In one of the regional banks I worked with, the CEO prided himself on having “a robust strategic plan.” Thick document. Perfect charts. The kind you present once a year and lock away. Then, one quarter, an AI-driven risk model flagged early signs of loan defaults that contradicted management’s optimism. The board ignored it: “Our experience says otherwise.” Six months later, losses doubled. The irony? The algorithm had seen what human intuition refused to admit.
That story is repeating itself across industries. From manufacturing to insurance, leaders are waking up to a new, uncomfortable reality: artificial intelligence is not a productivity tool; it is a strategic actor. Yet boards are still governing as if it were a line item in IT, not a co-pilot in decision-making.
The modern boardroom still loves the ritual of quarterly reviews and human deliberation. But what happens when the inputs are curated, filtered, or even generated by AI? When the narrative in the slide deck has already been optimized to confirm your biases? Many boards unknowingly validate the outcomes of algorithms without questioning their logic.
My take? “Human oversight” has become the most overrated governance safeguard in the AI era. Oversight without comprehension is theatre. AI literacy is now as fundamental to governance as financial literacy was in the 1990s. Yet only a fraction of board members can interrogate a data model, question algorithmic drift, or interpret predictive confidence intervals.
If you cannot understand how an algorithm arrives at a decision, you are not governing it; you are endorsing it.
Strategy is no longer a plan; it is a learning loop
Traditional strategy assumes a relatively stable environment: analyse, decide, execute, monitor. But AI doesn’t operate linearly. It observes, predicts, learns, and adapts continuously. The organization that learns faster now wins, not necessarily the one that plans better.
That is why the most resilient companies are abandoning static plans for dynamic learning systems. A leading manufacturer I advised stopped issuing five-year strategies. Instead, it built a rolling AI-driven “strategy lab” where market data, customer feedback, and operational metrics constantly feed new scenarios. The board’s role shifted from approving strategy to approving how the company learns.
It’s not a comforting shift. It demands humility. When the data disagrees with experience, who do you trust: the model or the manager? The winning companies are learning to test both.
The governance gap
AI has outpaced governance frameworks. Most boards have risk committees, audit committees, and ESG oversight. Few have algorithmic oversight. Even fewer have AI ethics policies that go beyond public relations.
Consider a telecom company that used AI to optimize workforce performance. The algorithm, trained on historic data, began rewarding aggressive behaviour and penalizing collaboration. Within a year, team trust collapsed. On paper, productivity rose. In reality, culture eroded.
That is the new governance challenge: AI amplifies what already exists, good or bad. It reflects the culture it finds. Therefore, if your culture prioritizes control over curiosity, the AI will learn to manipulate metrics rather than solve problems.
- Boards must start asking a different set of questions:
- What is the organization’s AI philosophy: efficiency, ethics, or exploration?
- Who owns the consequences of an algorithmic decision gone wrong?
- How do we audit a decision no human fully understands?
Without answers, you are running an unregulated brain inside your enterprise.
The false comfort of explainability
“Explainable AI” is the new buzzword in board packs. But explainability is not understanding; it is simplification. The board may see a neat graph or a probability score, yet never grasp the uncertainty baked into it.
True accountability demands traceability: the ability to follow a decision from input data to outcome. That requires an AI audit trail, just like a financial one. When AI models influence pricing, hiring, or risk, boards must demand that every automated decision be retrievable, reviewable, and reversible.
This is where most organizations falter. They invest millions in AI but almost nothing in AI governance infrastructure. The CIO focuses on implementation; the board assumes compliance will follow. But governance must lead design, not clean up its mess.
There’s a deeper threat no one wants to discuss: AI is quietly redistributing power inside corporations. Middle managers, once translators between data and decision, are being disintermediated. Algorithms now brief executives directly. Strategy offices are becoming data science hubs. Boards, meanwhile, still think they are steering the ship, but the navigation system is running on code they cannot read.
The organizations that will thrive are those that redefine governance as sensemaking, not supervision. They will invest in AI translators; leaders who can bridge technical depth with strategic judgment. They’ll treat AI not as a subordinate tool but as a system that requires moral and operational boundaries.
Mr Strategy’s recommended playbook for boards
- Redesign the board agenda. Every agenda should include “algorithmic impact review.” How is AI influencing capital allocation, talent, and ethics?
- Appoint a board AI adviser. Just as boards once added financial experts, they now need digital anthropologists who can interpret how algorithms reshape human behavior.
- Demand explainability by design. Before approving any major AI deployment, require an assurance statement detailing data provenance, model governance, and ethical safeguards.
- Integrate culture into AI oversight. Audit whether AI applications align with organizational values, not just performance metrics.
- Train directors in data literacy. You cannot oversee what you do not understand. A one-day “AI 101” course will not cut it.
Executives love to say, “AI will not replace leaders; leaders who use AI will replace those who don’t.” True, but incomplete. The real shift is that leaders who understand how AI learns will outlast those who only use it. AI is not neutral. It rewards velocity and punishes complacency.
The same algorithm that forecasts demand can reinforce inequity. The same chatbot that delights customers can destroy privacy. Boards must evolve from approving strategy to curating moral boundaries. That is the new frontier of corporate governance.
In a decade, regulators will not ask whether you used AI. They will ask how you governed it. And that is why my Algorithmic Accountability Framework
Clarity: Define what decisions AI can and cannot make.
- Control: Establish traceability; every AI decision must leave a governance footprint.
- Culture: Monitor how AI affects fairness, trust, and human dignity.
- Competence: Build AI literacy across the board.
- Courage: Stop hiding behind consultants. Make AI governance a standing item, not a project.
In the end, the companies that win will not be those that deploy the most AI, but those that retain moral and strategic clarity amid automation. Because the real risk is not that machines will start thinking like humans. It is that humans will stop thinking at all.
