The Architecture of Algorithm-driven Persuasion

Abstract

Persuasion is a process that aims to utilize (true or false) information to change people’s attitudes in relation to something, usually as a precursor to behavioural change. Its use is prevalent in democratic societies, which do not, in principle, permit censorship of information or the use of force to enact power. The transition of information to the internet, particularly with the rise of social media, together with the capacity to capture, store and process big data, and advances in machine learning, have transformed the way modern persuasion is conducted. This has led to new opportunities for persuaders, but also to well-documented instances of abuse: fake news, Cambridge Analytica, foreign interference in elections, etc. We investigate large-scale technology-based persuasion, with the help of three case studies derived from secondary sources, in order to identify and describe the underlying technology architecture and propose issues for future research, including a number of ethical concerns