Get the App
SLTechnology News&Howtos  ›  IT Information  › 

Research shows that self-driving software is "biased" and people of color and children face a higher risk of collision

Shulou Source: shulou.com Published: 2023-11-24 19:02:44 11月17日 Update

CTOnews.com, August 27, the development of artificial intelligence (AI) has brought many conveniences to human beings, but it has also caused some serious problems. Recently, a new study found that pedestrian detection software in self-driving cars may have different recognition effects on pedestrians of different skin colors, genders and ages, leading to higher safety risks for people of color and children, Businessinsider reported.

The study, conducted by research teams in the UK and China, tested the accuracy of eight popular pedestrian detection software in different situations. The results showed that gender had little effect on the recognition effect, but there were significant differences in skin color and age. Dark skin and children were 7.52% and 19.67% less likely to be detected than light skin and adults, respectively.

The research team believes that this bias is due to biases in the open source AI systems used by many companies. These open source AI systems are trained based on large amounts of data, but these data do not necessarily represent diversity in the real world. Therefore, when these AI systems are used in self-driving cars, there may be unfairness or discrimination against certain groups.

The research team called on lawmakers to regulate self-driving car software to prevent bias in its testing system.

CTOnews.com with the address of the research paper: click here direct.

Tags: Research systems skin color detection bias software driving differences teams cars pedestrians children age gender situation effects data people of color race risk Apple Docker Huawei Linux macOS MariaDB Microsoft MySQL NVidia OPPO Reno macOS vpn NVidia Xiaomi Huawei