Cognitive Debiasing in Engineering Project Management: A Neuro-Symbolic XAI Framework for Mitigating Strategic Misjudgment

Large-scale engineering projects are chronically plagued by costly overruns, a multi-trillion-dollar problem often rooted in persistent human cognitive biases like the Planning Fallacy and Optimism Bias. While Artificial Intelligence (AI) has been increasingly applied to project management, current tools often operate as "black boxes"and fail to address the psychological drivers of poor strategic decisions, limiting trust and adoption. This paper introduces a novel Neuro-Symbolic Explainable AI (XAI) framework designed to act as a "cognitive mirror"for project managers, augmenting their judgment in real-time. The system integrates a Natural Language Processing (NLP) model to detect linguistic markers of bias in project documents, a causal Bayesian network to simulate the downstream impact of these biases on project risks, and an XAI module that generates actionable counterfactual feedback. The framework's efficacy was evaluated through a controlled, human-in-the-loop experiment. The results demonstrate that participants using the tool produced significantly more realistic costs and schedule forecasts, indicating a measurable reduction in biased decision-making. Our findings provide strong evidence for a new class of human-centered AI systems that augment, rather than replace, expert judgment, contributing to more resilient, transparent, and ethically sound project management practices.