NORMA eResearch @NCI Library

Evaluating Discrimination Bias In AI Decision Making Systems For Personnel Selection

Avgustin, Viktor (2022) Evaluating Discrimination Bias In AI Decision Making Systems For Personnel Selection. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
PDF (Master of Science)
Download (1MB) | Preview
[thumbnail of Configuration manual]
PDF (Configuration manual)
Download (752kB) | Preview


This research work aims at establishing a blueprint for conducting an ethical audit on Artifical Intelligence (AI) algorithms which are used for decision making when pre-screeing job applicant resumes to be considered for vacant positions. AI is widely used in industry for pre-screening CVs due to the cost-savings and ability to filter through large quantities of applicants. However, over reliance on these type of AI systems creates various ethical considerations. Most AI algorithms employed in CV screening are akin to ”black boxes” in that they lack transparency in the way decisions are made. This research work highlights the importance of incorporating ethical considerations when building and training the algorithms in order to ensure that the production version is free of any discriminatory bias. In order to illustrate this, three classifier systems are built - kNN, Linear SVM and CNN to match CV of job applicants with job descriptions. The results are compared by adding gender as a sensitive variable to determine if any of the algorithms are bias towards gender e.g. selecting a larger proportion of a certain gender. The paper finds that there is a wide variation in the gender proportion across the three classifiers for the same job category. This indicates that a particular gender may be at a disadvantage based on the classifier used in the selection process. This paper advocates the building of robust classifiers which incorporate discriminatory variables in the training process to ensure that bias is eliminated when the classifier is deployed.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence
Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence
H Social Sciences > HD Industries. Land use. Labor > Issues of Labour and Work > Job Seeking
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Tamara Malone
Date Deposited: 17 Jan 2023 17:21
Last Modified: 07 Mar 2023 11:12

Actions (login required)

View Item View Item