The agent hell: a scenario of worst-practices in agent-based software engineering

Notable advances in miniaturization, communications, artificial intelligence and agent-based computing enable us to imagine a world pervasive with technology. Everything - our cities, our homes, our clothes - will be augmented with a multitude of inter-connected computer-based systems. Each system hosting one or more software agents devoted to monitoring the surrounding world and delegated to automatically performing activities on our behalf. The point of our Agent Hell scenario is that exploiting ways to make agents more and more intelligent and autonomous, and the figuring out to what extent we can delegate agents to do our work is not enough. Equally important is the need to advance the discipline related to the engineering of such systems. This discipline needs to take on the role of both enthusiastically advertising the advances of agent technologies, and of studying the limitations and dangers possibly associated with them. It is our hope that the technological "heaven" just envisaged soon becomes a reality. However, our scenario tries to draw attention to the fact that without adequate engineering foundations and discipline, such a paradise could easily become a living hell.