Intelligent Automation in Collaborative Systems
Intelligent automation has been a source of research and debate within the design community for several decades. When adding intelligent automation to single-user systems, two critical issues must be addressed. First, sufficient knowledge must be acquired about the user and her context to make high-level inferences at runtime. Second, the automation must be useful and delivered in a manner that does not impair the user's domain activity. These issues are equally relevant for collaborative systems. However, collaborative systems offer a potential solution to these problems by virtue of their privileged position as mediating artifacts within a collaborative process. Because coordination information must be exchanged through the system, there is an opportunity for the system to gain insights into user activities and context. Because mediating artifacts add structure to the information that passes through them to improve coordination, this information is made more accessible to standard AI algorithms. Thus, within a design solution for coordination problems in groupware, a solution to some of the issues with intelligent automation can also be found. Empirical evidence from a testbed domain is presented that validates this approach, along with a discussion of how the approach can be generalized to other collaborative systems.