The future of AI could be one of five scenarios.

I have spent nearly thirty years in and around business and IT transformation. Long enough to have watched several technology waves reshape industries, and long enough to recognise a repeating pattern.

I am nearing the end of my AI Executive Program at Oxford uni, this week reviewing the UK Government's AI 2030 Scenarios report. It dawned on me that we are in the early stages of a cycle familiar to anyone who lived through the large enterprise platform era (think ERP or SharePoint). Organisations moving fast, embedding technology deeply, with governance, architecture and long-term thinking trailing behind. We know how that story played out. Customised implementations locked into brittle architectures, change management an afterthought and years unpicking decisions that made sense in the moment but became liabilities.

AI is following a similar pattern. Deployed at pace, without governance, safeguards or considered architecture, without a clear answer to who is accountable when something goes wrong. Unlike previous platforms, the cost of running AI at scale is not predictable. Costs accumulate in ways most organisations have not modelled and will not understand until the invoices arrive. But this cycle is more consequential. With enterprise platforms the uncertainty was in the implementation. With AI it runs deeper. Organisations are making strategic bets on a technology whose trajectory, pace and capability remain genuinely open questions.

The report maps five possible futures, worth reading not as a prediction but as a prompt. At one end, highly capable but unpredictable AI released into the world, open source and largely ungoverned, creating serious risks from misuse alongside potential benefits nobody has yet worked out how to safely capture. At the other end, AI disappoints, capabilities fall short, investor confidence cracks, and organisations committed to AI-led transformation are left explaining why the returns never materialised.

Reality rarely lands at either extreme. But most organisations are planning as though one of these futures is already decided. They have made a strategic assumption about where AI is going, built plans around it, and not seriously examined what happens if they are wrong. That is not a small risk. It is the same pattern after every wave of ungoverned technology adoption. Move fast, govern later, spend years on the consequences.

This applies as much to a retailer, a financial services firm, a healthcare provider or a manufacturer as to a technology business. Every organisation with a plan that touches AI has made assumptions about how that technology will develop. The question is whether those assumptions have been tested against more than one possible future.

Are you using scenarios to inform your strategic planning, or is your strategy quietly banking on just one?