AI and HR
For a long time, Artificial Intelligence (AI) seemed to be the new panacea within HR. This would allow you to select and match more fairly, i.e. without prejudice. Inclusion thanks to an algorithm. But now there is doubt. Can AI live up to that promise? As an HR manager, do you have sufficient insight into how AI works to be sure about the outcomes? Is algorithmization in recruitment really such a panacea? Jacqueline van Breemen researched the use of AI in selections and came to a different conclusion: Selection with AI reinforces existing prejudices, there is no algorithmic inclusion.
"Labor market discrimination often happens unconsciously," says van Breemen. "With selection based on an AI algorithm, the idea is that you exclude that unconscious discrimination because a computer does not see who it is. He cannot therefore reject applicants on improper grounds. That sounds nice, but in practice it is different, according to my research."
Algorithms have been used for some time in selections, rules in software for processing information. This information can consist of, for example, the psychological characteristics of an applicant. Usually the algorithms are designed by humans, but with AI the software does that itself, you work with a learning system, machine learning. Algorithms are created because the software discovers patterns in the collected information that are then used in subsequent assessments. For example, the data of existing employees can play a role in testing new applicants. The tricky thing is that the patterns found through AI and the algorithms that follow from them remain invisible. Van Breemen: "There is no transparency in the system. And that is the main disadvantage."
In addition, her research confirmed that the use of AI does not lead to the desired objectivity. Existing prejudices about groups of employees or students were simply adopted or sometimes even reinforced. The result was overestimation of the chances of success for candidates with a Dutch background and underestimation of the chances for people with a non-Western background. And that is at odds with the desire from practice to recruit and select as objectively as possible.
"Unfortunately, there is often a lack of knowledge about the operation of the selection methods and especially about the objectivity of those methods," says Van Breemen. "As an HR manager, it is important to know on the basis of which criteria you choose a certain selection process and a certain selection method. What are the relevant properties that you want to be able to measure? In which situation can you use which model? How do you arrive at an objective assessment? And perhaps most importantly: How do you ensure a transparent process?"
Jacqueline van Breemen was able to carry out her research thanks to a research prize from NSvP. She also received the funds to draw up a further development plan, intended to gather more knowledge about how AI works within HR. And, just as important, to disseminate that knowledge through workshops with recruitment and recruitment agencies. Van Breemen: "AI is very suitable for processing factual information and assessing it. But it is still too early to apply it to recruitment and selection, when human characteristics have to be assessed."
Read Jacqueline van Breemen's research here .
