What If the Doomers Are Wrong? The AI Risk Boards Are Not Modelling
If AI succeeds brilliantly, the greater danger may be who gets the benefits, and few boards are stress-testing that scenario.
Sign up for my Substack daily AI newsletter here.
Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/
I sit closer to the cautious end of the AI debate than the optimistic one. I believe in what AI can do, but I worry about how we get there. So let me test my own position and grant the optimists everything. Steven Boykey Sidley recently imagined a world in which AI cures disease, makes energy cheap and transforms education, then asked who gets access to it. AI can discover a drug, he writes, but it cannot deliver it to a clinic in Soweto. Boards prepare carefully for AI failing. Few have asked what happens if it succeeds for some and not for others.
CONTEXT AND BACKGROUND
The early evidence suggests the benefits are already landing unevenly. Microsoft’s latest diffusion figures show that more than a quarter of the working-age population in the Global North uses AI, against about fifteen per cent in the Global South, and the gap is widening.
The value is concentrated as well. Two economists writing for the Centre for Economic Policy Research estimate that the time AI saves is worth about two point seven trillion dollars a year. High-income countries capture ninety-six per cent of that value, and in sub-Saharan Africa the gains flow almost entirely to higher-paid occupations. The head of the IMF, Kristalina Georgieva, warned recently that AI “largely bypasses most others”, which increases the risk of widening inequality across the globe.
INSIGHT AND ANALYSIS
Success brings three risks that seldom appear in a risk register. The first is ownership. If AI drives a surge in productivity, the returns flow to whoever owns the models, the chips and the data centres. Africa holds less than one per cent of global data-centre capacity, while being home to roughly eighteen per cent of the world’s population.
I have previously written about this in the context of Europe, which wrote the world’s strictest AI rules while hosting only a small share of the compute. My argument was that regulation without compute is not sovereignty. South Africa should invest in the parts of the AI stack it can influence, such as power, trusted cloud and open models it can host itself.
The second risk is time. Sidley points out that early adopters gain an advantage that no later price cut can return. Organisations that learn first build better data, better processes and more capable people, and the gap compounds.
The third is depth. AI can arrive everywhere without changing much. South Africa leads Africa on AI use, yet only sixteen per cent of adults own a computer, and an entry-level laptop costs more than the median monthly wage. A chatbot on a phone is access, not transformation.
IMPLICATIONS
Boards should add a scenario most have never tested: AI works brilliantly for competitors, for some customers and for some staff, but not for everyone. Directors should ask three questions: Which customers and employees are being left behind by our own AI programmes? How dependent are we on foreign compute and models? Are we building an early advantage or deferring it? These are not compliance questions. They go to the organisation’s resilience, its future markets and the strength of its workforce.
For policymakers in one of the world’s most unequal societies, uneven success is not an abstract worry. It is a risk to social stability. Investment in devices, skills, power and local infrastructure is not charity. It is how the benefits reach the last mile.
CLOSING TAKEAWAY
I remain cautious about AI, and I still take its catastrophic risks seriously. Testing my position against the optimists’ best case has clarified something, though. The case for governance does not depend on who wins the argument. If AI fails, we need oversight. If it succeeds, we need it just as much, because benefits do not distribute themselves. Distribution is a decision, not a side effect of the technology. Whether the doomers or the optimists are right, the real test of AI will be who it leaves behind.
Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net
Comments