There’s something about the idea of talking to a machine that has captured the imagination since the first such software was trialled in the 1960’s. In 2020, 2 out of 3 online customers engaged with some form of bot and this shows no sign of slowing down. However, when customer care managers think of chatbots, they often get it wrong. Instead of creating a solution that works for their team and their customers, they get hung up on decisions that are inherently superficial such as what the bot’s name should be. They should rather be focusing on the nuts and bolts of what it’s designed to do.
The History
To understand the power of chatbots, it’s worth briefly revisiting how they were developed. Some of the myths I’ll discuss have a direct link back to how the technology evolved and, without this context, it can be easy to repeat previous mistakes.
The first chatbot is widely accredited to Joseph Weizenbaum in 1966, meaning they’re already over 50 years old. Named Eliza, this bot worked in a remarkably similar way to much of today’s technology. It used a form of Natural Language Processing (NLP) to take words entered by a user and associate them with a relevant, scripted response. The most basic of today’s bots work in exactly the same way, with the main change being the medium of entry (often social customer service) for the customer.
Following the development of iterative technologies – including Parry and Jabberwacky – a big milestone came in 2001 with the development of SmarterChild. For people of a certain (read: my) generation, SmarterChild was our first interaction with bot technology. An instant–messenger–based bot (available via AOL, MSN Messenger, and other platforms), SmarterChild was a contact method to ask questions and request data from a system which – at that time – was relatively rare at a mass scale.
SmarterChild worked in a very similar way to Eliza back in the 60’s. Users would ask questions which were processed in the form of keywords, and SmarterChild would return a response. One of the features of modern NLP is the ability to learn; as bots answer more questions, they can understand more meanings/contexts from the user’s language. Using this framework, SmarterChild was able to improve its accuracy throughout its lifespan. Although, all too frequently, it struggled to provide a valid response.
In many ways, SmarterChild can be seen as the direct predecessor of many of today’s leading bot networks including SIRI, Alexa, Google Assistant, and more.
The Myths
From SmarterChild in the early 2000’s, it’s now time to jump forward to the present day and the way bots are used in digital customer service. Firstly, an important point: when used right, bots are a key part of providing a high–class customer service solution. The following three myths are a guide to focusing in the right areas, not ignoring bots altogether.