The Duality of Algorithmic Discrimination
--
Algorithms are a term that I was first introduced to in a calculus class and it was defined as a process or set of rules to be followed in calculations or other problem-solving operations. Now, as technology developed engineers started using algorithms for computers to perform tasks that gradually grew more complicated.
In today’s highly digital age algorithms have now become an intrinsic part of our technology and digital space. It now goes beyond the menial tasks of past algorithms and has embraced machine learning.
Machine learning is basically a process in the case of algorithms where it starts recognizing patterns of behavior and through association learns the user's online persona to help give information useful to said user. Algorithms are embedded in applications we use every day such as Netflix (or other streaming services), Spotify, Google, and social media platforms. Algorithms are now filtering the information we get and it has both positive and negative impacts on our society.
For entertainment, it can be said that it has helped a lot with the fast-paced short attention of our society. Before the onset of streaming services, people had to go to their local Blockbuster in order to find a film to rent. This process also included walking and looking through all the different films in the store even though they were not necessarily the movie that they were looking for. This meant that people were exposed to genres and films beyond their knowledge.
With streaming services such as Netflix and Spotify, the algorithm pays attention to what the user streams and slowly curates a selection that is specifically tailored just for the user.
On Netflix, this can be seen all over the platform. The possibility to ‘like’ and ‘dislike’ shows, showing the percentage that a certain movie/show matches your interests and even creating your own list of films you like. All this information is processed by the algorithm to create a homepage that seems to have an endless list of films and movies that you would like to watch. This also happens on Spotify as it creates playlists that it believes you would enjoy and recommend new artists in that genre all of which accumulates at the end of the year where they show you a ‘wrapped’ where they show exactly how many times you streamed certain songs, top genres, artists, etc. These are caused by algorithmic filtering. The issue that can arise from this is that people are only exposed to the things they like and sometimes entire genres can be hidden from them.
It causes people to end up with completely different virtual realities from each other.
The entertainment streaming service algorithmic filtering may seem less of an issue; however, when this translates onto search engines, social media, and news sources that’s when it can have a deadly polarizing effect in society. The United States is a perfect case study for this phenomenon of algorithmic discrimination manifesting into our reality. These platforms, especially Google and Facebook, have been under a lot of media coverage recently regarding their data harvesting and algorithmic discrimination. These platforms notice what posts the user likes, who they follow and what they search and start creating an echo chamber of information that confirms this user’s views. This can be detrimental when this information includes false information that can quickly become viral with a quick press of the share button. Some claim that these platforms censor women’s bodies that do not fit the ideal type. Others claim that the platforms censor conservative posts and are solely for liberals. This has led to the creation of other platforms such as Parler that only cater to one viewpoint. In the 2016 elections, Facebook was questioned for pushing false information on its users who believed them.
The algorithm can quickly develop into a process that polarizes people without their conscious knowledge of it.
Algorithmic discrimination also manifests in other harmful ways. On Google when searching for certain terms it gives suggestions as you type. These suggestions can reinforce the negative connotations that society has with these terms. For example, when searching for a doctor or engineer the images provided are predominantly white men and a few women. In contrast, when searching for criminals the images are saturated with men of color. There is an ongoing debate whether it is the algorithm's fault or that of society that perpetuates these harmful stereotypes through the continuous search for these terms.
The truth is that the debate over the problem of algorithmic discrimination does not have a black and white answer but is a reflection of both the flaws of our technology and our society.
Algorithms can help us by making life easier but it also can damage our society to the point of extreme polarization if it goes unchecked. One way to address this issue may be to have a diverse development team that can create an algorithm that is less biased.
However, I do not believe it is possible to be completely unbiased as long as humans influence machine learning, algorithms, and AI there will always be an influence of natural human bias.
