SAS and IDC report: 97% of users override AI recommendations

A new study by SAS and IDC, titled ‘Data and AI Impact Report: The New Economics of Trust’, reveals that 97.2% of users override or correct artificial intelligence recommendations. This behavior is primarily driven by a lack of explainability, as professionals struggle to understand or verify how AI systems reach specific conclusions.

The report highlights a significant economic divide: organizations that implement trustworthy AI practices—focusing on data quality, governance, and auditability—are 15 times more likely to report a high return on investment (ROI) compared to those that do not. These leading organizations achieve at least double the ROI of their peers.

As AI systems evolve toward ‘agentic AI’ with higher levels of autonomy, the study suggests that establishing transparency and trust is essential to prevent manual reviews from negating the efficiency gains intended by AI implementation.