Data has often been described as the new oil but in truth it’s far more valuable than that.
It is the fuel that powers the energy transition on electricity grids, the recommender engines on ecommerce websites, the targeted advertising that underpins the success of the social media giants and so much else besides.
But data itself is useless unless its collected and curated properly and capable of being accessed, transferred, processed and analysed.
More than that, data is now being transformed by AI, and no business can successfully leverage the latter until it has mastered the former.
“When we are talking about data today, we can’t ignore the AI phenomenon,” says Dennis Tougas, vice-president, global data security, at IT services and consulting firm Kyndryl.
“What we see is that many customers who try to use the transformative potential of AI find themselves running into a challenge in terms of scaling it for enterprise value realisation, because they don’t have their data well organised. They don’t have it curated. They don’t have it permissioned properly, and therefore, it tends to dramatically inhibit – if not completely stall out – their AI initiatives.”
More than ever, when it comes to data, quality trumps quantity.
“It used to be that the more data you captured, the better. But now it’s really about the responsible capture and use of data along with the responsible use of AI,” says Tougas.
Any enterprise that doesn’t have its data estate in order – which means knowing where its data is, what data it has, having it properly classified, and sensitivity labelled – will increasingly struggle to compete.
“If they have issues with inconsistent, incomplete, or biased data sets, they can’t realise its potential,” he says.
By contrast, those that do have built a foundation for true data-driven insights and process enablement which, when coupled with AI, allows them to “start unlocking tremendous value from their data”, he says.
At Kyndryl, he sees clients benefit in this way across almost all sectors.
“One financial services customer uses machine learning to identify anomalous transactions and prevent fraud. They have trained a large language model how to see the different methods and patterns involved so that, as they are processing transactions, they know what to look for,” he says.
“And the better the data is, the better the model gets, the smarter it gets, the less false positives, and the more accuracy in terms of fraud detection.”
It has vastly reduced incidences where a customer uses a card so infrequently that it gets declined by a payments system, for fear its use is fraudulent. In this way, AI not alone keeps financial services customers safer, but it also ensures better customer experience.
“The best companies are doing this more and more intelligently,” says Tougas.
He cites better credit scoring as another case in point, where AI is helping to improve the speed and accuracy of decision making through data-driven models.
What’s more, better data boosts inclusivity. “It is critical to have high quality data to avoid bias, because you don’t want to be making discriminatory practices” that expose the enterprise to reputational damage, he adds.
Healthcare is another major practice area in which AI driven data management is paving the way for innovation, while data security remains paramount.
“We have a client who uses data to identify risks for chronic conditions using AI-driven analysis. They look at the entire pool of data they have trained their large language model on and use it to do predictive health management, looking at symptomatic issues and demographical information about a given individual, and feeding it into the database to look at risk factors and their levels, to develop a diagnosis, prognosis and treatment plan,” he explains.
Another client, a large multinational airline, is working with Kyndryl to leverage AI systems and well organised data to improve customer experience, tracking every touch point along the way, from booking through to post-travel experience.
“They want to know everything about their customer in order to facilitate an excellent experience, a journey where you feel truly and personally known by your carrier,” says Tougas.
In all such cases of process innovation, the first step lies in ensuring your data, and your data management and control systems, are fit for purpose.
That includes cross-border data transfers.
Under GDPR, the EU’s general data protection regulations, any data transferred across international boundaries, and in particular outside of the European Economic Area, must undergo a “transfer risk assessment”.
“This is to determine what data is being transferred, where it is going to, how it is going to be transferred, how it is going to be used, who is using it and whether they have the appropriate technical and organisational measures in place to protect that data from misappropriation,” says Tougas.
It also assesses issues around privacy and whether the data controller or other third parties partnering with the controller have legal permissibility for the data they are capturing.
“It’s why we tell our customers you have to be very thoughtful when you think about a piece of data, or a data asset, for the multiple potential ways in which you may wish to use that data beyond the initial purpose of capture,” he says.
“Because once you feed that data into a large language model, and train that model, extricating that data is virtually impossible, which is why making sure you have the right governance system in place – privacy by design and security by design – is absolutely essential.”
