How AI can eliminate bias in compensation practices
Eliminating bias, or even the perception of bias, is critical to avoiding costly risks to employers in terms of employee engagement, retention, innovation and brand reputation. Employees who perceive bias in their workplace are three times as likely to leave their current job, according to a 2017 Center for Talent Innovation study.
Using AI can be the key to eliminating bias and recruiting talent, says Tanya Jansen, chief marketing officer and co-founder of beqom, a cloud-based compensation provider. But companies struggle with not having a centralized, automated way to manage compensation.
“As a result, compensation decisions are being made without any kind of logical approach, which leads to bias and potential unfairness in how different employees are treated,” she says. “The more companies can leverage technology to automate that process, the more they can remove any bias that’s going to creep in because of other aspects that are uncontrolled.
Jensen spoke with Employee Benefit News about the steps that can be taken to prevent these biases.
What issues do you see in the workplace today in terms of bias?
When things [like compensation] are not being automated and when you're not bringing in enough data to make the right decision, you're by default leaving room for error. You're giving the manager decision-making capabilities that are not necessarily controlled and as a result, you're going to have unfair practices, whether they realize it or not.
[When employers are] not making decisions based on data, they're making decisions based on how they feel, and that's what we're trying to remove. Whether they know it or not, there's still some type of a bias that's going into the decision making process as long as you leave it completely subjective.
How can employers use technology like AI for compensation practices?
There's a lot of assets and data — years with the company, level of education and performance etc. — that goes into deciding why [employees] should get a certain amount of [pay]. If you put all these factors together, [it makes it easier for] the human brain to comprehend all that information, so that's where technology comes into play. AI is using the data to make proposals for things like how compensation structures should evolve in the future, and is removing any type of bias that may [have] resulted in unfair or unequal pay.
In a highly competitive industry where [companies] have very highly paid individuals that are highly performing, there's a big risk they could lose [employees] to their competitors, which is a high cost on them. [Employers] can use AI in order to predict that and understand what areas where they might have retention risks. So anywhere they see that employees could potentially be a flight risk, [technology can help them] understand how to leverage compensation in order to better keep those individuals from leaving their firm and going to the competitor.
Why is it important for employers to address bias?
Because their employees care. Employees are [increasingly] putting pressure on their organizations to prove that [employers are] taking measures to treat them fairly — related to compensation, promotion or other various different aspects — in the workforce.
There's always going to be bias unless you're providing your employees with full transparency on why a decision was made, why they're being paid this, how it's linked to performance, how it relates to their overall total reward package. The more awareness that's being put on the topic of pay equality and removing decision-making bias, the more employers are going to have to prove that they're taking the right steps to address that in order to better retain their people.
By looking at areas where there's been high turnover in the past, and predicting where there might be manager bias, [employers] can prevent [it from happening] in the future. That can allow them to take preventative action on things like turnover, low performance and any other things that might negatively impact the organization.