Algorithms that companies use to advise on who should fire the company in a round of layoffs sometimes lead to incorrect advice due to limited or incorrect data, experts warn.
Almost all companies in a survey among 300 US companies say they use algorithms for layoffs, writes The Washington Post. Yet by no means all those companies rely on those recommendations, because 41 percent of large companies trust that there is sufficient and good data. For smaller companies this is 25 percent.
For example, companies may include the risk that an employee would want to leave in the near future as a factor in the algorithm. If, on average, black employees leave more often due to problems with discrimination and the algorithm does not understand that cause, it could recommend more black employees for dismissal.
Moreover, there is a risk in using incomplete data, which means that the algorithms make decisions with the wrong input, which then leads to wrong conclusions. According to the newspaper, many HR departments use algorithms to relieve the workload.
A top executive of a company that makes HR software says in the newspaper that companies must be transparent about the use of algorithms and which factors participate in decisions. The use of algorithms in many places has been a topic of discussion for years, because incomplete or incorrect data, or by weighting certain factors more heavily or not at all, can lead to incorrect decisions. It is often unclear why an algorithm has reached a certain conclusion.
