Page 28 - The EDIT | Q3 2017

28
Discovery
At the moment, AI is still not ‘intelligent’ enough to be able to understand and process very complex queries, and although NLP (Natural Language Processing) is evolving rapidly, we are still far from having deep conversation with a Chatbot.
“In my opinion, Chatbots should not over-promise. It is critical to build up the ’realm’ and capacity right from the beginning so the user always knows what to expect.”
Moving forward, it is also essential to design the learning curve of the bot. It is another (common) mistake when the tutoring is not properly designed and the Chatbot does not benefit from other data
sources. We call this ’isolation’, where the Chatbot acts almost like a separate entity — and because of the poor data architecture, it does not make any use of all the other data that businesses nowadays have access to. A cloud data warehouse is a great way
to overcome this. This is a solution which basically collects all relevant data in one place, acting like a textbook from which the Chatbot can learn from. This inclusion is very successful and improves the accuracy enormously, while being a good example of how much more lies beneath the surface.
What is being done in the realm of Chatbots to make them feel more life-like, so that users feel like they are interacting with a person? Is this in fact an intention of Chatbots?
Up until a year ago, everything was around making Chatbots more ‘human-like’ and making them feel as though you are talking to a real person. But, if
THE EDIT ISSUE 7 | Q3 2017