Student opinions about algorithms in applicant selection
Discrimination still plays too large a role in recruiting. Fair selection of candidates is a major challenge for companies. The danger of subconscious and sometimes even conscious discrimination lurks in too many places. As a result, many young people fear that they will have fewer opportunities on the job market because of their gender, origin or sexual orientation.
To prevent this, many companies are starting to look into the use of algorithms in recruiting. Besides many potential benefits (see our validation studies of our CASE algorithm), the use also brings risks. This is because an algorithm can only make decisions based on the data with which it is trained. So if there are already undetected wrong decisions in the training dataset, the algorithm may incorrectly reproduce these decisions and lead to unfair decisions. While 61% of recruiters consider the future use of digital tools in personnel selection to be significant, the position of applicants in general and students in particular on the use of algorithms has not yet been adequately captured. In this regard, we asked ourselves what future employees think of an exclusively algorithm-based applicant selection.
As part of the "Fachkraft 2030" special evaluation (Maastricht University and Job Valley), around 15,500 students were asked in March 2020 about the acceptance of algorithms in personnel selection. The following criteria were used to ask whether humans, algorithms or a combination of both should be applied. In addition, specific application characteristics such as (1) transparency of decisions, (2) avoidance of wrong decisions, (3) reduction of discrimination, and (4) speed of decision-making were considered.
Students favor combination between humans and machines
From the results, it was found that the sole use of algorithms in personnel selection is viewed critically, whereas the majority favors the general use with human supervision. In addition, it was found that foreign students, with over forty percent, generally consider the use of algorithms in personnel selection to be (more) useful than domestic students. It is also striking that female students tend to be more skeptical about the general use of algorithms in personnel selection than male students.
Algorithms can help to reduce discrimination - students find
With regard to the criterion "reduction of discrimination", the picture is much more balanced than with regard to the general acceptance of algorithms. A clear majority of students prefer decisions to be made by algorithms (31%), or in combination with algorithms (34%). To avoid wrong decisions in the selection process, the majority of respondents prefer a combination of human and algorithmic decisions. Furthermore, with regard to transparency of decisions, 77.5% of students would like to see the use of humans or a combination. Lastly, it can be highlighted that in the criterion of "speed of feedback", almost half of the students (49%) prefer the algorithm.
The use of algorithms should be well considered, but in no case categorically rejected
Algorithms in applicant selection represent a helpful and time-saving tool in recruiting. However, they should be well-considered and should not have any discriminatory factors, which is considered rather difficult so far. In fact, this requires clean validation and data collection to improve the technology. So that there are also no worries with the employee of tomorrow in the application process, it is desirable if the selection process is never determined by algorithms alone, but at the same time human co-supervised, in order to be able to prevent inequalities, as it emerges from the Fachkraft 2030 study. That the CASE Score does not discriminate and why it can be a good tool for algorithmic support in personnel selection, we have highlighted here.https://www.candidate-select.de/en/blog/use-of-algorithms-in-the-application-process