A new survey from Dataiku, “Global AI Confessions Report: Data Leaders Edition,” reveals that 95% of senior data executives admit they cannot fully trace AI decision-making, raising concerns about governance, explainability, and trust in AI deployments.
The Harris Poll study surveyed 812 data leaders across the U.S., U.K., France, Germany, UAE, Japan, Singapore, and South Korea, finding that 52% have delayed or blocked AI agent deployments due to explainability concerns, while only 19% always require AI agents to “show their work” before approval. The survey also highlighted that 80% consider an accurate but unexplainable AI decision riskier than a wrong but explainable one.
CIOs and CDOs face high stakes: 46% are most likely to be credited for AI gains, yet 56% are most likely to be blamed for business losses caused by AI failures. Around 60% fear their jobs are at risk if AI does not deliver measurable results within two years.
Data leaders also reported that AI is already causing challenges in practice, with 59% stating AI hallucinations or inaccuracies have caused business issues in the past year, and 82% believing AI can outperform their boss in analysis, though 74% would revert to human oversight if AI errors exceed 6%. Furthermore, 89% said there is at least one business function they would never delegate to AI.
The survey underscores a disconnect between CEOs and data leaders. Only 39% of data leaders say their C-suite fully understands AI, while 68% believe executives overestimate AI accuracy, and 73% say executives underestimate the difficulty of achieving AI reliability before production.
Florian Douetteau, Co-founder and CEO of Dataiku, said, “An alarming revelation of the report is that enterprises worldwide are betting on AI they don’t fully trust. The good news is that most failed AI initiatives suffer from common blockers that can be overcome with more explainability, traceability, and governance.”
The report highlights a growing need for stronger AI governance frameworks to ensure reliability, transparency, and accountability in enterprise AI deployments.
