The future of privacy using conversational bots


  Aberer Karl

Two parallel trends are evident in today’s tech-driven world: the rapid deployment of Artificial Intelligence, and the growing use of conversational interfaces. A seamless amalgamation of these two developments is a conversational bot, or a chatbot. In this context, researchers at EPFL are working on the future of privacy using conversational bots.

The research assumes great relevance when we consider the current method of sharing privacy terms and notices adopted by websites and mobile apps. When you register on a website or install an app, you are presented with a lengthy block of text that you must accept before using the app or services on the website. It is impractical to pause and peruse the whole document, and you invariably tend to mechanically accept the notice and move on. In short, the purpose of the notice or privacy policy is defeated by the very way in which it is presented to the user.

In a paper presented at a conference in Denver last year, researchers Hamza Harkous, Kassem Fawaz, Kang G. Shin, and Karl Aberer proposed Conversational Privacy Bots (PriBots) as a novel and more effective way of delivering notice. The method involves a conversation between the user and a chatbot.

In contrast to the exiting method of delivering notices, PriBots enable users to set their privacy settings through an intuitive and interactive interface. Instead of having to navigate through complex app menus and technical jargon, users can interact with the PriBot in a natural conversational environment to determine the effectiveness of a particular preference and the implications of using that preference. For instance, to set the privacy level of your birthday on Facebook, you need to scramble through several complicated steps on both the desktop and the mobile app. Conversely, with the PriBot, all you need to do is ask a question and set the limitations accordingly. The PriBot can even read the intention of the user based on usage pattern, and then suggest privacy settings. In other words, the PriBot functions as your own privacy consultant.

PriBots have immense potential for the future, especially with the rapid development of advanced machine learning tools and natural language processing (NLP) techniques.

Further reading: