Human Agency on Algorithmic Systems

Abstract

The majority of large web service platforms promote service personalization as a means of providing user convenience. Parameter settings are commonly inferred from user behavior rather than explicit requests. As a consequence, users have little direct control of the system behavior. In this paper we discuss the implicit trade-off between user convenience and agency over the online experience. We will present results from multi-stakeholder engagement workshops exploring concerns and routes to solutions as raised by participants from industry, civil-society, teachers and academics. Finally, we will place these results within the context of the wider debate about User Trust on the internet and specific efforts to develop a standard on algorithm bias considerations.