Risks and rewards of AI in HCM

In an article looking into artificial intelligence (AI) in human capital management (HCM) published in February, Rod Garlick, head of innovation technology and the future of work at Citi Global Insights, and John Burke, global head of business and professional services at Citi Commercial Bank, said that while AI can help in over 90% of a typical hiring process, it is a support tool, and thus there are many sensitive and personal HR areas where AI is inappropriate.

“There are many other areas when it is imperative that humans make the final decision, such as hiring or firing,” they wrote.

The pair stated that AI can be used to judge personality traits more accurately than humans via psychometric testing. 

They noted that deep learning and generative AI (GenAI) has accelerated the capability and usefulness of AI. For instance, by taking a previously analogue task such as reviewing a CV and digitising it, more meta-data is available to train AI models. “This is not just taking place for HR teams but for candidates, so we expect candidates to apply for more roles using AI matching engines,” they added.

AI is also being applied in talent intelligence, matching internal staff with vacancies and identifying upskilling opportunities.

Like many HCM software providers, Oracle has embedded AI capabilities into its HCM software. The AI in Oracle HCM is being used to construct knowledge base articles from frequently asked questions; and develop survey questions to understand employee sentiment and develop long-form content on hiring topics, such as benefits, the workplace environment, and company culture. Oracle also provides a chatbot to enable employees to ask HR-related questions.

In March, Nokia said it would be standardising its HCM systems onto the Oracle HCM platform. The telco equipment provider is deploying Oracle Fusion Cloud Human Capital Management (Oracle HCM) to consolidate 14 different HR applications on the cloud as part of a company-wide digitisation programme. The new HR platform, named NokiaME, will replace on-premise systems previously operated at business unit or country level and streamline more than 100 HR processes in a single integrated HR management system.

When asked about using AI in human resources, Lisbeth Nielsen, head of people experience at Nokia, said: “We’re looking at different use cases for GenAI at the moment, but we haven’t decided where to go first.”

Among the areas she expects the company will look at is deploying AI chatbots such as Oracle’s digital assistant and using AI in HCM analytics. “Recruitment and skills is another area, but we are very conscious about responsible AI,” said Nielsen. 

She said Nokia has had a number of open discussions on where and where not to use AI: “These are very open and active discussions. I think analytics and the understanding of the company is definitely one side of that. You really need to ask if this really is something we want.”

Nielson recognises that if an HCM system is going to be successful, employees need to use it and trust it is working on their behalf rather than collecting data for the company to scrutinise. For instance, a chatbot could offer staff a way to ask questions about their personal circumstances they would not wish to share with their employer. “We have to be very open and clear with people about what we’re actually doing so people can trust the system and trust us as a company,” she said.

AI systems are trained on data and use data to make decisions. If that data is biassed, the systems will be, too. But humans are also biassed. One piece of research shows medical school candidates were 10% less likely to be offered a place when they were interviewed by a human on rainy days versus sunny days.

Citi’s Garlick and Burke believe that if AI algorithms can be designed to remove subjectivity via training and fine-tuning, it may work better than human biases in recruitment. “DEI [diversity, equity and inclusion] can improve with better data and measurement,” they said. “Good implementation, audits and supervision are key.”

However, they recommended keeping humans in the recruitment process. “AI recommendations are just that – a tool to help inform, with approvals and decisions made by humans.”