The question is not only what AI can do, but who is shaping the systems that may guide our future.

For much of the twentieth century, large parts of our future were imagined and organised within the public sphere. Governments and institutions had a central role in setting rules, defining collective interests and shaping a vision of society.

Artificial intelligence has the potential to disrupt some of these pillars, because it can shift influence towards those who produce knowledge, those who make recommendations and, ultimately, those who shape decisions.

AT present, a significant part of public life has already moved into privately owned platforms, where billions of people now communicate, work, exchange information and make decisions. Artificial intelligence represents an even deeper shift. We now share our projects, problems, queries just to mention a few.

AI can also change where knowledge is produced and, indirectly, where power becomes concentrated.

Let’s take healthcare as an example: AI can increasingly influence diagnosis, triage, treatment recommendations, drug development and resource allocation. Used correctly, this could represent an enormous benefit for patients: better diagnosis, earlier detection, more personalised treatments and better use of resources are all realistic possibilities. But there is another side worth considering.

If these systems are developed and controlled by private organisations, even a small systematic bias in the way a model prioritises one treatment, diagnostic pathway or technology over another could potentially influence millions of decisions. And move millions of pounds. The patient may still receive a perfectly safe and clinically appropriate treatment, but the economic consequences could be very different.A relatively small preference embedded within a recommendation system, multiplied across millions of users, could redirect significant amounts of healthcare spending towards one treatment, diagnostic technology or provider.

This does not necessarily require bad intentions. But equally, where so much economic value can be influenced by a system, the possibility of a risk to spark bad intention cannot be ignored.

And that is exactly why the questions become interesting.

• Who controls the infrastructures?
• Who owns the models?
• Who controls the data?
• Who defines the objectives?
• Whose interests are embedded in the technology?
• And if an increasing part of knowledge and decision-making is mediated by AI systems, is regulation alone enough?

Different parts of the world are already approaching these questions in very different ways. In the United States, much of AI development is driven by private technology companies and commercial interests. This brings enormous capacity for innovation, but also raises questions about control, data, influence and accountability. China represents a very different model, with a stronger connection between technological development, the State and centralised control.

Europe, at least from my perspective, appears much more cautious and heavily focused on regulation. Can Europe really regulate technologies that it does not own?

And this is not only a political or economic question. For Pharma, it may become very practical.

If increasingly important parts of our knowledge, analysis and decision-making are supported by AI systems developed outside our organisations, then our dependence is no longer limited to the technology itself. We may also become dependent on the people who control the model, the data behind it, its development and ultimately its future direction.

So we are no anymore simply discussing where AI is taking us. Perhaps we are discussing :

Who is actually shaping the systems that will help us decide where to go?