Columbia Group is urging industry leaders to implement artificial intelligence in a way that ensures digital processes do not hinder progress in attracting more women to shipping, arguing that the technology must be embedded carefully if it is to support inclusion rather than reinforce past bias.

The company said AI is built on historical data that may already reflect discrimination against women, making it imperative for businesses to integrate it thoughtfully so that shipping becomes more appealing to future generations of female talent.

Newly appointed group head of AI Christina Orfanidou said artificial intelligence is often presented as though it were neutral, when in reality it depends entirely on the data it is trained on.

“Artificial intelligence is often spoken about as if it were a perfectly neutral system, yet it remains entirely dependent on the information we choose to feed it,” she said.

In shipping, she added, where women make up only a very small share of the global workforce, that information inevitably reflects decades of imbalance, adding that “If we allow AI to learn uncritically from that past, we risk creating tools that appear modern while quietly reinforcing patterns we are striving to change.”

Orfanidou said historical underrepresentation can itself become a problem when fed into AI models, warning that, “If historical data shows women appearing less frequently in certain roles or ranks, a model may interpret that scarcity as a natural rule.”

At the same time, she stressed that AI can reduce administrative work and improve decision-making when applied properly in areas such as crewing, knowledge management and operational planning.

In that respect, Orfanidou said her focus is on embedding AI within the business rather than treating it as a separate function.

She added that it is vital for tools to be built in partnership with people who understand the realities of vessel operations, crewing, safety and human resources, because they are best placed to identify where bias may be hidden.

AI can help reduce time spent on routine tasks and free people up to focus on work that requires judgement and expertise, Orfanidou said, adding that, “when used responsibly it can make the industry more appealing, more modern and more aligned with how people today expect to work.”

Orfanidou, who the company described as an Oxford University-educated doctor of machine learning and information engineering, also warned against over-reliance on the technology, saying it is important to know when to challenge AI because it can be “persuasive even when inaccurate.”

While the lack of globally standardised regulation remains a concern, the company noted that regional frameworks are beginning to take shape. It pointed in particular to the European Union’s AI framework, which includes requirements on non-discrimination.

However, Orfanidou said the outcome will depend on how businesses choose to use technology.

“If we handle AI well, it can support inclusion rather than undermine it,” she said, adding that it can also “broaden access to opportunity, reveal emerging talent and create more consistency in how decisions are made.”

At the same time, she warned that “if we handle it poorly, it can quietly close doors that the industry has spent years trying to open.”

In that sense, she believes AI can either become a force for progress or a mechanism for exclusion, depending on the choices companies make. “At Columbia, our commitment is to build AI that reflects the industry we want to see, rather than the one the data remembers,” Orfanidou concluded.