Social Transparency in Network Monitoring and Security Systems

Abstract

System administrators (sysadmins) are key to keeping computer networks safe. As networks grow in size and complexity, partial workflow automation with the help of AI has been introduced to assist them. However, AI-aided tools often lack transparency, which may lead to the sysadmin's reluctance to use the new software, slower response time in case of a security breach, and increasing errors. Related work suggests that the concept of social transparency (ST), when applied to the IT-security context, enables peer support and could provide the missing knowledge to the user facilitating explainability of the system and improving human-AI trust. In this paper, we investigate the profile of sysadmins and confirm that ST can indeed yield benefits for them but only when coupled with relevant contextual information and only when it adheres to the sysadmins' quality standards. Finally, we contribute design recommendations for incorporating ST into the existing workflows of sysadmins.