The Board's Duty to Chain the AI Hype
NEWP&L MANAGEMENTEXECUTIVE MANAGEMENT
The same AI systems that enable massive productivity gains also enable the large-scale substitution of human labor with autonomous agents. The risk is that our incentive systems may unintentionally accelerate this transition without sufficient guardrails.
If we weight executive compensation too heavily toward cost reduction, margin expansion, or headcount efficiency, we inadvertently incentivize our C-suite to aggressively replace human capability with AI agents. While financially rational in the short term, this can rapidly erode institutional knowledge, weaken organizational resilience, and create immense reputational and regulatory exposure.
We face a structural tension in modern reward design: optimizing for AI-driven efficiency may simultaneously increase long-term fragility. A company can become highly profitable while quietly hollowing out the human systems that sustain adaptability and trust.
For boards, the challenge is no longer just modernizing incentives for AI or applauding survey metrics. It is redefining performance itself—balancing ruthless efficiency with institutional resilience, and ensuring that our drive for AI leverage does not inadvertently produce organizations optimized for speed, but entirely fragile in substance.
About the Author: An Accredited Director with the Singapore Institute of Directors, Board Director, former Big Four Partner, CIO Advisor, and SaaS entrepreneur and investor. He frequently writes and speaks about AI, emphasising the critical need for strict, multidimensional discipline to achieve true AI transformation.
As a board member my time is increasingly spent separating executive enthusiasm from operational reality. Recently, the new "2026 Gartner CEO & Senior Business Executive Survey" crossed my desk, detailing the sentiments of 469 global leaders.
At first glance, the data is impressive. We are allegedly witnessing a wave of "Micro Confidence," with 88% of CEOs planning to increase AI investments and 27% aggressively expecting to run largely autonomous operations by 2028.
But from a governance perspective, this top-down optimism is a dangerous mirage. If we govern our enterprises based solely on this survey's metrics, we are steering the ship straight into a wall. The baseline of this entire conversation is fundamentally flawed because it ignores the messy, bottom-up reality of how AI is actually being used today.
The Shadow AI Blindspot
The Gartner survey captures what leadership thinks is happening, but it completely ignores the elephant in the boardroom: Shadow AI.
In reality, up to 80% of enterprise AI activity—the daily employee prompts, the localized small agents, the unvetted tools—is happening entirely off the books. Consequently, up to 60% of our actual AI costs are buried in fragmented, unnoticed SaaS subscriptions or hidden in massive token consumption.
Because we are blind to this shadow activity, we are suffering from three massive institutional blindspots:
Unknown Value: We have zero visibility into the actual ROI embedded in the current AI usage.
Entrenched Compliance Risks: Corporate data is already flowing freely into third-party, uncontrolled models.
Invisible Talent: We have absolutely no clue who our most talented and knowledgeable employees are. Our staff is aggressively upskilling themselves in the dark, and because we refuse to acknowledge Shadow AI, we cannot formalize their grassroots maturity into a cohesive transformation strategy, although AI savvy talent will be key.
The Executive Illusion of Fast Progress
When we view the Gartner findings through the lens of this flawed baseline, several massive logical gaps emerge in the current C-suite narrative. There is a growing, dangerous tendency for executives to push for top-down "autonomous operations" simply to skirt the incredibly messy, difficult work of bottom-up change management. They are prioritizing the illusion of fast progress.
Consider these three glaring contradictions:
The Missing LLM ROI: The survey champions the fact that 88% of CEOs will increase their AI spending. Yet, why is there absolutely no boardroom discussion about the millions already being sunk into making enterprise LLMs available to all employees today? We are projecting future spending while seemingly ignoring the utilization and value generation of the tools currently in our employees' hands.
The 19.9% "New Revenue" Fallacy: CEOs boldly estimate that by 2030, 19.9% of total revenue will come directly from AI agents or machine customers. Let’s be pragmatic. AI does not automatically sell more of our existing products, nor does it magically create net-new physical goods and build out their required supply chains overnight. It is far more likely that this figure merely represents the automation of Order-to-Cash (O2C) and treasury operations—which 25% of CEOs explicitly expect AI to handle. We must stop confusing operational efficiency with net-new top-line revenue.
The Accountability Void: An astonishing 39% of CEOs already view AI agents as employees, a fundamental change in how they define the workforce. Coupled with the desire to design humans "out of the loop" for core transactions, a glaring question arises: If AI agents operate independently, who exactly is responsible when things go sideways? We cannot automate away fiduciary duty. If a machine makes a catastrophic trading or payment error using programmable money, who is legally and financially accountable?
The Governance Challenge: Redefining Incentives and Guardrails
The Gartner report correctly notes a critical implication for the C-Suite: enterprises cannot simply cut their way to an autonomous business future. The transition requires a new and specific talent pattern, protecting high-value human roles and automating transactional ones.
For the board, this means AI is forcing us to rethink a core assumption behind our leadership reward models: the idea that value creation is primarily driven by humans managing other humans. That assumption is breaking down. Output is increasingly decoupled from headcount. Small teams amplified by AI can outperform large traditional structures, and execution itself is shifting from people to autonomous agents.
This creates immediate pressure on our compensation frameworks. If leadership impact is now less about managing scale and more about orchestrating systems of human and machine intelligence, we must move toward contribution-based reward models. Instead of paying executives for organizational size or budget ownership, compensation must be tied to measurable value creation, system leverage, and transformation outcomes.
However, this shift introduces a deep structural risk that is only just surfacing in governance discussions.
Join thousands building the future of crypto intelligence
Start using AI-powered tools that elevate your trading and decision-making.
