School of Computing Research Colloquia

Diversity in Action: The SmartSociety perspective


Michael Rovatsos, University of Edinburgh


SmartSociety is a major EU-funded initiative that investigates diversity-aware collective adaptive systems composed of humans and machines. The project is driven by the vision that in order to build massive-scale collaborative systems that can help solve hard societal problems, machine intelligence should complement human intelligence in ways that respect human values and human diversity. In this talk I will give an overview of the work of SmartSociety, focusing, particularly, on diversity in human decision making and how we can build AI algorithms capable of accounting for this diversity.

I will discuss the fundamental challenges involved in balancing personalisation and global optimisation in large-scale, multi-stakeholder systems, and present some initial results in terms of concrete task recommendation algorithms we are developing.

The work I will present mostly draws on techniques from multiagent systems, decision-/game-theoretic AI, and recommender systems, but also touches on human factors, distributed systems, and machine ethics issues.


Michael Rovatsos is Director of the Centre for Intelligent Systems and their Applications and a Senior Lecturer in the School of Informatics, where he leads the Agents Research Group. His research is in multiagent systems with a particular focus on methods to avoid and mitigate conflict among agents with divergent objectives or beliefs. This focus has recently shifted toward human-oriented AI, both in terms of building intelligent systems to support human collaboration, and in developing human-inspired methods that enable artificial agents to learn to communicate like humans.

His work in these areas has attracted £2 million in external research funding and led to over 80 publications since 2000. He received his PhD in Informatics from the Technical University of Munich in 2004, and his first degree from Saarland University in 1999.