AI-based and inclusive recruitment
AI systems are widely used in recruitment and have great potential to streamline processes. However, because AI is trained on historical data, the systems tend to reinforce pre-existing biases, which risks leading to unintentional discrimination. We want to counteract this.
The aim of the project, funded by the Norwegian Directorate for Children, Youth and Family Affairs (BufDir), is to raise awareness and knowledge among companies and organisations about how to recruit people with disabilities and avoid the risk of unintentional discrimination associated with AI-based recruitment.
By identifying and highlighting the risks, and by working with both employers and employees, we will develop concrete recommendations to employers on how to ensure that their recruitment processes are inclusive and accessible.
The project will result in a guide on how to make AI-based recruitment processes inclusive and a checklist to increase accessibility in digitalised recruitment.
This is a super important area to work with, we need to increase knowledge so that it does not fail and exclude people, says Malin Hammarberg, Senior UX designer at Funka Foundation. I look forward to spreading knowledge about both the problems and how we can solve them.
The background
AI systems are rarely trained to take individual needs into account, which means that candidates with accessibility needs may have difficulty handling different types of tests. People who use assistive technology may also be disadvantaged as the system may incorrectly analyse the use of the technology.
Many employers lack insight into how AI systems work in practice and the implications of their use for persons with disabilities. A key shortcoming is that employers often rely on AI systems to be ‘neutral’, without realising that these models are based on historical data that may contain biases. This can lead to the reinforcement of pre-existing patterns of exclusion, making it even more difficult for underrepresented groups to enter the labour market.
Employers also need to better understand the barriers that AI systems can create in practice. These can include inaccessible test formats that exclude candidates with visual, hearing or cognitive disabilities, or systems’ interpretation of body language, tone of voice or linguistic abilities that lead to misjudgements. Many do not realise that these factors can affect the fairness of the selection process, and they do not have the knowledge to identify and address the problems.
Contact
Sara Kjellstrand
sara.kjellstrand@stiftelsenfunka.org