The article addresses the significant problem of algorithmic bias, asserting that algorithms are indeed biased and can produce incorrect or unjust outcomes. It presents various examples, such as Google's algorithm which historically displayed images of a couple of color when searching for 'gorillas', and Applecard's algorithm which allegedly offered higher credit limits to men over women. The article emphasizes the argument that, rather than being objective and fair, these algorithms often embed and amplify existing biases against non-majority groups.

The text explains that there are several reasons for this bias in algorithmic design including poor algorithmic design practices, the use of discriminatory criteria that involve direct or indirect factors, and training algorithms based on historical statistical data which itself may incorporate bias. This creates a cycle of reinforcement where existing prejudices become part of the decision-making processes governed by algorithms. The implications of this are far-reaching, impacting how technologies affect various demographics and potentially entrenching social inequalities.

Furthermore, it raises questions about the European legal framework aimed at regulating these algorithms and ensuring fairness in their application. The ongoing discussions in Europe regarding the establishment of laws to combat algorithmic discrimination highlight the urgency for accountability and transparency in algorithmic systems to protect vulnerable populations from potential harm created by these biased designs.