5 Chatbot Myths Demystified

The first chatbots were based on scripts, following a yes/no, if-this-then-that approach to conversations. For a huge number of use cases, this is still a perfectly valid way of holding a customer conversation, helping users get the required information, or to the right part of the conversation, quickly and efficiently.

Features like natural language processing, machine learning and AI can all be applied to bots as part of the service to make them smarter, to help them learn to hold better conversations. However, they are not necessary for a successful chatbot project, and they can make a simple project more complicated or more costly.

As AI services become smarter and easier to implement, bots can be upgraded or overhauled to offer a better service.

We’ve yet to come across anyone in a bar moaning that a chatbot took their job. And the technology isn’t likely to be making the front pages for wiping out a generation of careers. Instead, the key benefit of a chatbot is that it helps remove some of the basic parts of a receptionists’ or customer service agents’ tasks, freeing them up to focus on the important areas, and dealing with customers who really need their help and specialist knowledge.

Most larger customer-facing businesses are rolling out their chatbots alongside their usual customer support services, and those agents remain on hand to deal with questions the bots can’t handle. Smaller businesses can deploy bots to save on hiring in some limited areas and to help automate parts of the business the current workforce don’t have time to deal with. But we are a long way off if ever, seeing redundancies at scale due to a business launching a chatbot.

In a recent example, Marks and Spencer is moving its staff to the shop floor as a chatbot takes over some customer service duties.

Headline writers love a good story. And much of the feedback from early user surveys suggested that people were uncomfortable or frustrated using chatbots and didn’t get the information they were looking for, which made for the usual sensational headlines.

As with any technology, it takes a while for people to get used to it, and with the rise of Messenger and other chat tools as platforms, people are now more used to conversing online, both to other people and bots. Also, as more companies launch chatbots, people are used to seeing them around and become familiar with the technology.

Now, most potential users are over the getting-to-know-your-chatbot phase, they understand the benefits and limitations of bots and both acceptance and user satisfaction scores will rise both thanks to that and as businesses learn to create better bots that meet customer needs.

One of the benefits of a bot is that the creator can imbue it with some of the brand’s, business owner’s or department’s personality. While most people accept this, the idea that a bot should sound and act like a person is generally a bad idea. Aside from some welcoming pleasantries, users are talking to the bot to get information, not have a fireside chat or trade jokes.

One area where this is less true is in the charity and mental health area where bots can help tell a story or ask for more information than is usual. In these conversations, building a rapport with the person is a key part of the bot’s role and by giving the bot a greater sense of personality, people are more likely to open up, donate or feel engaged and get involved.

Chatbot training is a key part of launching any AI-enabled bots, as discussed in the first myth. However, script-based bots, while they require plenty of testing to make sure they work, don’t need to be trained with masses of data that your company might not have anyway.

The training of AI-powered bots using archived or live customer chat conversations help it learn the right responses and even offer new ones, based on what it learns. So, yes enterprise-class bots should probably be very well trained before going live to the public, but not all bots require such an effort.