Chatbots have been around for quite a while now, probably longer than many of us realise. Despite this, it is only in the last decade that they have really gained in popularity among users and businesses.
Popularity is probably a misleading term in this instance as the chatbot is proving to be an extremely divisive technology. Though rather than loving or hating them, it is maybe more a case that people are either indifferent or, quite vocally, hate them.
So how does this technology cause such an emotional response to the extent that many businesses shy away from giving it a try? In this blog we will explore the origins of chatbots, what good and bad ones look like and try to understand why they are so divisive.
Let us start with a simple definition: a chatbot is a software programme that is able to interact with a human in order to provide information. In that way it attempts to replicate human activity and therefore is considered Artificial Intelligence in a broad sense.
Chatbots were actually “born” over 50 years ago. The first one was developed by a professor at the Massachusetts Institute of Technology (MIT) called Joseph Weizenbaum in 1966. The chatbot was called ELIZA and it was designed to demonstrate that communication between a human and a machine or computer was only trivial.
Using a relatively simple decision tree of questions, ELIZA was able to do a parody of an initial psychiatric interview but without any real understanding. Despite this, many users were convinced that the chatbot was intelligent! Possibly due to the lack of general access to computers in the 1960s.
Since ELIZA there have been many iterations of chatbots as the vision and underlying technology have evolved. It was not until 2010 that the progression of chatbots really accelerated with the introduction of Siri by Apple. An intelligent personal assistant that uses natural language processing to engage with the user, respond to their questions and learn from each engagement.
This paved the way for chatbots, also referred to as virtual assistants, that use AI to enhance their capabilities. Siri was quickly followed by Google Now in 2012, Cortana from Microsoft and Alexa from Amazon in 2014.
Alexa is arguably at the forefront of chatbot technology that is working its way into our everyday lives. The underlying technology is impressive and has many potential applications both in the home and across the business world.
Despite this success – chatbots still have a bad reputation with many people.
This is probably because, as well as the good, there are some really bad chatbots out there.
High profile examples include Lee Luda. Developed in South Korea, it was able to converse with a natural tone, but quickly had to be taken offline after it started using hate speech. A similar fate befell Microsoft’s Twitter based chatbot Tay that survived only 16 hours after it was manipulated into posting racist tweets.
On a smaller scale there are countless examples of chatbots being designed or implemented badly, leading to them ruining the customer experience they were supposed to be improving. Common issues usually relate to over-ambition in terms of the technology capabilities. For example, natural language processing that allows the user to engage the chatbot in their own words rather than selecting options. This is really exciting when it is done well but has the potential to make the interaction long and frustrating.
Even the good chatbots get criticism though. Often that is because expectations have become too high. Perhaps it is the name “chatbot” but users regularly believe they can put any question to a chatbot, no matter how badly worded or spelled and expect it to be able to respond. Limitations of the technology mean that this can really trip chatbots up.
As AI technology continues to develop so will chatbots. Consequently, they will become more powerful and more useful. In particular, AI capabilities such as Natural Language Processing (the ability for a chatbot to analyse a sentence and understand the intent of the user) and Machine Learning (the ability to learn from previous interactions and identify the answer that is most likely to be successful) are driving innovation.
As a result, future chatbots will be able to interact more intelligently with users but also be better at finding the right answer, quicker.
Beyond this, the next generation of chatbots will be agnostic of the platform. Rather than just sitting in the corner of the website, users will be able to engage, and have equivalent interactions, on the website, Facebook Messenger, SMS, WhatsApp etc. etc. Giving users the choice to engage on their platform of choice.