Algorithmic Bias: The Code that Shapes Our World

Alpha
Oct 2, 2023
4 min read

Updated: Oct 1, 2024

In the 21st century, a major part of the average citizen’s life is heavily influenced by algorithms. From online shopping and video consumption to social and political views. The virtual content we consume is served to us by a faceless and emotionless algorithm that can, at times, seem to know our tastes better than us. However algorithms are built and coded by people, and as such are known to carry their biases. In this article, we will explain the causes for said bias, explain its ramifications and consequences, while presenting some of the offered solutions.

ree — PHOTO CREDIT: by Markus Spiske from Pixabay

The Making of an Algorithm:

Algorithms are man-made. As such there is a process for creating these “machine learning” programs, which revolves around the exposition of the program to data. If an engineer wants to create an algorithm that can identify people, it first needs to feed said algorithm information about what people look like. Consequently, if said engineer only shows a single ethnic group, gender, or even physical build to said algorithm, it won’t recognize the others as human. As such, when creating an algorithm, its accuracy, method of identification, and decision-making will be based on the information that it has been exposed to. Meaning, that an algorithm can have a bias that results from its creating process.

Said bias can result from a lack of an original data pool or the biases of those that feed it information. As was explained by computer scientist and assistant professor at the University of Washington, Aylin Caliska [1]: “Many people think machines are not biased, but machines are trained on human data. And humans are biased.” [2]. Fundamentally, algorithms are the reflection of those who create them and in essence, of those people’s views of the world.

The Consequences:

Seeing as though algorithms are a facet of our everyday lives, all biases shown by these programs also have an impact on our daily lives. For instance, if the platform that you use to collect data is biased towards content that will resonate with your viewing habits, other meaningful content might be pushed out in favor of generating engagement. In particular, if a company uses a biased algorithm to seek and hire potential workers, it will favor a selection of individuals over another based on those biases [3]. Such was the case of an algorithm produced by Amazon, that sought to cut down the time behind the hiring process but was eventually scrapped [4].

Nevertheless, with the implementation of machine learning in increasingly more aspects of our societies, the cracks formed by these biases have begun to spread even further. May it be in smaller aspects, such as text-to-speech [5], or more fundamental parts of our lives, namely face recognition [6]. That is without mentioning the impact that these programs have in the distribution of online information and the effects that these integrated disparities in the presentation of information can have [7].

Solutions:

At the end of the day, the bias that is shown by these algorithms is either an error that results from its creation, a consequence of its creator's own biases, or it is purposefully designed. Thus, there are a few solutions that might help in eliminating said biases from these algorithms in their creation or navigating them. By having a diverse team behind the algorithm's creation, and validating all the information independently, the team behind the creation of these programs can reduce the likelihood of having their own biases bleeding into these programs. However, if the algorithm is biased by design, understanding these biases and calling out for transparency, while sharing your knowledge and warning others can go a long way.

Conclusion:

Machine learning programs such as AI and algorithms, are a cornerstone of the online environment. It is these programs that dictate in which way we should be exposed to information and as a consequence, they hold immense power over our daily lives, ideologies, and social activity. Since these issues have become more prominent, more and more calls for transparency have been made.

However viewing back the sociopolitical climate of the last few years, and the proliferation of biased, oftentimes outright false information, pushed forward by these programs, it is crystal clear that there is a long road ahead of us. As consumers, there is little that we can do when it comes to their development. Nonetheless, we can protect ourselves against the dangers of these biased programs, but learning how to navigate them and calling out for more impartial algorithms. These programs are a vital part of our modern society, and as such we deserve to get the very best. And it is within our right to be displeased when we don’t

Explore more with us! Join our Newsletter, support us on Patreon, listen to our Podcast, and discover our other captivating content. Together, let's shape a better future. #Politdigm