10108856Inclusive Innovations: Transforming CV Screening for EquityClosedCollaborative R&DInnovate UK_"72% of CVs are never seen by human eyes. Computer programs flip through them, pulling out skills and experiences, scoring each one as a match for the job opening. The more candidates they eliminate with this first screening, the fewer human-hours they'll have to spend processing the top matches."_ --- _Cathay O'Neil, Weapons of Math Destruction (2007)_ This pioneering initiative targets the pressing issue of bias and discrimination within AI systems used in recruitment. The project won't just tackle algorithmic bias at a technical level; the socio-technical approach to developing a bias monitoring toolkit encompasses structural and human biases that can influence hiring decisions. Beyond research and technical development, this initiative features a dynamic programme of community engagement, workshops, knowledge-sharing activities, and conference presentations to drive awareness and adoption of best practices within the machine learning community. At the heart of the project is the development of an innovative open-source toolkit and service to detect and mitigate biases, both human and AI-driven, in recruitment processes. We will create two new datasets: a GDPR-compliant anonymised or synthetic CV dataset; and a dataset of UK-wide demographic benchmarks across various occupations. The primary focus will be to evaluate and reduce biases inherent in automated CV screening algorithms, which are prevalent in recruitment but prone to biased outcomes. The project will deliver two key outputs: a Minimum Viable Product (MVP) demonstrator and a comprehensive bias detection toolkit designed specifically for UK data protection and equality laws. The MVP will showcase our service's practical application and effectiveness in real-world settings, whilst the toolkit will provide practical tools and guidelines to identify and address bias. A standout feature of this solution is its capability to infer protected characteristics from anonymised CVs, providing insights into potential areas of unfair discrimination. The system will offer explainable insights on hiring decisions, shedding light on the often opaque 'black box' of AI decision-making in recruitment. The open-source toolkit and associated service are designed to benefit a wide user base, including Heads of HR in large corporations, law firms, compliance officers, and developers seeking tools compliant with current legislation. It will also be a valuable resource for auditing hiring processes, particularly in legal scenarios. Our goal is to encourage responsible and fair AI practices in recruitment, leading to a more equitable society. We are committed to regularly monitoring and evaluating our system to address ongoing biases and adapt to emerging challenges.