STEMM-CHANGE

An AI Communications Toolkit for Diverse Recruitment Practices in Multinational Workplaces

Following two years working on language bias in job advertisements and descriptions, researchers on the language strand have partnered with Diversely to inform development of an AI-driven Job Description Analyser. This tool, currently under development, embeds specialist knowledge on exclusionary language patterns, the severity of bias encoded and suggestions for more inclusive alternatives in the platform used by the organisation’s multinational clients. These suggestions encourage holistic communication strategies that view prospective employees as individuals with intersectional backgrounds, characteristics, life experiences and world views. This approach makes it easier to avoid wordings which unconsciously stereotype and perpetuate workforce imbalances and channels the ability of language to shape conceptualisations of appeal, fit and belonging to generate interest from a broader range of qualified applicants.

 

UoN’s Applied Linguistics IP has been instrumental to develop Diversely’s job as bias analyser. The tagged bias phrases data have been used to develop a contextual NLP library which identifies and highlights bias phrases in company’s job ads. Recommended alternative phrases speed up the time for our users to write a more inclusive job ads that appeal to broader and more diverse applicants by at least 50%. We look forward to continued collaboration with UoN (Louise and Jacqueline) in updating bias linguistic patterns and providing data driven feedback from our applicant pipeline to demonstrate the impact being made in hiring diverse talent across industries and the globe.

Hayley Bakker, Chief Product Officer at Diversely.io