Enabling decentralised decision-making in AI – an agent environments perspective

Should we be concerned about how to build decision-making platforms that are decentralised ? If so, how can we build them to support people make better decisions ? While some in the mainstream decision-making community dismiss these concerns, I will argue instead that a fundamental re-work of the practical aspects of decision-making is required to take these concerns into account. Instead of building systems that simply optimize a set of objectives, we need to learn how to build systems that will, in fact, interact with other systems and take decisions side-by-side with humans, so that the combined decisions can be beneficial to us. To address these concerns, I will use the idea of events processing in agent environments to show that it is useful to imbue systems that support explicit mechanisms connecting the decision-making process with the environment in which it takes place. I will then explore how by decentralising these mechanisms we can create deployment platforms for decision-making applications that can, in turn, benefit the humans they are designed to help.