Slogans to promote the importance of voice and natural language interaction are everywhere. Many ask themselves, how to keep up with the trend and build a useful, profitable, cool chatbot on their own. The confusion starts with the term chatbot itself, since this is mostly not quite, what we are building, when we are into task/marketing automation or virtual assistants. Most of the so called chatbots are nothing more than just conversational agents and they are not meant to be anything else either. To interact in a dialogue with them has nothing to do with real organic conversation. And users‘ expectations seldomly get met. If we add “more AI inside” and “ML on top”, everything will get better – that‘s the promise. But in reality we‘re missing out on important efforts and simple strategies to turn a bot into a more intelligent product. Without these efforts and strategies, we can use as much AI as we want and won’t succeed. Why is it, that most of the first Turing test Winners came without any AI and strike us to perform better? They seem to be more empathetic and to understand more than most of the AI bots nowadays. What are we missing out on? Let’s have a look!
Why You Don’t Need AI to Build a Good Chatbot
As a former Senior BE Engineer, I led several projects and teams. I used to build Chatbots for one of the first and most innovative companies for voice in Europe, that is Alexa Dev Partner and I did this in cooperation with a Google Expert Dev for Assistant (one of 40 worldwide), which enabled me to always have the latest technological insights. I wrote scientific work about Best Practices and Design Patterns for Voice User Interfaces and I do a lot of knowledge transfer and workshops on the topic to enable people to build good bots and to prevent them to fall for marketing magicians.