AI safety: It starts with people, not technology

I spend a lot of time discussing AI and, if you’re like me, you’ll quickly come across a familiar narrative: automation, efficiency and the belief that technology will continue to reshape how organisations operate and grow.

From my position at Cadent, that is not how it looks at all.

In CNI (Critical National Infrastructure), the conversation about AI is fundamentally different. It’s not about replacing people. It’s about how we use technology to better support them. Because ultimately, our colleagues are the ones carrying responsibility for keeping gas flowing safely and reliably.

One of the first things I learnt many years ago in this industry is that the margin for error is incredibly small. We’re not dealing with abstract systems or purely digital outcomes. The decisions we make have real consequences; for safety, for resilience and for the trust people place in our network every day.

This has inevitably shaped how we approach AI.

There’s a lot of emphasis in the ‘AI world’ on speed, including getting solutions live quickly and developing from there. In our world, that mindset must be balanced with a very different set of priorities: assurance, governance and having the confidence that what we’re putting in place will work as intended.

That doesn’t slow progress down unnecessarily. But it does mean progress looks different.

The real value of AI isn’t automation

The most meaningful impact of AI I’ve seen so far isn’t about removing people from processes. It’s about helping them make better decisions.

Whether it’s engineers diagnosing issues, teams prioritising work in the field, or operational colleagues managing complex systems, the common thread is the same: AI provides insight at a scale we simply couldn’t achieve otherwise.

But insight on its own isn’t enough.

Context, judgement and accountability still sits with people. And in a CNI environment, that’s exactly where it should sit.

For me, the real opportunity is in that combination, where technology handles complexity and scale, and people apply experience and judgement to act on it.

The issue is not simply headcount; it is about aligning capability, capacity and support with the work that needs to be delivered.

There’s a lot of external focus on what AI might mean for jobs. Internally, the conversation feels very different.

What we’re really focused on is how roles are evolving, and what that means for the skills we need as a business.

We need more digital capability. We need greater confidence in working with data. And we need people who can operate comfortably in an environment where technology is playing a much bigger role in day-to-day decision-making.

At the same time, the knowledge built up over years of operating complex infrastructure remains absolutely critical. That experience – understanding how systems behave in the real world, how risks emerge, how situations unfold – isn’t something you can easily replicate.

So, this isn’t about replacing one with the other. It’s about bringing those things together.

That’s why, for me, investment in people; reskilling, development, and creating the right opportunities are just as important as investment in technology.

Trust still comes down to people

If there’s one thing that underpins everything we do, it’s trust.

Customers and stakeholders don’t think about algorithms when they turn on their heating or invest in our business. Regulators don’t hold technology accountable. And our colleagues need to feel confident in the tools they’re using if they’re going to rely on them in critical situations.

AI can support better outcomes, but it doesn’t own those outcomes. People do.

That’s why transparency, clear accountability and human oversight aren’t just governance requirements, they’re essential to maintaining trust.

Raising the ‘pressure’, not removing the responsibility

I’m optimistic about what AI can enable within our industry. It has the potential to make our networks smarter, more proactive and more resilient.

But. I don’t see a future where people become less important. If anything, the opposite is true.

As we scale AI, the ‘bar’ rises. The need for good judgement, for accountability and for leadership becomes even greater.

Technology will continue to evolve. That’s a given.

But in CNI environments like ours, one thing doesn’t change: responsibility will continue to sit with our people.