IBM Study: The AI Workflow Gap Is Not Only Skills Training—It Is Who Owns Judgment

Has AI made organizations more productive? Start by checking who owns judgment and accountability

On September 21, 2026, the IBM Institute for Business Value published a study conducted with Oxford Economics. It combines two surveys: 1,500 CHROs or equivalent executives responsible for workforce strategy, and 8,800 full-time employees across 28 countries. IBM reports that 60% of employees worry AI is eroding their skills. It also reports that 71% of CHROs consider supervising, validating, and overriding AI outputs the workforce’s most essential skill, while only 29% of employees rank judgment as important.

These are survey findings reported by IBM. They are not WayToClawEarn’s own efficiency test, and they do not by themselves prove that AI caused skills erosion or will produce productivity gains in every organization.

What did the study actually find?

• A judgment-versus-execution gap: CHROs place more weight on supervising, validating, and overriding AI outputs than employees place on judgment.
• Incomplete organizational design: IBM says 46% of organizations do not involve the CHRO when AI strategy is defined, and only 28% of CHROs report a joint HR–IT roadmap with a shared operating cadence.
• Unclear accountability: 43% of employees say they are blamed when something goes wrong with AI, while 36% of CHROs say unclear accountability is complicating deployment.
• More invisible work: 80% of CHROs believe AI creates invisible work such as validating recommendations, fixing errors, adding context, and managing exceptions. Forty-two percent of employees say AI increases their workload or that their extra work goes unrecognized.
• Treat outcome figures carefully: IBM says organizations that clearly define human-led, AI-assisted, and AI-executed workflows report 18% lower risk and 20% better quality. Those are reported organizational outcomes and associations from the study, not a randomized controlled trial or a guarantee for every company.

What does this mean for people building AI products and services?

The most useful signal is not whether AI will replace a particular job. It is what enterprises may pay for as a deliverable:

1. Make decision ownership a product field. For each AI output, record who approves it, who reviews it, who can override it, and when human escalation is mandatory.
2. Make invisible work visible. Estimate not only generation time but also validation, rework, exception handling, and training costs.
3. Sell a workflow, not just a chat window. Define the boundary between human-led, AI-assisted, and AI-executed work before choosing models, tool permissions, and audit logs.
4. Validate value on real tasks. Track model versions, inputs, human edits, failures, handling time, and final quality before claiming that a workflow improved.

This is a productization recommendation, not an IBM-proven revenue case. The study does not provide a guaranteed income number for independent builders and does not prove that any particular AI service will be profitable.

Conclusion and next step

IBM’s data supports a limited but actionable conclusion: enterprise AI adoption may be constrained not only by model capability, but also by decision ownership, accountability, and the new work created after automation. It does not support turning survey associations into universal causal claims.

If you deliver AI automation, create a workflow accountability map for your next project. For every step, state who decides, what AI may do, what triggers human takeover, and how failure and rework are logged. After recording real tasks for a defined period, decide whether to expand automation or make an efficiency claim.

Sources

• IBM Newsroom official announcement (September 21, 2026): https://newsroom.ibm.com/2026-09-21-new-ibm-chro-study-ai-puts-critical-thinking-at-the-center-of-workforce-priorities
• IBM Institute for Business Value study page: https://www.ibm.com/thought-leadership/institute-business-value/en-us/c-suite-study/chro
• Methodology: IBM and Oxford Economics conducted two complementary surveys from April to June 2026 across 21 geographies, 23 industries, and 28 countries. This article does not present the survey findings as our own test.