Who Is Actually Running Your Organisation?
- H.E. Prof. Ambassador Tal Edgars, PhD
- 2 days ago
- 5 min read
The Most Important Question Boards Are Refusing to Answer
Let me ask you a question that most boards are too uncomfortable to answer honestly. When your organisation made its last major strategic decision, a capital allocation, a restructuring, a market entry, a credit approval, a procurement award, who actually made it?
Was it your CEO? Your board? Your executive committee? Or was it an algorithm?
If you hesitated before answering, you are already living inside the most consequential organisational crisis of our time.
THE ILLUSION OF HUMAN AUTHORITY
Across the world, in boardrooms that project confidence and institutions that project authority, a quiet and largely unacknowledged transfer of decision-making power is underway.
Artificial intelligence systems built by a handful of technology companies headquartered predominantly in the United States and China are now embedded in the credit models of our banks, the procurement systems of our governments, the performance management frameworks of our corporations, the diagnostic tools of our hospitals, and the surveillance infrastructure of our cities.
The humans nominally in charge are, in many cases, ratifying decisions they did not originate, do not fully understand, and cannot meaningfully interrogate.
We have created a sophisticated theatre of human authority. The curtain is beginning to fall.
THE LIABILITY VACUUM THAT IS DESTROYING GOVERNANCE
Every governance framework in existence, the King IV Report, the OECD Principles of Corporate Governance, the Companies Act, the Public Finance Management Act, was built on a foundational assumption: that human beings make decisions, and human beings bear accountability for them.
That assumption is now structurally broken.
When an AI-driven credit model denies ten thousand small business owners access to finance, disproportionately black, disproportionately female, disproportionately rural, who is accountable? The bank's CEO?
The board's risk committee? The technology vendor in San Francisco who sold the model? The data scientists who trained it on historical data already saturated with systemic bias?
The honest answer, in most jurisdictions today, is: nobody.
This is not a theoretical concern. It is the lived reality of millions of people across Africa and the developing world who are being governed, assessed, financed, and excluded by systems they have no visibility into, no recourse against, and no democratic mandate over.
Boards that are not wrestling with this question are not governing. They are presiding.
THE AFRICAN DIMENSION THAT THE GLOBAL CONVERSATION IS IGNORING
The Western discourse on AI governance is, at its core, a conversation among the powerful about how to manage the power they already hold. It is a debate conducted by those who built the systems, own the infrastructure, and control the data.
Africa is largely absent from that conversation. Not because we have nothing to say, but because we have not yet fully recognised that we are already the subjects of it.
Consider the architecture of decision-making on our own continent.
Development Finance Institutions disbursing billions across African economies are increasingly using AI-assisted risk models and impact assessment frameworks developed in Washington, London, and Brussels.
National treasuries are procuring enterprise resource planning systems built on algorithmic logic they did not design.
State-owned enterprises are adopting performance management platforms whose underlying models reflect assumptions about productivity, value, and human behaviour rooted in entirely different cultural, economic, and institutional contexts.
We are, in effect, allowing our most consequential institutional decisions to be shaped by intelligence that does not know us, was not built for us, and is not accountable to us.
This is not innovation. In its current form, it is a new architecture of dependency. It demands the same rigour of scrutiny that we applied to every other system of external control that has sought to govern African institutions from a distance.
THE BOARD COMPETENCE CRISIS
I have sat in enough boardrooms to know that the majority of board members, accomplished, experienced, and well-intentioned as they are, do not possess the technical literacy to govern AI systems. This is not an indictment. It is a structural reality that our governance frameworks have not yet confronted honestly.
How does a board exercise meaningful oversight over an algorithm it cannot read, a model it cannot interrogate, and a dataset it has never seen?
How does a risk committee assess the ethical dimensions of a system whose decision logic is, by design, opaque?
How does an audit committee provide assurance over processes that unfold at a speed and scale that no human auditor can match?
The answer, in most organisations today, is that they do not. They rely on management assurances, vendor certifications, and the comfortable fiction that because a system was procured through due process, it is therefore legitimate.
Due process and legitimacy are not the same thing. They have never been.
WHAT GENUINE GOVERNANCE REQUIRES
I am not arguing against artificial intelligence. I am arguing for accountability, the same accountability we demand of every other force that shapes organisational outcomes.
Genuine governance of AI in organisations requires five non-negotiable commitments.
First, algorithmic transparency. Organisations must be able to explain, in plain language, how consequential AI systems make decisions. Black box models that affect human lives and institutional outcomes are incompatible with legitimate governance.
Second, bias auditing. Every AI system deployed in a consequential domain must be subjected to regular, independent auditing for discriminatory outcomes across race, gender, geography, and socioeconomic status. In the African context, this is not optional. It is a constitutional imperative.
Third, human override authority. No AI system should have the final word on a decision that materially affects human dignity, access to resources, or institutional direction. Human override must be not merely possible but structurally mandated.
Fourth, data sovereignty. African organisations must assert ownership and control over the data that feeds the systems governing their institutions. Data exported to foreign platforms, processed under foreign jurisdictions, and used to train models that are then sold back to us represents a sovereignty deficit we can no longer afford to ignore.
Fifth, board-level AI literacy. Governance bodies must invest urgently in the capability to understand, interrogate, and provide genuine oversight of the AI systems operating within their organisations. This is not optional enrichment. It is a fiduciary duty.
THE QUESTION THAT WILL DEFINE YOUR LEGACY
History will not remember whether your organisation adopted AI early or late.
It will remember whether you governed it wisely or recklessly.
Whether you used it to expand human capability or to abdicate human responsibility.
Whether you deployed it in ways that advanced equity or entrenched exclusion.
The leaders who will be remembered well are those who, in this moment of profound technological disruption, had the courage to ask the uncomfortable question and the discipline to demand an honest answer.
So let me ask it again.
Who is actually running your organisation?
If you do not know with certainty, you have your most important governance task ahead of you.
"Africa has a unique chance to shape the AI revolution... Africa is no longer just adopting technology; it is setting the rules. If Africa supplies the minerals, why shouldn't Africa also build the value chains? Why not assemble GPU systems on the continent? ... Africa cannot afford to watch the intelligence age unfold from the sidelines.
Comments