The field of Human Resources is not immune to the allure of automation and with online job portals exponentially increasing the number of applications, ATS have become standard. When these systems integrate AI, they’re able to parse down text into structured data, analyse career progression, and provide a numerical match score to the job listing to separate successful candidates from a large pool. While efficient for HR Departments, this seemingly objective approach can perpetuate biases against female applicants. The systems seeking matches are reliant on profiles of previously successful candidates, underlying base model training and semantic analysis, all of which contain inherent biases.

The example career trajectories of successful tech candidates would be historically patterned after men. Even with the removal of gender identifying CV components like names or pronouns, systems that work with patterns often identify proxy variables which can carry through biases that align “successful” candidates with profiles of those who previously held the same positions. When extracurriculars, networking groups, or even certain educational institutions are included, they may carry gender implications. Months-long gaps in a candidate's work experience are often due to maternity leave or elder care and therefore the trajectory of these applicants' careers will not map to the typical male career patterns dominating the training data. Even the difference in the way men and women typically write about their achievements creates gendered silos based on the analysis of linguistic patterns.  

The EU classifies these systems as high risk and requires bias auditing, transparency and human oversight. With entry level jobs disappearing and decision-making becoming increasingly concentrated in the field, hiring practices that contribute to fewer women in tech despite educational trends need to be systematically corrected. Individually, women can proactively neutralise gender indicative credentials, anchor accomplishments with data, use impact driven language, and contextualize career breaks.