Managerial expectations of generative AI adoption vs lived experiences of knowledge Workers

Abstract

Generative artificial intelligence tools such as ChatGPT, Microsoft Copilot and Claude have been adopted rapidly in knowledge-work settings, where they are often presented as instruments for productivity, speed and cost efficiency. This thesis examines the gap between these managerial expectations and the lived experiences reported by IT related knowledge workers. The research problem is that organizations are adopting generative AI tools faster than they are developing evidence-based practices for managing their effects on work, skills, autonomy and professional identity. The study uses a multivocal literature review design that combines formal academic and practitioner literature, 150 manually screened Reddit posts from professional communities, and five semi structured interviews with IT workers. The material was analyzed thematically and compared across the three evidence sources. The analysis identifies five recurring themes: deskilling and skill erosion, workload transformation, surveillance and control, identity threat, and management disconnect. The findings show that generative AI does not simply reduce work; it often shifts work toward validation, review and correction. Workers also report dependency on AI tools, uncertainty about monitoring, and a perceived gap between leadership narratives and day-to-day practice. Management disconnect was the most frequent theme in the Reddit data and was also strongly reflected in interviews. The thesis contributes a role sensitive framework for understanding generative AI adoption in IT work and argues that successful adoption requires training, transparent guidelines and realistic managerial expectations rather than a narrow focus on productivity metrics.