In order to avoid confusion: there are two types of chatbots. The first is the ‘traditional’ bot, which works on the basis of preprogrammed rules. So if a question is put to this chatbot which is not preprogrammed, the bot can’t help you. The second type is learning, smart chatbots, on the basis of AI or Machine Learning. The strength and possibilities of this second type of chatbots are increasing all the time.
The ‘human touch’
Not everyone enjoys talking to a bot. There is a debate going on about the extent to which chatbots can replace the ‘human touch’. For this reason, chatbots are an interesting subject for discussion in the context of the ‘Turing test’. A machine, such as a chatbot, which passes this test is able to mimic human intelligence. Are chatbots going in that direction?
There is no definitive answer to that question for the time being. But there are already functionalities which can bring a chatbot closer to the human touch in terms of customer contact, particularly in combination with AI or machine learning. Three aspects which show that chatbots can make customer contacts stronger:
1) Standard questions and speed
Customer service employees spend a great deal of time answering standard questions. Chatbots are very suitable for taking over these tasks. At ING, it is now possible to provide customers with help immediately if something goes wrong with a bank card while abroad. That takes a great deal more time through the helpdesk. When a chatbot is used, the customer service department has more time to deal with complex problems or questions. This allows them to pay more attention to the human touch when this is really needed.
Sometimes you would prefer a service desk employee to be sitting next to you at all times while you are carrying out a complicated activity for a company. Chatbots make this possible. For example, the chatbot of Eneco helps customers submit their meter readings. In the same chat screen, customers can submit their readings and get answers to their questions about them. This makes a complicated assignment a lot easier.
2) Measuring mood and language recognition
This technology is still in its infancy, but thanks to machine learning methods and AI, chatbots are increasingly able to better measure people’s moods in conversations, and recognise language. The technology for this is already being used on a large scale. All developers can already make use of this intelligence by getting their software to tap into services from larger players who are developing this technology and making it available, such as Google or Microsoft. In this way, these developing technologies are already being integrated. And the human touch is gradually getting within reach.
3) ‘Talking’ to software
Another way in which a chatbot adds the human touch is at a moment when no human being would normally be involved. It is possible to get into a discussion with software via a chatbot. This can be extremely useful if a customer needs help and can literally ask the system a question. For example, when configuring new systems, it can be more pleasant to get help and instructions in a ‘natural language interface’ than by looking through a help menu for the answer. This is especially useful for less frequent users.
In short: a chatbot is more than a glorified answering machine, particularly in combination with AI or machine learning. The technology is offering more and more possibilities in this area. In the debate about chatbots, it is probably better not to think in extremes: when chatbots cannot provide the answer, human service desk employees can jump in. So it is not a question of either/or, but of both working together. When the human and the machine can work together in the right way, customer contact through chatbots will only improve in the future.