AI in Hiring: Risks of Bias and Inequality

The integration of artificial intelligence and facial analysis technology in job interviews is becoming more prevalent. This innovative approach, created by a US-based firm, assesses candidates by analyzing their vocal tone and facial expressions during video responses to interview questions.

Initially launched in the UK in September, this technology has been in use worldwide for several years, with companies like Vodafone, Hilton, and Urban Outfitters testing its effectiveness. While it offers benefits such as accelerating the hiring process by up to 90%, it also poses significant risks that could reinforce existing inequalities in the job market.

The AI system relies on algorithms that evaluate applicants against a vast database of approximately 25,000 data points related to facial expressions and language patterns, derived from interviews with individuals deemed "successful hires." This includes 350 linguistic features, such as tone and sentence structure, alongside various facial characteristics like eyebrow movements and smiles.

A critical concern raised by opponents of AI in hiring is that the technology is built upon a foundation of societal biases and discrimination. The datasets used to train these algorithms often reflect these biases. For example, searches for terms like "successful manager" typically yield images of middle-aged white men, while searches for "housekeeping" predominantly show women. This indicates that algorithms can learn and perpetuate harmful stereotypes, potentially reinforcing them in hiring practices.

Consequences for Equality

French sociologist Pierre Bourdieu emphasized that various forms of capital, including economic and cultural, contribute to the reproduction of inequalities. Elements such as upbringing, educational opportunities, and extracurricular involvement shape individuals' intellectual capabilities and self-esteem, ultimately influencing their confidence and prospects in life.

Erving Goffman, another influential sociologist, described this phenomenon as a "sense of one's place," where individuals from less privileged backgrounds may conform to their perceived roles. This is often reflected in their body language and communication styles during interviews. Candidates who exhibit higher confidence and better linguistic abilities possess what Bourdieu termed "symbolic capital," which can lead to greater success, regardless of whether these traits are genuinely relevant to the job at hand.

The use of AI in hiring could worsen this situation by favoring candidates who mirror the profiles of previous successful hires, thereby reinforcing traditional hiring patterns and potentially sidelining diverse applicants. The rigid nature of algorithms leaves little room for subjective evaluation or the consideration of unconventional talent, which could ultimately harm businesses by limiting their potential for innovation.

Wider Ramifications

This technology may unintentionally disadvantage skilled individuals who do not conform to the expected candidate profile, resulting in missed opportunities for innovative contributions. More alarmingly, it risks excluding individuals from diverse backgrounds, favoring those with greater economic and social capital who can develop the skills that are often prioritized in interviews.

In conclusion, the reliance on AI in hiring processes highlights broader issues inherent in technology. When developed using data from a biased society, AI is likely to reproduce those biases in its outcomes, perpetuating inequalities rather than mitigating them. Addressing these challenges is crucial to ensure a fairer and more inclusive job market.