Many believe that the app bubble is bursting or about to burst. In some cases, the evidence shown for this is financial. The WSJ cites a stagnation in app start-up funding. In other cases, the evidence presented is based on user statistics, such as the fact that from April-June this year, 49% of US smartphone users downloaded ZERO new apps.
If the app bubble is truly bursting, people will need new ways of interacting with companies, trying out new digital products, and gathering information.
Enter the chatbot.
Chatbots have been around for a while; who hasn’t been routed through one on a phone call to a customer support number? However they currently lack the maturity of the app ecosystem because making a realistic and useful chatbot is a complex problem. Natural Language Processing (NLP) has reached a point where chatbots can be trained with limited conversational skills using deep reinforcement learning. Even with deep learning, though, NLP experts like Jason Baldridge are skeptical that deep learning will give chatbots human level competence in all aspects of a conversation.
This disconnect in people building chatbots and the chatbots’ capabilities can result in a user experience that feels canned, as if the chatbot is simply picking from a list of pre-made responses (and in many cases it is). In addition, many chatbots are simply not that efficient at collecting information from users. Apps have been optimized to make collecting information as easy as swiping, pushing buttons, and scrolling. Chatbots on the other hand require forming a coherent intent. And because many chatbots don’t support speech-to-text, they also require a lot of typing. While this may seem trivial, the information transmitted per screen press is much lower in an average chatbot vs. an average app.
This means that while chatbots are excellent at some things at which apps perform poorly, like pushing out important alerts, one-time uses, and interactions where a human element is important, chatbots may be bad at other things at which apps excel, such as gathering information, navigating menus, and playing games.
While current circumstances may call for tools in between the functionality of apps and chatbots (think Google instant apps), chatbots hold a place in scenarios where the human element is important, but scalability is a concern. Motivation building, remote therapy, and customer service are all examples of applications where hiring enough people to service millions of users may be too expensive, but a human element is important. Chatbot use will certainly grow in the coming years as the next billion users connect to the Internet, but they have a long road to go before they can replace the ease and design of apps.