Automation in IT Service Delivery: Human Factors Challenges

Throughout the session, each panelist introduced some element of existing knowledge or techniques for improving the human factors of automation. These elements included contrasting semantic-based planning versus policy-based planning, management of increased interaction complexity caused by automation versus simplification, and abstraction and simplification of system
interfaces.

The panelists generally agreed that, to this point, there is no grounded theory that provides a framework for human interaction with automated IT systems. However, it is evident that a lack of system transparency, observability, and resilience are issues that need to be addressed for the most effective interaction balance between the human and differing degrees of automation. Many questions were posed but left unanswered. How do you recognize when the boundaries of IT system automation are being challenged, and how do you bring in the right resources to recover?

How do you effectively coordinate resources that are pulled in to deal with anomalies? Is there something we can learn from mission control, which became very good at recognizing anomalies that might occur, and when they did occur, fanned them out to the appropriate areas of expertise for analysis followed by reintegration and planning? How can we tell who is failing – the user or the
automation? Thus, it appears that there is still much to learn in the area of
human factors in automation as applied to the domain of IT systems and service delivery.

Although human factors issues are beginning to be addressed in some professional forums (e.g., International Conference on Autonomic Computing ICAC and USENIX conferences), there is still a need for greater focus on
human factors. Toward that goal, as part of IBM, we’re working with other industry and academic leaders to establish a conference dedicated to this topic, targeted for the late first quarter of 2007.