abstract
-
This dissertation addresses the issue of user adoption of interface agents for electronic mail (email). Interface agents are reactive, continuous, collaborative, and autonomous software entities that act on a user's behalf by communicating directly with a person offering assistance and advice in performing computer-related activities. The study presents and empirically validates a model that describes user adoption behavior, offers insights on important features of this technology from the end-user perspective, reports on critical incidents of agent usage, and offers recommendations for developers and marketers.
As means of investigating this phenomenon, a survey of actual users of an interface agent-based email system was conducted. Emphasis was placed on identifying user needs and key factors that influence their adoption decisions. Data analysis involved quantitative and qualitative techniques (Partial Least Squares, descriptive statistics, classical content analysis). An extended version of the Technology Acceptance Model was introduced and tested, and the user context surrounding email agent adoption was explored. Survey findings suggest that existing management information systems and social sciences theories, models, and methodologies may be fruitfully applied to investigate user adoption of novel interface agent technologies.
By combining and synthesizing results of a deductive and inductive analysis of the survey data, a new, grand model of interface agents adoption and use is suggested that is the central contribution of this research. According to this model, in voluntarily usage conditions, two general types of factors influence user adoption behavior - user perceptions and agent operability. User perceptions are either positive or negative mental reflections of several properties of an agent, such as perceived enjoyment, usefulness, ease of use, intrusiveness, and attractiveness. Agent operability embraces factors pertaining to operational characteristics of an agent, such as compatibility, system interference, reliability, and personalization.
Findings also suggest that to foster the diffusion of highly useful agent systems, developers and marketers need to become aware of the importance of individual user characteristics, enhance their understanding of factors influencing people's adoption decisions, and demonstrate the functionality of interface agents through non-agent technologies.