First published on
08 December 2020 (last update on: 10 December 2020)

The online Mutual Learning Seminar examined the challenges and opportunities of using AI in selection and recruitment processes, with a focus on gender bias in algorithms. Government representatives and experts from 21 EU Member States participated in the seminar. With the aim of developing a coordinated approach, the European Commission published a White Paper on AI in early 2020, which sets the European approach to trustworthy AI, including a proposal for regulatory framework.

The seminar provided an overview of the main concerns related to the use of AI in labour market recruitment processes and potential sources of gender bias, many of which are hidden or implicit. Because decision-making in the labour market is already biased, AI risks replicating and amplifying those biases. The Netherlands presented their recent online ‘hackathon’, aimed at developing solutions to prevent biases in AI used in recruitment processes. The two winning teams have been offered an opportunity to elaborate on their work in an incubator programme with support from the Dutch government.

In the discussions, participants exchanged ideas on different approaches to address the potential risks of discrimination and strategies to raise awareness of gender bias in algorithms. Solutions proposed included checklists or guidelines, certification and auditing systems, as well as outcome reporting. Cooperation between governmental bodies, interdisciplinary approaches, new legal tools at national and EU levels, awareness-raising and training with different stakeholders, and education from early childhood onwards to tackle gender stereotypes were considered important. Further research on different aspects is needed and it was recommended that a repository of good practices be established.


DownloadPDF - 680.4 KB
DownloadPDF - 756.2 KB
DownloadPDF - 705.5 KB
DownloadPDF - 619 KB
DownloadPDF - 585.3 KB
DownloadPDF - 387.2 KB
DownloadPDF - 557.3 KB
DownloadPDF - 553.2 KB
DownloadPDF - 469.9 KB
DownloadPDF - 855.5 KB
DownloadPDF - 577.8 KB
DownloadPDF - 679.3 KB
DownloadPDF - 556.5 KB
DownloadPDF - 729.8 KB
DownloadPDF - 611.9 KB
DownloadPDF - 566.5 KB
DownloadPDF - 672.4 KB
DownloadPDF - 605.4 KB
DownloadPDF - 656.9 KB