Implicit Bias in Predictive Data Profiling Within Recruitments

Abstract : Recruiters today are often using some kind of tool with data mining and profiling, as an initial screening for successful candidates. Their objective is often to become more objective and get away from human limitation, such as implicit biases versus underprivileged groups of people. In this explorative analysis there have been three potential problems identified, regarding the practice of using these predictive computer tools for hiring. First, that they might miss the best candidates, as the employed algorithms are tuned with limited and outdated data. Second, is the risk of directly or indirectly discriminate candidates, or, third, failure to give equal opportunities for all individuals. The problems are not new to us, and from this theoretical analysis and from other similar work; it seems that algorithms and predictive data mining tools have similar kinds of implicit biases as humans. Our human limitations, then, does not seem to be limited to us humans.
Document type :
Book sections
Liste complète des métadonnées

Cited literature [30 references]  Display  Hide  Download

https://hal.inria.fr/hal-01629166
Contributor : Hal Ifip <>
Submitted on : Monday, November 6, 2017 - 10:12:03 AM
Last modification on : Monday, November 6, 2017 - 10:15:45 AM

File

447020_1_En_15_Chapter.pdf
Files produced by the author(s)

Licence


Distributed under a Creative Commons Attribution 4.0 International License

Identifiers

Citation

Anders Persson. Implicit Bias in Predictive Data Profiling Within Recruitments. Anja Lehmann; Diane Whitehouse; Simone Fischer-Hübner; Lothar Fritsch; Charles Raab. Privacy and Identity Management. Facing up to Next Steps : 11th IFIP WG 9.2, 9.5, 9.6/11.7, 11.4, 11.6/SIG 9.2.2 International Summer School, Karlstad, Sweden, August 21-26, 2016, Revised Selected Papers, AICT-498, Springer International Publishing, pp.212-230, 2016, IFIP Advances in Information and Communication Technology, 978-3-319-55782-3. ⟨10.1007/978-3-319-55783-0_15⟩. ⟨hal-01629166⟩

Share

Metrics

Record views

59

Files downloads

17