Research Insight · From Section 3 of the SGaaS White Paper
The AI Governance Gap
Two-thirds of boards cannot evaluate what management tells them about AI. The cognitive defect Herbert Simon identified in 1947 — satisficing under bounded rationality — is operating, measurably, in the risk category that the WEF identifies as showing the sharpest increase in perceived severity between the short and long term.
0%
of boards report they have “limited to no knowledge or experience” with AI — the empirical signature of bounded rationality operating in the fastest-rising risk category.
695 board members & C-suite executives
56 countries
Jan–Feb 2025 survey window
Source: Deloitte Global Boardroom Program (2025)
Dial 1 of 3 · Knowledge
Two-thirds of boards cannot evaluate the technology
66% of board members and C-suite executives report that their boards have “limited to no knowledge or experience” with AI. Without that knowledge, evaluating management’s AI strategy is not possible — only accepting it is.
Why this is the satisficing pattern
Bounded-rationality theory predicts that decision-makers who lack the cognitive resources to evaluate a risk domain will engage with it episodically rather than continuously, and will default to accepting management’s framing rather than challenging it. The 66% figure is that prediction, confirmed in the domain where governance stakes are rising fastest.
0%
of WEF-surveyed risk experts expect a calm global outlook
“The risk category rising fastest is the one boards are least equipped to discuss.”
WEF Global Risks Report 2026 · 1,300+ risk experts surveyed across 100+ countries. Neither the two- nor the ten-year horizon is expected calm. The risk environment that boards must govern through is one the practitioner consensus considers anything but benign.
What the gap means
Three findings, one structural defect. Engagement frequency is determined not by risk severity but by board cognitive capacity.
01
Boards cannot challenge what they cannot evaluate
The 66% knowledge gap is the operational ceiling on challenge. When evaluating management’s AI claims requires expertise the board lacks, the only available board behaviour is acceptance.
02
Engagement frequency follows capacity, not severity
The 31% who don’t discuss AI at all are not making a risk-prioritisation decision. They are demonstrating that cognitive cost exceeds the perceived cost of inattention.
03
17% is the floor, not the standard
A continuous-risk landscape requires continuous board attention. Only 17% of boards currently operate at that cadence. The 83% gap is the structural defect — not an effort problem.
04
Adversarial governance closes the gap structurally
A retained principal whose mandate is to challenge AI strategy carries the domain expertise the board lacks and operates on the cadence the risk landscape requires — without asking the board itself to become technologists.
The Structural Defect
Two-thirds of boards cannot evaluate what management tells them about AI. One-third do not discuss it. The satisficing pattern identified by Simon in 1947 is operating, measurably, in the risk category that the WEF identifies as showing the sharpest increase in perceived severity between the short and long term.
From Strategic Governance as a Service, Section 3 — Marentis Labs, 2026.
From the SGaaS White Paper
Read the full research
The interactive above is one of seven drawn from the 90-page Strategic Governance as a Service research paper. Download the full paper or schedule a confidential conversation.