Data quality stands as the silent executioner of enterprise AI initiatives, responsible for an estimated 60-73% of AI project failures. This article presents a comprehensive economic framework for understanding, measuring, and mitigating the costs of substandard data in AI systems. Drawing on fourteen years of enterprise software development and seven years of AI research, I examine the hidden cost multipliers that transform minor data quality issues into multi-million dollar failures.
Oleh Ivchenko (Thu,) studied this question.