Chatbots are alternatively being hyped as a great opportunity and a great disappointment. There is a plethora of options on how to build and design chatbots and a lot of platforms where they can run. To make things even more complicated there is a lot of confusion around the lines between chatbots and artificial intelligence and how the two intersect. This talk aims to clarify the range of concepts and give you a conceptual framework that will make it much simpler to guide strategic decisions around the application of chatbot technology.
We go through different approaches to chatbots development. The presentation consists of a few Jupyter notebooks with code samples. We show and compare a few examples of neural networks architectures that can be used for chatbots. Based on our experience and presented architectures we give some advice on which architecture should be used where.
Key Takeaways
Chatbots are replacing humans in many domains from online sales assistants to online learning. However, they can often be very stilted and mechanical in their interactions – how can we give a chatbot a more human persona? As with all systems, using a chatbot involves a level of mutual trust and this depends upon identity. How can the human be sure that the identity of the chatbot and how can the chatbot be sure of the identity of the human? This panel will discuss these topics