"How are you doing?" It is the most routine question you can get, and it comes with an equally routine answer: "Fine, thanks." We nod, smile, and move on. Nobody expects more, and few of us offer it.

Lately, if I am honest, my real answer is "Not great”. That is an awkward admission for someone who has spent more than 30 years helping organisations through digital transformation and who now spends most days advising leaders on AI. I am supposed to be the calm one with the answers. Yet right now I feel overwhelmed.

It is not hard to see why. Every day brings another story. AI systems behave in ways their designers did not intend. AI agents take actions nobody sanctioned. Criminals use AI to scale fraud and impersonation, and security officials warn that it could help those planning acts of terror. And don’t even get me started on what seems to be an unstoppable AI arms race between the US and China for dominance in developing and applying AI.

As if that isn’t enough, the existential headlines follow. Geoffrey Hinton, the Nobel prize-winning computer scientist often called the "godfather of AI", has said there is a 10 to 20% chance that AI could lead to human extinction within the next 30 years. This month, a new Lancet Commission assessment named the malicious use of AI, including AI-enabled bioweapons, as one of only two threats that could end humanity by 2100. The other is nuclear war.

Even if we pull back from the brink, the immediate economic forecasts are just as worrying. Elon Musk predicts that within 10 to 20 years, AI and robotics will make work optional and money irrelevant. Others are waiting for the AI bubble to burst, and even the Bank of England warns us that AI valuations look stretched and that a sharp correction could hit the wider economy soon. Both cannot be right. Yet both are delivered with complete confidence.

If you feel the same way, you are not alone. Surveys show that people are getting more concerned than ever about where AI is leading us. In all this turmoil, what approach should I adopt to AI to make sense of it all?

Two responses are tempting, and both are mistakes.

The first is to put your head in the sand. Treat it all as hype, wait for it to blow over and get on with the day job. Given how often technology has been oversold, that instinct is understandable. However, AI is already changing how work gets done, how fraud gets committed, and how decisions get made. Looking away doesn't stop any of that. It simply means the important choices get made by someone else.

The second is to run around shouting "the sky is falling!". That means responding to every alarm, freezing investment or banning tools outright. It feels responsible, but it rarely is. Panic crowds out judgement. Furthermore, organisations that freeze or introduce blanket bans often find their people using AI anyway, out of sight and without any guardrails at all.

So, what lies between?

Here is how I'm trying to avoid being overwhelmed by AI, in five steps.

1. Name it. Admitting that you feel overwhelmed is not weakness. It is an accurate reading of an information environment that rewards the most extreme claims. It also matters for leaders. If you are feeling this, your teams and the people around you almost certainly are too, and most of them are answering "fine, thanks" as well. Saying "not great" out loud gives others permission to be honest. Honest conversations are where sensible decisions start.

2. Sort the worries. Much of the feeling of being overwhelmed comes from treating every AI story as the same kind of problem: equally urgent and equally yours. They are not. Some harms are here now and well documented, such as AI-enabled fraud, deepfakes, and agents acting beyond their authority. Others are serious but speculative, such as extinction risk, a post-work utopia or an economic collapse. The first group calls for action this quarter. The second calls for attention, preparation, and public engagement, not daily anxiety. Sorting your worries this way does not make them smaller. It makes them manageable.

3. Read the contradictions as a signal. When credible people confidently predict both paradise and collapse, the honest conclusion is that nobody knows. That is not a reason for paralysis. Rather, it is a reason to stop betting on any single forecast. Good strategy under deep uncertainty looks for "no-regret" moves, meaning things worth doing whichever future arrives. Strengthening governance, building AI literacy across the workforce, designing for resilience and avoiding dependence on one vendor or one narrative all qualify. None of them requires you to know who is right.

4. Focus on what you can control. You cannot change Geoffrey Hinton's estimate or the Bank of England’s forecast. You can decide how your organisation deploys AI agents, what it measures, who is accountable when something goes wrong, and how quickly you would know. Most of the anxiety lives in what we can see but cannot influence. Most of the useful work lives in what we can influence. Put your energy there.

5. Ration your attention and engage collectively. Doomscrolling is not research. Choose a small number of trusted sources and review them on a regular basis, rather than reacting to every headline. Then channel your concern into places where it can shape outcomes: professional bodies, public consultations, sector networks and support for institutions such as the UK's AI Security Institute. Individual worry achieves little. Collective, informed pressure is how societies have tamed powerful technologies before.

Pilots facing an emergency are taught a simple order of priorities: aviate, navigate, communicate. First keep the aircraft flying. Then work out where you are going. Only then talk about it. The lesson is not to ignore the alarms. It is to deal with them in an order that keeps you in the air and moving forward.

That feels like the right way to address AI today. Not the ostrich, and not Chicken Licken, but something closer to calm urgency. Take the risks seriously, act on the ones in front of you, prepare for the ones that might come, and refuse to let the noise make your decisions for you.

I am still not "fine, thanks". But I feel a little less overwhelmed now that I've written this down. Now I just need to not get distracted by the next AI story I read.