Artificial intelligence (AI) has arrived in the mainstream and is augmenting decision making in many industries. From data-driven understanding of students’ needs in education, to surrogate operators in combat and training in aviation, to robo advice dispensed in financial services, AI is available in multiple flavours to all takers. While the boardroom discusses how best to implement it, chief data officers seem poised to take centre stage when it comes to the rollout of AI.

The emergence of the data chief reflects the starring role of data within commercially savvy organisations. A chief data officer (CDO) newcomer, Caroline Bellamy at mapping company Ordnance Survey, explains the mission: “Customers demand we do things [with their data] correctly and ethically. A lot of it is common sense, such as respecting individuals and rights, doing what we say we do. As data professionals, we have to hold in our hearts that we are doing the right things for both individuals and the organisation.”

Algorithmic outcomes can be notoriously unpredictable, and that is the galvanising truth for CDOs — remember the Tesla driverless car that mistook a tractor-trailer for the sky and fatally crashed? The Microsoft bot that learnt from bad actors? And the chatbots that may have influenced the 2016 US election? For this reason, Mrs Bellamy counsels transparency as the guiding principle for any organisation considering implementing AI.

“Outcomes can be unpredictable using new technologies such as AI, but activities can and must be contained and constrained,” she says. “Critically, it’s important to communicate to customers about algorithmic activity that uses their personal data so that they are aware and can give their consent.” Outcomes can and should be contained, she argues, by business leaders and data scientists focusing on areas where AI drives the most advantage.

Ordnance Survey harnesses AI to speed up imaging data, to do things faster and to respond quicker in ways unimaginable to the public. “Our ability to process data and see patterns is way ahead of what the public and employees imagine,” says Bellamy.

And knowledge and AI are accelerating, she points out, saying: “We have to bring in and embrace the new thinking inside our organisation. We have to keep abreast of capability, while wrapping it with ethics and legal obligations.”

Another issue for CDOs to consider, especially those working in or partnering with the public sector, is how algorithmic learning can be shared for the public good. Typically, in-house governance concerns itself with a narrow set of duties, says Bellamy, who queries how the benefits of AI can be used for the civic good.

“To sense-check AI governance, we need ethics bodies,” she says. “When [decision-making] happens only in closed rooms with one agenda — then we’re in trouble.”

For these reasons — the immature and unpredictable nature of AI, coupled with its scalability — the key role of the data officer is to hold his or her organisation to account, says Bellamy. “A core part of the job is to fly the flag for ethics, and this may entail difficult calls such as putting a project on pause to consider ethics,” she says.

It isn’t something that is necessarily written into the job description, but it should be the essence of the CDO’s DNA, she adds. “If you don’t take care of it, it will ruin everything going forward and lose the faith of your customer, the public and society.”