The pros and cons of using AI in HR

There may be a lot to get your head around, but artificial intelligence in human resources is already a reality. Four experts debate the challenges and opportunities

PROS

How AI can transform recruitment

After several stop-starts, artificial intelligence, or AI, is now the single most important driver shaping the future of work, but has proved itself to be evolutionary rather than revolutionary. So says Ian Jones, co-founder and chief strategy officer at the business software developer AMPLYFI.

While it cannot transform a firm overnight, nor be “a silver bullet to business leaders’ painpoints”, it can deliver “incremental improvements to productivity or planning” provided realistic expectations are set, Mr Jones says.

AI in HR can make a positive impact across everything from recruitment and retention to fostering staff wellbeing, enriching jobs and protecting against brain drain, he believes.

As information overload intensifies, the ability of AI to process big data at a speed, scale, and accuracy that humans cannot match will reduce stress among all knowledge workers.

Increasing productivity and cutting stress

Use of AI to monitor the emotional health of staff has already led some call centres to introduce automated feedback. AI systems are also able to recommend breaks if employees look stressed, frustrated or tired when dealing with customers.

To Dan Marsh, director of people at e-learning provider Sponge, “the roles that businesses are recruiting for today may not exist in the near future”. He notes that AI enables learning platforms to replicate the approach of consumer content providers such as Netflix, thereby “improving learning and outcomes at a time of reduced shelf life for human skills”.

Sue Lingard, marketing director at HR software vendor Cezanne, agrees. “AI in HR may well play a valuable role in helping employees work more productively or suggesting training courses to match career preferences,” she says.

Ms Lingard notes that LinkedIn Learning already recommends courses based on users’ profiles and training histories, and she predicts more of the same.

AI in HR may well play a valuable role in helping employees work more productively or suggesting training courses to match career preferences

HR technology not one-size-fits-all

Andrew Spence, HR transformation director at Glass Bead Consulting, sees great benefits in using technology “to capture a broad range of personal data at work” and says organisations that “remain transparent, empower the workforce and share the benefits” will win out.

“I see employees harnessing HR technology to make better decisions, match projects and tasks on digital work platforms, and use coaching tools to improve insight and performance,” Mr Spence predicts, adding that the entire nine to five will inevitably be redefined.

“Work will be carried out by people, networks, contractors, augmented humans, robots and automated systems all working together,” he believes.

Too many AI initiatives start with businesses feeling pressurised to “implement AI, blockchain or the internet of things simply to keep up with the herd”, says Mr Spence.

“My advice is to go back to business strategy and refuse to allow your organisation to be distracted by all the hype and shiny new tools.”

CONS

AI can learn from discriminatory data

Cezanne’s Ms Lingard notes wryly: “Businesses are yet again being told to change their HR technologies and, as always, some are predicting that those organisations which fail to do so will go the way of the dinosaurs.” While she agrees that AI in HR will become increasingly important, she identifies three fundamental problems.

As the recent pulling of Amazon’s new recruitment AI system, which deselected female candidates on the grounds that data indicated most existing tech workers are male, has demonstrated, inherent bias in AI is hard to eradicate.

“It’s not just about the fact that historic data isn’t necessarily a reliable predictor of the future,” she says. “It’s also about how the framing of a problem, how the initial choices are made and the social context, or lack of it, will all influence what a machine appears to ‘discover’ on your behalf.”

For computers to learn and replicate, they need to crunch a lot of data. With hundreds of thousands of people-related datapoints at their disposal, the AI tools deployed by major corporations may indeed “find patterns that could help predict flight risk or future potential”, says Ms Lingard.

The framing of a problem, how initial choices are made and the social context, or lack of it, will all influence what a machine appears to ‘discover’ on your behalf

Does AI in HR make sense for small firms?

In contrast though, smaller companies just don’t have enough data to benefit from true AI, unless it’s aggregated across a number of organisations and even then there may be problems.

While some recruitment software vendors offer augmented AI video interviewing tools to help streamline mass-recruitment exercises, “most experienced HR directors will have horror stories of candidates who ticked all the boxes, but failed to thrive in a particular business culture, or outliers who looked to be a high risk, but turned out to be superstars,” she says.

The question of whether machines should be allowed to make decision about the future of human beings remains paramount in her view. “Of course, people managers need to get to grips with AI, not least because of the impact it is already having on the workforce, but all HR professionals must be prepared to ask the tough questions,” Ms Lingard adds.

Tech adviser Barry Flack believes HR professionals will eventually “use algorithms to locate the right candidate for a role in a matter of seconds”, but says it is the often-corrupted data relied on by the business community which is of greater concern.

While taking time to clean existing datasets of inherent human bias is now long overdue, full-scale adoption of AI in HR “will remain a figment of the over-active imagination of marketers and geeks for the foreseeable future”, Mr Flack concludes.