Support or Surveillance? A Mixed Methods Study of AI Implementation Framing, Employee Trust, and Adoption Intention

This thesis examines whether and how managerial framing of workplace artificial intelligence is associated with employees’ perceived surveillance, trust in AI systems, and adoption intention. The study compares supportive AI implementation framing with surveillance-oriented framing using a mixed methods design. First, five semi-structured interviews with professionals from different organisational backgrounds explored perceptions of AI implementation, transparency, monitoring, trust, wellbeing, and adoption. Second, an online between subjects survey experiment with 60 participants compared responses to supportive and surveillance-oriented AI implementation scenarios. The qualitative findings suggest that employees respond more positively to AI systems when they are perceived as supportive, transparent, practically useful, and human centred. Monitoring, behavioural tracking, and organisational control concerns were identified as barriers to trust and adoption. The quantitative findings showed descriptively higher trust in AI and adoption intention under supportive framing. Therefore, H2 and H3 were directionally consistent with expectations, but these differences were not statistically confirmed. H4 was supported as a correlational association, as perceived surveillance was negatively associated with adoption intention. H5 was strongly supported as a correlational association, as trust in AI was strongly positively associated with adoption intention. However, H1 was not supported, as perceived surveillance remained highly similar across both framing conditions. The study contributes to organisational AI adoption research by showing that supportive framing may be associated with more favourable trust and adoption perceptions, but may not be sufficient to reduce surveillance related concerns. The findings highlight the importance of transparent communication, human oversight, trust building, and responsible data governance in workplace AI implementation. The findings should be interpreted cautiously due to the exploratory design, limited sample size, self-reported data, and scenario-based experiment.