Artificial intelligence should not be used in employee remuneration just because it’s an exciting new technology.

“We need to identify real opportunities and problems first and then decide if and how we can apply AI to them,” says Andre Daniels, Chartered Reward Specialist and Exco member at the South African Reward Association (SARA).

Avoiding FOMO (fear of missing out)

Many South African reward professionals and employers might feel that they are being left behind, which could lead to panic-driven AI deployments. This mindset risks investing large amounts of time, money and resources into planning and implementing a solution that never yields a net positive result.

“The use of AI in remuneration should rather be purpose driven, focused on solving specific problems and creating tangible value, but not adopting it for its own sake,” says Daniels. 

Automation vs AI

Speaking at SARA’s 2024 Conference, Johan Steyn, founder of AI for Business, warned that AI should not be mistaken for automation. “There’s a very good chance that AI is not what you need to achieve efficiencies and effectiveness,” he said. After analysing a given problem, it may be found that simple automation is a more appropriate solution than AI.

AI can certainly contribute to efficiency, reward system transformation, bias and fairness measurement, annual salary reviews and streamlining other processes. Yet, it promises far greater utility than classic automation.

Data utilisation

Daniels says he was inspired by Steyns’ presentation and assertion that AI is best utilised to discover meaningful patterns in corporate data.

Because of this, Daniels’ main focus currently is the value potential of AI in remuneration that lies in its ability to mine insights from remuneration and related data. In this way, it’s much like LiDAR, a laser mapping technology that can reveal archeological sites hidden deep underground.

“Similarly, AI can be used to reveal hidden patterns in remuneration data that inform powerful strategic decisions, reward policies, business processes and equity approaches,” says Daniels.

Building on flexibility

Even now, many organisations implement ERP solutions, usually at immense cost, only to see them fail to deliver the envisaged business-critical outcomes. Definitely, any AI-based implementation should be flexible enough to adapt to both current and future needs.

This will ensure that employers are investing in a bespoke solution that can evolve within its changing business environment, rather than something generic that doesn’t do everything it needs to.

Starting with the obvious

There is an opportunity to identify quick wins for some of the more obvious applications.

One is managing complexity. Do we really still need people to manually compile remuneration disclosures, or could an intelligent agent do this far more efficiently and accurately?

Another is managing bias. For example, can an AI model with access to an employee database carry out equal-pay-for-work-of-equal-value (EPWEV) analysis and judgements better than a human?

Or could AI help you refine your remuneration processes so that time-consuming admin and labour are eliminated?

Certainly, AI has been shown to quickly turn raw data into rich insights to inform and empower decision making at all levels of the organisation, including remuneration.

Beyond the obvious

After the quick wins, organisations can start looking at less obvious applications.

Could AI assist with the alignment of reward strategy to remuneration practices, ensuring cohesion between these concerns?

Might it also integrate remuneration with other HR practices, such as analysing the talent profiles of individual workers to help reward practitioners understand the reward preferences of a workforce? Or allowing them to understand how and when people prefer to be recognised?

Would employees, who sometimes feel embarrassed discussing their personal challenges with another human, find comfort in seeking guidance from an impartial AI agent? Not the limited chatbots with preprogrammed responses, but a language model that is trained to offer, say, realistic advice and assistance in a non-judgemental way.

Starting off right

Before considering AI as a one-size-fits-all solution, organisations need to identify what they are doing that doesn’t add value or isn’t working, and opportunities to institute step change in their practices. What can they do that would take them to the next level?

This means identifying concrete business problems that must be solved or opportunities that may be exploited, rather than starting with an AI-first approach and expecting to retro-fit the model as and when required.

Only then should they consider which tools or solutions to employ – which may or may not include AI.

Involving reward professionals

For SARA, the burning question is: if AI is identified as a solution in remuneration, what is the role of the reward professional in the future?

This is critical because, if we cannot show them what their future looks like alongside AI, they see technology as an enemy that makes them redundant, and fear progress instead of embracing it. “However, if there is a shared vision of their role in the future, it becomes an exciting opportunity to work towards,” says Daniels.

With clarity on how the reward professional of the future adds value to the business, they can fully understand how to become better strategic enablers in organisations, and communicate their worth confidently.