Only 35% of Executives Trust AI to Deliver Reliable Outcomes

The reliability gap

As artificial intelligence moves from pilot programs to production environments, a significant trust deficit has emerged among business leaders. Only 35% of executives say AI consistently delivers intended business outcomes while remaining under control and earning confidence from regulators, customers, and leadership, according to research from HFS Research and TCS published Monday.

The survey of more than 100 C-suite and technology executives across the United States and Canada reveals a stark reality: enterprises are struggling to prove AI can reliably perform at scale. Just one in four executives reported having the governance and controls necessary to support enterprise-wide AI deployment, while only one in six trust autonomous AI systems for critical business functions.

"We've spent the last few years asking whether we can begin to deploy these tools — the harder question we have is whether we can begin to depend on them," Dana Daher, executive research leader at HFS Research, told CIO Dive.

Why it matters

This reliability crisis arrives as companies face mounting pressure to justify AI investments amid rising costs, sluggish workforce adoption, and evolving regulatory requirements. The gap between deployment and dependability threatens to stall AI transformation initiatives just as enterprises move beyond experimentation. Organizations that fail to address trust and governance issues risk both wasted capital and operational disruptions from AI systems that executives cannot confidently scale.

Where trust breaks down

The research identifies several critical failure points. Companies are tracking technical metrics like uptime and accuracy more frequently than actual business outcomes, creating a measurement mismatch that obscures whether AI delivers value. Vendor fragmentation and inconsistent pricing models compound cost challenges, while low employee understanding of AI tools—just 16% of information workers highly understand AI according to separate Forrester research—limits effective adoption.

As AI systems grow more complex, explainability becomes increasingly difficult. Executives struggle to articulate how AI reaches its decisions, creating what Daher describes as "a really big tension around trust" as organizations deploy AI across operations without understanding how it generates responses.

The human-in-the-loop imperative

Confidence in AI remains highest when humans maintain oversight, the report found. Reliable AI doesn't mean eliminating every failure, researchers noted, but rather building systems that can quickly detect, explain, and control problems when they occur.

Daher emphasized that organizations need clearly defined human-in-the-loop strategies that outline accountability and trust levels. "You need to train the person, figure out what they're accountable for," she said. This approach requires evaluating AI based on business outcomes delivered rather than purely technical performance metrics.

Governance frameworks will prove crucial as AI systems scale. However, most U.S. companies currently lack mature AI governance structures, creating vulnerability as regulatory scrutiny intensifies and stakeholder expectations for responsible AI deployment grow.

The findings were first reported by ESG Dive and reflect a broader industry shift toward more pragmatic AI approaches as the initial deployment wave gives way to harder questions about dependability, control, and sustainable value creation.