When everyone has AI, what sets you apart?

Dubai: William Gibson famously observed that "the future is already here — it's just not very evenly distributed."

It is a useful lens through which to view artificial intelligence. Because AI is already here. Access that would have seemed extraordinary only a few years ago is now available to almost anyone, and increasingly to almost any business. Closer to home, the Gulf has moved with remarkable speed to lay the foundations for an AI economy.

As access spreads, however, the divide is shifting. It will be less about who can access AI, and more about who is deliberately building capability around it.

Access is not capability

If the world's best models can make an employee more productive, help a company serve customers better or enable a government to deliver better services, they should be used. But access to somebody else's intelligence is not the same thing as developing your own capability.

Satya Nadella often talked about "tech intensity": adopt the best technology available while building your own unique capability. AI has made the adoption side of that equation dramatically easier. That makes the capability side more important, not less.

Take a young analyst preparing a recommendation. AI can now gather information, analyse data and draft an argument in minutes, replacing hours of work. Some of those hours were simply inefficiency, and eliminating them is a positive development. But some were where that analyst learned to interrogate information, recognise weak assumptions and form judgment. If AI consistently allows people to skip the difficult cognitive steps, there is a risk of becoming more productive without becoming more capable. It is like going to the gym and getting a forklift to lift the weights: no muscle gets built.

The answer is not to preserve inefficient work for its own sake. It is to ask a more deliberate question: which capabilities should AI replace, which should it amplify, and which do people still need to develop for themselves?

Organisations need to build, not just consume

The same tension exists inside companies. Most organisations will soon have access to broadly similar models and tools, and deploying them cannot, by itself, create sustainable differentiation. The advantage comes from what gets built around them.

Technology can be outsourced and tasks can be automated, but the learning through which an organisation becomes distinctive cannot be outsourced. And that learning only compounds if it lands somewhere the organisation owns — in its data, its workflows and the way it makes decisions — rather than in individual prompts and chat histories that disappear when the session ends.

The data already suggests a divide. PwC recently found that just 20 per cent of organisations are capturing 74 per cent of the measurable economic value created by AI. The models are not reserved for that 20 per cent; the study's own answer is that the leaders are using AI to grow and reinvent, not just to become more efficient.

A simple way to think about where any organisation sits is consume, adapt, create. Consume the best technology the world has to offer. Adapt it to a specific context including customers, data, people and ways of working. And where it genuinely matters, create something distinctive: intellectual property, products or business models that somebody else cannot access simply by buying the same subscription.

The point is not to own every layer of the technology stack, but to know which layer of capability cannot be outsourced.

The same question applies to countries

This becomes even more important at a national level. The Gulf has moved fast: the UAE and Saudi Arabia have invested in compute, research, talent and partnerships with the world's leading technology companies.

But long-term competitiveness cannot come only from being an exceptional consumer of technology created elsewhere. The next question is what capabilities countries want to build locally, and which globally relevant companies can emerge from this region. There are encouraging signals. Falcon, Jais and ALLaM are models built here for the region's language and context, and national champions like G42 and HUMAIN are building AI infrastructure and capability at serious scale.

Capability, though, is deeper than a handful of models: it is the accumulated ability of an economy to understand technology, adapt it, build on it and eventually create something others value.

This is not an argument for technological self-sufficiency; no serious innovation ecosystem works that way. The argument is for intentionality. Use the world's best technology. Attract the world's best talent. Build with the world's best partners. But be equally deliberate about what knowledge, skills, companies and capabilities should be developed yourself.

What are we becoming?

AI adoption and productivity gains will naturally be measured. The questions that matter more are different. What can people do today that they could not do a year ago? What organisations and economies are now capable of building themselves?

And which human capabilities — deep domain expertise, knowing what to ask, deciding which problems are worth solving, spotting weak assumptions and recognising when the machine is wrong — become more valuable as machines become more capable?

The future is already here. Increasingly, technology will be available to everyone. The capability to shape what is done with it will not be. That is something that needs to be built.