Many studies regarding the face detection, can categorize people biasedly according to their characteristics such as skin color and gender. MIT is working to improve this problem with a better algorithm, rather than solving it manually.


MIT has been able to achieve reducing the categorical bias problems by 60 percent, with its new algorithm developed on the face detection. However, efficient results can be achieved in a shorter time on a larger data set. These improvements became important because the face detection systems became widespread and its reliability ratio is increased in many organizations and companies. Failure of solving these problems can also cause major problems such as difficulties in using the face detection products, producing false results for people, and making wrong decisions.

Artificial intelligence, developed by MIT’s CSAIL researchers, is improving with bias-free educational materials.

Face detection technology has entered our lives in a short period of time, however,  it is claimed that the systems approach with a racial bias in some cases. The face detection system developed by Amazon, which was started to use by the US Immigration and Customs Enforcement in recent months, has caused controversies. Senate members are questioning how Amazon is testing the system and the possibility of creating a racial bias. The improvement algorithm, that MIT is working on, could be the solution movement for this technology, which will become more widespread.