access to data, for use cases like underwriting and risk decisioning, is needed, which will be difficult. Customers are hesitant to provide data directly to financial institutions. And broad use of social data raises questions. Also cited were severe data problems. Integration points need to be developed, data needs to be converted and cleansed, and relevant data from outside the organization must be accessible for use cases like AI engagement and decisioning. Sixty percent of AI leaders cited data issues as a top problem. And of course, skilled implementers or subject matter experts are in short supply. Too many projects depend on the Alpha developers.
And these were just some of what we found. The financial services knowledge base is even a bigger hurdle. Knowledge bases define the context, likely sentiment, and data to personalize an interaction to give advice, or complete a transaction. But our research found that few vendors had knowledge sets as starting points to build AI solutions. Certainly the major platforms - IBM Watson, Infosys Mana, Microsoft Azure machine learning, and Wipro Holmes - do not.
Henry Truong, the chief technology officer at TeleTech (a contact center and business process outsourcing [BPO] provider), put it this way: "The vendor approach to prospective users too often starts like this, ‘Give us $2 million, six months, and all your content, and we will build something.' To me, it's all about the knowledge base, and the vendors I speak with don't have one.
(About the author: Craig Le Clair is a vice president and principal analyst at Forrester Research. This post originally appeared on his Forrester blog, which can be viewed here).
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access