It
was interesting to read the article which developed on the existence of
so-called algorithmic bias in the computing world. Numerous studies by different researchers were cited which confirm the existence of a bias in decision-making capabilities of artificially intelligent (AI) computing systems. It works in a closed-loop feedback system wherein the application data, as it grows, trains the logic of machine learning and deep learning algorithms. The growing data, especially in the case of social media, reflects mass mentality and inherent biases of the society, be it related to race, skin color, gender, political views or religion.
The reflection of apparent trends based on bulk data which gets generated with millions of posts on social media platforms does not necessarily portray the myriad voices and opinions which help in giving our society a balanced character. The suggestions, filters, and recommendations on such platforms affect the independent choice-making capabilities of individuals. There have been vigorous stirrings in different parts of the world vowing to preserve the neutral character of the internet, and there have been instances when social media giants have been stopped from tinkering with the neutral order in place.
There is a cue in place for programmers trying to leverage the capabilities of machine learning algorithms, that the data-sets they use to train their models are not free of biases because the source of that data is far from perfect. Efforts and means need to be put in place which could indeed help in minimizing it.
