In 2017, Google’s DeepMind Artificial Intelligence famously proved how well it understood (and mimicked) human intelligence by repeatedly beating the world’s top players of the hugely complex board game, “Go”.
In 2018, it is the turn of human beings to return the favor.
This is the year that we will refine our understanding of AI technology and begin to appreciate its diverse components and capabilities, central to which are a data and analytics foundation, machine learning, deep learning, natural language processing and generation, and visual recognition. Having gained early experience in using AI, in 2018, enterprises will talk specifics when discussing their future plans.
When it comes to applied solutions of artificial intelligence, some will fare better than others. Robotic Process Automation, which is entering its fifth year and is therefore quite mature – one survey claims 34 percent adoption in financial services1 – will continue to attract interest. So will machine learning and its subset, deep learning, which received 60 percent of AI investment in 20162. Leaders in adoption, such as ICICI Bank with more than 200 business processes already automated today, will bring more under the coverage of robotic automation, while the fence sitters will finally hop on to join the early adopters and fast followers.
Natural language-based applications – chat bots, smart assistants etc. – will also find their way into banks that don’t have them yet. There are two factors in their favor – progressive banks have given NLP and NLG the thumbs up, and vendors have built a substantial domain-specific knowledge base that gives latecomers more confidence to adopt. In 2018, FinTechs in the AI space will also look to collaborate with banks more effectively for customer data, an asset they chiefly lack.
The flip side of AI maturity is the new concerns that will emerge along with new opportunities. Machine learning is widely used, but there is limited understanding of how it arrives at a result – a credit score, for example – or why it forecasts one thing and not the other. What complicates this is that the bank will not be able to validate the result for at least a year, the typical length of a credit cycle. So in 2018, along with embracing the potential of AI, banks must also beware of its pitfalls.
This is the year that use cases will multiply beyond the established applications in risk management, fraud prevention and customer service. For example, in the era of PSD2 and open banking, when banks will have unprecedented access to customer information, they could AI to accurately forecast future spends based on a customer’s activity dating back several years. They could also process massive amounts of information to identify trends and their implications, such as the impact of retirement of Baby Boomers on the workforce, years in advance. That being said, cyber security will still be the foremost use case for AI in banks in 2018.
Along with increasing their reliance on AI, banks will combine it with human intelligence resources – experience and expertise – to arrive at a point of “right” intelligence, from where they will offer the right product at the right price on the right channel at the right time to the right customer, while staying on the right side of the regulator.
Application of AI has matured and seen wide adoption in areas such as biometric authentication, fraud detection and customer service. However, there are several new areas where AI can have significant impact. In 2018, progressive banks will further explore how AI can be applied in these use cases, which may be new for banking but proven in other industries. These are non-traditional credit scoring, documents classification, product recommendation and enhanced automatic trading among others. Indeed, AI will prove to be a huge differentiator for banks that understand the technology better and invest early.