'I'm an advocate for human in the loop': How a critical infrastructure provider is handling AI

And why they’re looking beyond it

Gen-AI offers “a massive opportunity,” but needs to be supported by “the right quality data” - otherwise, “sometimes you just make the wrong things go faster.”

Those errors at a critical infrastructure provider like National Grid could have ramifications for the whole country.

It's something National Grid CIO Sarah Milton-Hunt is fully aware of, and why she's thinking very carefully about both traditional and generative AI.

"It offers a massive amount of opportunity for business and for sustainability: that ability to analyse more data, and to develop out the gen-AI, machine learning and language models."

AI has the potential to be a true game-changer for capabilities and productivity, but - she warns - we have to understand its limitations, especially when it comes to data.

"The reality is you can put brilliant capabilities on top of it, but if you're running it from biased or skewed data...sometimes you just make the wrong things go faster."

Sarah believes AI can be transformative, but like most IT leaders, she remains cautious.

"I'm an absolute advocate of AI, but I am also an advocate of having the human in that loop. We need that intervention; everything isn't binary, and sometimes [AI] needs some critical thinking that we cannot as yet industrialise into an algorithm or into a learning capability."

Using AI safely comes down to pairing the human and the system to "bring all the nuances of analysis together." Data, inevitably, underpins that.

National Grid is "moving to a very strong strategy around data mesh." Not only will this support the organisation's use of AI, it will also add to its ESG credentials by consuming data in situ instead of being forced to move it around for analysis.

"Large replications of datasets just mean that we're taking up space in datacentres we potentially don't need to take up."

While more or less every IT leader is investigating AI, in a sector as volatile as energy it pays to be even more forward-looking. That's why Sarah is also considering how artificial intelligence will play with tools like quantum computers.

"If you stick AI capabilities on top of quantum computing, the potential that offers us is fundamentally, monumentally changing. It changes what the possibilities can and could be."

Even though commercial quantum computers could be ten years away, National Grid is building an energy environment that will need to last for decades more.

"It has to be on our radar," says Sarah.