BEYOND AUTOMATION: GENERATIVE AI'S INFLUENCE ON HUMAN DECISION-MAKING, TRUST, AND ACCOUNTABILITY IN INTELLIGENT CYBERSECURITY SYSTEMS

Abstract

Generative Artificial Intelligence is getting involved in more and more cybersecurity decision-making processes. However, the efficiency is just as important, but this needs to be balanced against issues including dependence, measurement of trust, and responsibility. The study examines how the use of GenAI-supported cybersecurity systems affects 412 cybersecurity professionals, IT security professionals, security operations managers, security managers, and related professionals from their perspectives of human decision-making, calibrated trust, automation bias, and transitioning to an accountable state. A structured questionnaire on five-point Likert scale was used for the study design which was cross sectional quantitative. The analysis of data was done by descriptive statistics, Cronbach's alpha, Pearson correlation and multiple linear regression. There was a strong relationship between GenAI reliance and perceived decision efficiency (r = .52, p-value < .001), and calibrated trust (r = .47, p-value < .001), and also showed a significant association with automation bias (r = .39, p-value < .001). At the multivariable level, reliance on GenAI was the greatest determinant of decision efficiency (β = .41, p < .001), explainability was the greatest among the determinants of calibrated trust (β = .34, p < .001), and human oversight was negatively associated with automation bias (β = −.29, p < .001). The strongest predictor of the diffusion of responsibility was indicated as responsibility ambiguity (p < .001, β = .31). When combined with explainable outputs, active human participation and well-defined accountability, the findings indicate that GenAI can enhance the cybersecurity decision processes. The findings confirm that the role of AI in the workplace must be focused on humans; it should never be a replacement for the essential cybersecurity decision-making.