Just Hiring! How to counter discrimination in AI-assisted recruiting within your organisation. A toolkit for HR professionals
Contains bibliogr. pp. 68-73
"In this Toolkit, we offer introductory information and concrete recommendations on how to mitigate the risk of discrimination in algorithmic hiring systems within organisations. This Toolkit addresses HR professionals and recruiting agencies and intends to help them gain a better understanding of what they can contribute." (Page 4)
RECOMMENDATIONS
1 Taking action against discrimination is key and should be done at multiple levels. Key recommendations for HR professionals, 9
PART 1: WHY IS ALGORITHMIC DISCRIMINATION PROBLEMATIC?
2 Algorithmic discrimination in hiring is a lose-lose scenario that can't be tackled on a technical level only, 16
3 From the job ad to the final selection – discrimination can occur at each stage of hiring processes that involve AI. Sources of bias and other challenges in addressing discrimination in algorithmic hiring, 19
4 Algorithmic discrimination in hiring is real - and significantly affects real people. Experiences of navigating in opaque hiring processes, 24
PART 2: WHAT DOES THE LAW SAY AND WHAT LEGAL LOOPHOLES DOES IT LEAVE?
5 EU Non-Discrimination law: important safeguards, but with limits. A long tradition that needs some updates, 32
6 AI Act: AI hiring tools fall under the high-risk category. The EU Artificial Intelligence Act (AI Act) and AI–based hiring, 36
7 AI Act: Deploying AI hiring tools comes with a special responsibility – and with legal obligations. The AI Act’s obligations for providers and deployers of AI high-risk systems, 38
8 GDPR: Monitoring for discrimination using sensitive data is a necessity – but also a dilemma. Detecting discrimination in AI systems once they are in use, 41
PART 3: WHAT CAN BE DONE TO COUNTER ALGORITHMIC DISCRIMINATION IN PRACTICE?
9 Implementing an AI hiring tool should be a well-considered choice. Good decisions start with the right questions: defining the problem, setting goals, evaluating risks, 48
10 Algorithmic impact assessments should become the new normal. The necessity to evaluate risks before and during the system is put to use, 50
11 It is crucial to monitor AI hiring systems once they are in use. Secure Multi-Party Computation: an effective and GDPR-compliant protocol, 51
12 Algorithmic auditing should become standard practice. Audits, a structured way to uncover discrimination and prove compliance, 56
13 Transparency is essential to making hiring fairer – but not sufficient. Good standards in transparency and explainability benefit all stakeholders involved, 59
14 Your perspective as an HR professional is needed, from development to monitoring. Tackling the complexity of discrimination needs an interdisciplinary approach, 62
1 Taking action against discrimination is key and should be done at multiple levels. Key recommendations for HR professionals, 9
PART 1: WHY IS ALGORITHMIC DISCRIMINATION PROBLEMATIC?
2 Algorithmic discrimination in hiring is a lose-lose scenario that can't be tackled on a technical level only, 16
3 From the job ad to the final selection – discrimination can occur at each stage of hiring processes that involve AI. Sources of bias and other challenges in addressing discrimination in algorithmic hiring, 19
4 Algorithmic discrimination in hiring is real - and significantly affects real people. Experiences of navigating in opaque hiring processes, 24
PART 2: WHAT DOES THE LAW SAY AND WHAT LEGAL LOOPHOLES DOES IT LEAVE?
5 EU Non-Discrimination law: important safeguards, but with limits. A long tradition that needs some updates, 32
6 AI Act: AI hiring tools fall under the high-risk category. The EU Artificial Intelligence Act (AI Act) and AI–based hiring, 36
7 AI Act: Deploying AI hiring tools comes with a special responsibility – and with legal obligations. The AI Act’s obligations for providers and deployers of AI high-risk systems, 38
8 GDPR: Monitoring for discrimination using sensitive data is a necessity – but also a dilemma. Detecting discrimination in AI systems once they are in use, 41
PART 3: WHAT CAN BE DONE TO COUNTER ALGORITHMIC DISCRIMINATION IN PRACTICE?
9 Implementing an AI hiring tool should be a well-considered choice. Good decisions start with the right questions: defining the problem, setting goals, evaluating risks, 48
10 Algorithmic impact assessments should become the new normal. The necessity to evaluate risks before and during the system is put to use, 50
11 It is crucial to monitor AI hiring systems once they are in use. Secure Multi-Party Computation: an effective and GDPR-compliant protocol, 51
12 Algorithmic auditing should become standard practice. Audits, a structured way to uncover discrimination and prove compliance, 56
13 Transparency is essential to making hiring fairer – but not sufficient. Good standards in transparency and explainability benefit all stakeholders involved, 59
14 Your perspective as an HR professional is needed, from development to monitoring. Tackling the complexity of discrimination needs an interdisciplinary approach, 62
