The future of hiring is shifting to accomodate for the advancements in AI technology, which are increasingly being embraced by businesses worldwide. The global landscape of generative AI in the HR market is predominantly led by the recruiting and hiring segment, comprising 28% market share. However, with this increase in adoption,AI bias has become a rising concern, discriminatory data and algorithms are built within these systems, therefore presenting adverse consequences for organisations.
AI bias refers to AI systems that generate biased results that reflect and sustain societal biases, including historical and current social disparities. Bias derives from initial training data, where, typically a Developer will input a dataset when setting up an AI model or algorithm, with large amounts of data to help it learn patterns and generate outcomes aligning to business needs.
Applicant screening is a common use case for AI in recruitment, historical hiring data from the company may be used when building an AI system to support this process. This historical data may have favoured male candidates over female candidates for particular roles, and therefore can present skewed results that inadvertently prioritise male candidates over equally qualified female candidates. Consequently, this may lead to uninformed hiring decisions, due to the limited pool of talent, accessible to hiring managers. For example, Amazon discontinued the use of a hiring algorithm, after discovering that it favoured applicants based on words like “executed” or “captured,” which were more commonly found on men’s resumes. In recent studies, Shortlist.net reports on a UK-led research, revealing an increase in the proportion of masculine words as the salary or title seniority of a role increased, while expressions of EDI and flexible working arrangements dropped. Job ads for senior roles exhibit traits typically associated with men, potentially reinforcing existing stereotypes that discourage women from applying to such positions.
Similarly, companies often use AI-driven job advertising boards such as Seekout which analyses user data e.g. demographics, skills, preferences, and online behaviour. Employers are then presented with talent that matches their organisational requirements. However, if businesses fail to diligently select unbiased recruitment tools, they risk falling victim to the negative effects of AI bias. Resulting in a tarnished employer brand and difficulties in attracting and retaining top-talent, as job-seekers will gravitate towards employers who share the same values and prioritise fairness throughout the hiring process.Therefore, employers are encouraged to ascertain the data source and quality, whether it be their own or from an external vendor, ask questions prompting the disclosure of AI tools employed and monitor the performance of target job advertisements. Is it capturing relevant candidates that align with your requirements? Ultimately, this improves inclusivity and diversity in the talent pool, resulting in higher quality talent.
To achieve fair results and mitigate bias, prioritising AI governance is imperative. Organisations should refer to Australia’s AI Ethics principles when implementing AI strategies into their operations and ensure compliance with data privacy rules regarding generation and storage of significant personal data. Additionally, human intervention is paramount when leveraging AI to automate scheduling processes, streamline candidate screening and resume scanning, as AI may lack a nuanced understanding of human behaviour. Therefore, with human intervention fostering fair and personalised interaction throughout the hiring process, hiring managers can gain a better understanding of individual circumstances that influence their decision-making. Upholding integrity, accountability, inclusivity and trust among both candidates and stakeholders, reinforces the importance of AI Ethics principles, while highlighting the potential of AI to bring significant positive changes in the recruitment process.
Share this blog