Complex Conversational Exchanges with a Bot – Are we there yet?

In the age of Siri, Cortana and the Google Assistant in particular, which can make appointments and personal phone calls on a user’s behalf, customers have started demanding more from banks. Though Eliza, the first chatbots came in the 1960s, chatbots are a relatively new – decade and a half old – term in the corporate arena. However, now customers have started expecting for more from their banking partner. Conversational bots can provide you with solutions for your portfolio or provide information for your queries. However, customers want more. How much more? – We don’t know. Neither do we know about the banks readiness to take the next step into conversational workflows between bots and humans. Banks, however need to ask themselves – Are we there yet?

The Constant Evolution of Chatbots

Chatbots have evolved over the years and their definitions have been written and rewritten over and over again, across various technological dictionaries. The basic simplistic chatbots exist and provide customers with basic responses to questions like “Show me all of your loan plans or credit approval plans?” However these bots have a very limited knowledge base, where the workflow is restricted to two or three solutions, which might not exactly match with the particular portfolio of the customer. There are also those bots which can filter solutions based on particular parts of speech from the user. Questions such as “Show me loan plans above 1000 dollars and less than 10000 dollars for a period of 1 year” belong in this category. However we’ve come to the stage where, customers want intelligent robotic advisors or chatbots which can understand through the immense amount of data available on the exact solution for the exact portfolio.

Where we’re currently placed

At this stage in the banking sector, is chatbots the future? Well according to Grand View Research, the chatbots market is going to see a 1.25 billion valuation and a 25% growth rate by 2025. Banks need to figure out how to create intelligent conversations with their bots in such a way where, they can completely erase human interaction in most of the front office operations, which the customer will never recognize. 40% of customers don’t care if they’re attended to by a chatbot or by a human, as long as their issues are solved. So, when customers don’t mind the lack of human interaction and banks can benefit from a 30% savings in customer support, banks really have nothing to lose. However, only 20% of banks have employed chatbots within the banking sector.

Intelligent Conversations with Chatbots – How it Starts

So, how does the technology behind a chatbot work and what are the necessary attributes that a bank needs to focus on while implementing one? The first major step for a chatbot is to understand the input of the customer. This could be done in the form of different language processing techniques such as artificial neural networks, natural language understanding and different machine learning algorithms. Once the chatbot has understood the need of the customer, there come into factor the different responses between the bot and the human, which would require a different type of NLP process known as natural language generation (NLG).

The success rate for bots after having these NLP processes in the banking space, will reach up to 90%, giving it the highest success rate amongst all sectors, if implemented by banks.

Common Challenges

AI and customer data go hand-in-hand. One can’t survive without the other. Hence, the first and major challenge that banks will have to face would be Security. With the level of intelligence that is being incorporated into a chatbot, banks would be hoping for the most complex conversational engagements with customers being solved without the use of human resources. However, banks need to ensure that experience and security aren’t lost at the expense of the other. Banks would need to ensure that hackers can’t hack into the chat interface, which would otherwise prove to be disastrous for the bank. In a non-banking scenario, Microsoft’s chatbot Tay was hacked by a group which went to abuse customers racially.

So how can Chatbots be protected?

Chatbots could be protected through the different algorithms, end-to-end encryptions, two-factor authentication or even through self-destruction mechanisms, where personal and confidential data could be deleted after a certain point in time, or in the case of the threat of a hack. Gartner has even predicted that the use of passwords and tokens would drop by 55% due to the increase in the use of biometrics and novel recognition technologies.

Final Thoughts

Banks need to ask themselves on the efficiency of their front office operations, their expenses and the success rate of each customer query. Banks need to ask themselves whether they need to delve more into the customer data and whether their AI processes are up to the task. Banks need to ask themselves, whether they would find themselves redundant 5 to 10 years from now, with several non-banking parties providing segmented banking services and solutions. If the answer to all those questions is a negative one, then banks need some serious introspection, starting with their provision of chatbots in the front office.  

Follow Me