AI and employment: conclusions of the working group on automated decisions

Conclusions of the working group on the use of artificial intelligence in recruitment and people management processes, with recommendations for companies and public administrations.

The document starts from an uncomfortable premise: most automated screening systems are deployed to gain speed, not to improve the quality of the decision. That difference in objective shapes everything else, starting with what is measured to know whether the system works.

Recommendations

Audit bias before and after deployment, not only at the design stage. Keep a record of rejected decisions, not just accepted ones. Guarantee human review with a real ability to reverse decisions. Inform candidates in plain language.

Sample document for repository layout.