Riding the Paradox: How AI Consultants Manage the Tradeoff Between Explainability and Performance

This study examines how AI consultants manage the often-cited tradeoff between explainability and predictive performance in machine learning models. While simpler models are easier to explain, they frequently underperform compared to complex, opaque ones. Through interviews with experts at a global IT consulting firm working across diverse AI projects, we explore how they evaluate explainability, perceive the explainability–performance paradox, and address the challenges of using black-box models. The findings uncover previously undocumented practices, including situations where explainability can be counterproductive. These insights offer fresh contributions to academic discourse and provide actionable implications for AI practitioners navigating this critical tradeoff.