Can we really be responsible for consequences we cannot yet imagine?

How many time a week, or perhaps a day do we talk about or mention risks?  We have been told to identify them, assess them, consider them, weight their possible consequences and decide what to do about them. Them , them and them.

But what about if we don’t know them. If we neither can see or predict “them”. It does not mean that the risks don’t exist.

This is something I have been thinking about in relation to artificial intelligence simply because we are still learning what it can do, how it behaves and what its longer-term impact might be.

And perhaps there is an interesting philosophical angle to this. Living with uncertainty.

A few years ago, I read a book by Zygmunt Bauman, Liquid Modernity, which I actually destroyed (it ended up in the washing machine after I finished reading it, but that is another story). Bauman described a society where things are constantly changing and where the structures and certainties we used to rely on become less stable. I find this idea quite relevant to what we are experiencing today. We are introducing technologies that could change how we work, make decisions and interact with each other, while we are still trying to understand their implications.

Of course, uncertainty has always been part of life. But perhaps what is different today is how quickly things can change and how far the consequences of a decision might reach. It makes me wonder whether we are becoming better at dealing with uncertainty or simply getting more used to living with it. These days, I see young people protesting and rebelling against a world full of uncertainty that, to some extent, we have created for them. Whether I agree with their methods or not, seeing them questioning things gives me some hope. But back to our discussion, knowing how to do something doesn’t necessarily mean knowing whether we should. Aristotele made an interesting distinction between different forms of knowledge, particularly practical wisdom and technical knowledge. In simple terms, being capable of doing something does not necessarily mean knowing whether it is the right thing to do. (No, I haven’t read all of Aristotele’s works yet, but I have started exploring some of his ideas and already found quite a few interesting points to reflect on.)

And does the ability to make a decision better or faster automatically mean that we should delegate that decision? I don’t think the answer is always straightforward.

There is still a difference between what technology allows us to do and what we consider appropriate to do with it.

The more we can do, the more responsible we become?

Hans Jonas explored something similar in The Imperative of Responsibility. His argument was that modern technology has given us powers that previous generations could hardly imagine. And with those powers come responsibilities that may extend well beyond the people making the decisions. And this is were thing get messy. Think about something relatively simple.

A decision made by a small group of people developing or introducing an AI system could eventually affect thousands, perhaps millions, of people who have never heard of that company or had any say in the decision. Most of those consequences might be positive. But what about the ones nobody anticipated?

We cannot realistically expect anyone to predict everything. Yet, if we already know that our understanding is incomplete, perhaps that should influence how we approach responsibility.

And this brings us back to the same question: what level of risk is acceptable, and who gets to decide?

AI could bring substantial benefits to society, but it also introduces risks that may be difficult to fully understand. Would someone benefiting from AI make the same decision as someone potentially harmed by it?

In pharmaceutical Quality, part of our responsibility is to assess risks objectively, without allowing commercial interests or personal preferences to influence our judgement. But AI is often developed and introduced by businesses, where commercial priorities may not always align with ethical considerations. This does not mean that business leaders lack ethical principles, but it raises questions about how these competing interests are balanced.

And if we do not fully understand how an AI system reaches its conclusions, how can we recognise when it is misleading us, even unintentionally? On the other hand, refusing to introduce a useful technology can have consequences too. We could miss opportunities to improve lives or prevent harm.

What matters is how we arrive at that decision, whose interests we consider and how prepared we are to reconsider it when new information becomes available.Not knowing is part of the responsibility. Recognising the limits of our knowledge is not a sign of weakness. It may actually be an important part of making responsible decisions. And maybe the ethical challenge with AI  is understanding how much confidence we should place in our ability to recognise those risks in the first place.

If we know that there are things we don’t know, how do we decide when we know enough to move forward?

---

---

Further reading on EPAI

• Are We Building the Guardrails Fast Enough?
• Responsible Progress in the Age of AI