KPMG’s latest Global AI Pulse shows that the share of organizations “driving adoption” jumped from 13% to 22% in one quarter. Yet only 7% describe themselves as having established ROI.
Yesterday, I argued that a company is already an algorithm.
The KPMG report provides some interesting evidence for why that matters as AI moves deeper into organizations.
AI adoption is accelerating rapidly. Value realization is not keeping pace.
And one of the strongest signals is organizational clarity.
Organizations with clearly defined executive accountability report established ROI at more than 3x the rate of those without it.
Yet only around a third report very clear arrangements for questions such as:
- Who owns AI data quality?
- When should humans override AI?
- Who can intervene or stop an AI-driven decision?
- Who understands and owns the economics of AI once it is embedded in workflows?
These look like AI governance questions. I think they point to something deeper.
They are questions about how the organization reasons:
- Who decides?
- On what evidence?
- Using which criteria?
- Within which constraints?
- Who can override?
- What gets escalated?
- How do outcomes change future decisions?
For decades, organizations could leave much of this implicit.
People learned the real operating system through experience, relationships, incentives and observation.
AI changes that.
You cannot reliably insert agents into an organization whose decision logic exists primarily in people’s heads.
As model capability improves, the bottleneck increasingly moves outside the model: from AI capability to organizational reasoning capability.
The companies extracting disproportionate value from AI may not be those deploying the most AI.
They may be those that make their decision logic explicit enough for humans and machines to reason within it together.
That is a much bigger transformation than implementing AI.
It is making the organization itself legible.