Artificial intelligence has become the go-to solution for modern recruitment, handling everything from screening CVs and scoring candidates to dispatching automated emails and analysing video interviews. With its promise of speed and efficiency, AI is streamlining HR operations at a pace few imagined just a few years ago.

But while the technology is charging ahead, the path to workplace equity remains riddled with obstacles. New research led by Connie Zheng, Co-Director of the Centre for Workplace Excellence at the University of South Australia, suggests that AI – for all its power – can only advance diversity when embedded within the right organisational environment. In other words, technology alone won’t fix what culture resists.

Chasing speed at the cost of fairness

Much of AI’s appeal in recruitment lies in its ability to fast-track high-volume hiring. As Zheng puts it, “The main goal of using AI is to expedite the process, particularly when dealing with large volumes of job applications.” That focus on speed and cost reduction, however, often leads to blind spots in diversity and inclusion efforts.

When efficiency is prioritised above all else, diversity goals tend to be sidelined. “The problem when the main goal is efficiency is that diversity issues often then take a backseat,” Zheng warns. The risk is that organisations, in their quest to cut corners and quicken hiring timelines, may inadvertently reinforce existing inequalities.

Conditions that make AI work for diversity

Zheng’s findings are drawn from two separate but related studies that go beyond the simplistic question of whether AI or humans make better hiring decisions. Instead, they focus on the conditions that allow AI to support equitable outcomes.

According to the research, AI tools improve diversity only when three key ingredients are present:

1. Transparency and explainability – AI decisions must be traceable and understandable, especially in relation to diversity outcomes.
2. Qualitative over quantitative goals – Hiring should not be reduced to a numbers game. Broader, values-driven goals must guide decision-making.
3. Robust organisational frameworks – Clear and actionable DEI policies need to steer both human recruiters and AI systems.

“These factors encourage HR professionals and decision-makers to reflect more carefully on their choices,” Zheng says. Without these guardrails, AI is just another tool – powerful, yes, but directionless.

Data bias: The ghost in the machine

One of the most pressing concerns around AI in hiring is its dependence on historical data. Algorithms trained on past hiring records often inherit the same biases that HR leaders are trying to eliminate. It’s a case of “bias in, bias out” – and it’s fuelling widespread hesitation among HR departments.

Many organisations are still sitting on the fence, hesitant to fully embrace AI in recruitment. As Zheng observes, part of the reluctance stems from a belief that existing teams are managing well enough. But that sentiment shifts quickly when headcounts shrink or the pressure to fill roles intensifies. “Attitudes change when HR teams face cuts, increased workloads or growing pressure to hire quickly,” she explains.

Tech is only as good as the people behind it

Even the most well-intentioned AI won’t achieve much if human decision-makers aren’t on board. Zheng’s partnership with HUMAINE, a European research network focused on human-centred AI, underscores this point. Their studies show that AI built with diversity in mind still falls short unless organisations actively engage with DEI principles.

“Unless the organisation and its hirers are conscious about diversity and justice issues,” Zheng cautions, “using AI for talent acquisition isn't going to lead to more diverse and inclusive outcomes.” In other words, AI might open the door – but it won’t push it open unless people do.

Turning promise into practice

For business and HR leaders, the lesson is clear: AI is a promising tool, but it isn’t a magic wand. Used wisely, it can support fairer hiring practices. Used blindly, it risks automating old prejudices at scale.

To get the most out of AI, leaders must resist the urge to chase quick wins and instead invest in building an inclusive infrastructure that supports both human and machine decision-making. That means crafting policies with teeth, measuring success beyond quotas, and making transparency the norm – not the exception.

The future of hiring may well be digital, but inclusion still requires a human touch.