Engineers’ Professional Responsibility in Using Machine Learning

As engineers increasingly delegate decisions to machine learning (ML) systems, the profession is starting to face foundational questions about accountability and ethical responsibilities. Given the limited work in this area, this paper examines how ML deployment transforms engineering judgment through philosophical, ethical, and virtue-theoretic lenses. We begin by analyzing ML systems’ distinct opaqueness contrast with traditional computational (e.g., finite element solvers) tools’ characteristic opacity. Building on this analysis, we propose a three-tier safeguard protocol mapped to established duty-of-care thresholds to align with those expected in engineering societies. We then examine how cognitive offloading to ML systems risks eroding core engineering virtues of diligence, prudence, and practical wisdom. In response, we propose interventions and workflow designs that preserve engineers’ deliberative role while leveraging computational augmentation. We also emphasize disclosure norms for ML-assisted design that balance client comprehension with regulatory requirements and argue that transparency obligations should scale with the degree of ML contribution. Finally, we conclude with best practices and specific recommendations for updating professional licensure standards.