The promise of a bias-free meritocracy thanks to artificial intelligence in selection processes is crumbling. A recent analysis reveals that platforms using video with AI to evaluate candidates reintroduce discrimination based on gender, race, or appearance by showing the face before skills. The supposed technological neutrality clashes with reality: algorithms replicate human prejudices.
Forced Anonymity: The Key Technical Solution Against Visual Bias 🛡️
The technical solution is to require these platforms to anonymize the video until advanced stages of the process. This involves applying facial blurring filters, replacing voices with neutral synthesis, and hiding physical markers such as clothing or background. Only after evaluating technical competencies through blind tests would visual identification be allowed. This prevents the image from conditioning hiring and perpetuating labor inequalities.
Blind Hiring, But Not So Much 😅
In other words, so that AI is not racist or classist, we have to ask it to look the other way. It's like being on a blind date with a bag over your head until you prove you can cook. In the end, technology promised to eliminate human bias and ended up needing us to teach it not to look. Ironies of progress: for the machine to be fair, we have to cover its eyes.