How to Reduce Bias in AI Systems
Tue, April 20, 2021

How to Reduce Bias in AI Systems

Sofia addresses AI bias by making sure that their data are clean, unbiased, and completely diversified / Photo Credit: Gustavo Frazao via 123RF

 

Artificial intelligence tools are widely used now in hiring as many companies believe that they are free of bias. But it’s not entirely true. While AI tools can sometimes be bias-free, it is still prone to unintentional bias because the systems are loaded with human information. This should be addressed as soon as possible because when AI makes a mistake, it can be devastating. For instance, Microsoft unveiled an AI chatbot in 2016 that turned from being a jolly bot to advocating the destruction of feminists and Jewish people in just 24 hours. 

Thus, many companies are attempting to eliminate bias in AI systems. Sofia, one of the leading information technology service providers with a broad spectrum of services that help clients closely align IT with their business objectives, deals with bias. Sony SungChu, the head of the applied data science team and creator of Sofia, stated that to address this problem, one has to guide AI’s learning from scratch. This requires creating stringent conditions and boundaries within which AI can learn. 

According to HR Daily Advisor, an online site that offers free webcasts, articles, and reports on topics important to HR and compensation professionals, companies should also safeguard sensitive topics such as private health information, sexual harassment records, and more to prevent bias from entering the AI systems. Companies should not limit their standards to people who come from a certain socioeconomic class. This will alienate quality talent and lose out on a diversity of thought. “So as a community, we’re still trying to strive to improve those safeguards,” SungChu said. 

He also compared teaching AI to teaching a child. “When you’re teaching a human child, you expose them to different ideas. Let them experience the world a bit; they intake ideas either with your guidance or without your guidance, right? And then they synthesize that information so that they can get cues to think,” SungChu said. Thus, teaching AI the same bias humans have will not be good. To address this, companies should make sure that the data is as clean as it possibly can be, unbiased, and completely diversified.