AI is transforming work. The questions boards are not asking

In the last piece, we made a clear case: replacing people with AI is accelerating, but governance is not keeping pace with what is being changed and what is being lost. The more consequential question is whether organisations are reshaping work in ways that remain governable, explainable, and resilient under real-world complexity. Productivity gains are tangible, and some firms report meaningful cost savings. Yet displacement remains uneven, benefits often depend on the successful redeployment of human capacity, and real-world failures show that poorly governed AI can scale error, bias, and liability faster than it scales value.

Boards are asking sensible first-order questions. What is the strategy? What is the return on investment? Who owns the program? What are the legal, cyber, data, and vendor risks? Do we have the right talent? These reflect current governance norms and are necessary.

They are also insufficient.

The critical blind spot is not in the answers boards are getting. It is in the questions they are not asking. As AI systems take on a greater share of work – making decisions, shaping outcomes, and redefining roles – the risk is not only the loss of control. It is that boards are not interrogating the underlying redesign of the enterprise. And in governance, what is not asked is what goes unchallenged and, eventually, what goes wrong.

Boards are not asking what work must remain human and why.
The dominant lens is substitution: what can be automated. The more consequential question is preservation: what must not be? In areas involving ambiguity, ethical judgement, and reputational consequence, human intervention is not optional; it is a control. Without defining where human authority must remain, organisations risk building systems that perform efficiently in routine conditions but fail under pressure.

Boards are not asking how capability is being redistributed or eroded.
Efficiency gains often come from removing layers of work that historically developed judgement and expertise. What appears as productivity in the short term can translate into capability gaps over time. This is not workforce optimisation; it is a continuity risk. If expertise is compressed faster than it is rebuilt, dependence on opaque systems increases.

Boards are not asking how decision accountability is being restructured.
AI redistributes decisions across systems, making accountability harder to trace and defend. Traditional governance assumes clear ownership; AI disrupts that model. Boards must require not just named accountability but accountability that is demonstrable – traceable, documented, and defensible across the full decision chain.

Boards are not asking what happens at the edges – where systems are most vulnerable.
AI performs well within defined boundaries but degrades in novel conditions. When systems encounter edge cases with no historical precedent, they apply past logic to new realities. Without escalation paths, fallback mechanisms, and rollback triggers, errors do not stay contained; they scale. This is where governance is tested.

Boards are not asking how value is created and who captures it.
AI can improve output while reducing labour, but it also redistributes value across vendors, platforms, and data ecosystems. At the same time, removing human roles without reinvestment weakens internal capacity to create differentiated value. Focusing only on cost savings risks short-term gains at the expense of long-term advantage.

These are not abstract concerns; they are structural gaps in how AI-driven work is being governed, with direct implications for resilience, accountability, and long-term value.

The shift required is one of perspective. AI-driven workforce transformation must be governed as a capital allocation and control decision at the highest level of the enterprise, not delegated as a technology rollout or treated as a cost-efficiency programme. Boards must move beyond endorsing adoption to asserting control over how work is redesigned, how decisions are made, and how accountability is sustained.

This requires discipline in questioning and clarity in expectation. In an AI-driven enterprise, control is not established by implementation. It is established by the ability to explain, challenge, and intervene—consistently and with evidence.

Boards that do not recalibrate now will oversee organisations that optimise for efficiency while eroding the conditions that sustain performance: capability, accountability, and resilience.

This is a governance obligation. Control must be demonstrable. Accountability must be explicit. And every material AI-driven decision must be defensible by design, not after the fact.

Amaka Ibeji, Founder of DPO Africa Network, is a Boardroom Qualified Technology Expert and Digital Trust Visionary. She advises boards, regulators, and organizations on privacy, AI governance, and data trust, while coaching and fostering leadership across industries. Connect: LinkedIn amakai | [email protected]