The first thing many people think about when they hear the word ‘algorithm’ is their feeds on popular social media apps such as TikTok or Facebook. The order of content is often discussed as the ‘algorithm’.
What is an algorithm?
An algorithm is a well-defined set of step-by-step instructions or rules that are designed to perform a specific task or solve a specific problem. This provides a systematic approach to problem solving, enabling either computers or humans to achieve the desired outcome via a series of logical and precise operations.
Algorithms have become integral to modern life through their heavy use in the online world we interact with daily. Algorithms drive online modern activity. Social media feeds are curated by an algorithm, and recommendations of music, films to watch, and items to purchase are recommended by an algorithm. Even online banking uses algorithms to detect transaction processing and fraud detection.
Are algorithms harmful?
At its core, an algorithm is a mathematical equation. They become harmful when they are programmed maliciously or when they learn harmful content.
Algorithms can adapt and ‘learn’ an individual or a group of users’ habits through a process called machine learning. This works by an algorithm adapting by analysing data patterns. It will repeatedly process all data inputted and adjust data further based on the outcome vs the predicted outcome.
How do algorithms work?
There are several ways that algorithms work. The first is a technique called supervised learning where the algorithm learns from labelled data. The second is called unsupervised learning where the algorithm learns from finding patterns without labels. The third is called reinforcement which learns by rewarding desired behaviours and/ or punishing undesired ones. It will be able to perceive and interpret its environment by taking actions and learning through trial and error.
In simpler terms, if you shopped online for clothes on the same website and always looked at grey hoodies, the algorithm would learn that you are interested in grey hoodies and would show you recommendations for grey hoodies. If you then looked at blue hoodies on the same website one day, the algorithm would learn that you might also like blue hoodies and may recommend some of those depending on the frequency of you looking at them.
This can be helpful, as it provides recommendations that you may actually enjoy, such as new content to watch on video streaming services or new music on music streaming platforms but it can also be harmful.
For this to be effective a lot of data is required about you, which will be collected and retained by the service. You may consider this to be harmless data, however, it can reveal a significant amount of information about you, what your interests are and your use of the online world.
How are algorithms supporting business decisions?
Companies may also use your data as well as other user’s data to make business decisions. An example of this is Netflix, which uses the data its algorithm has learnt to pick and create new content that will be of interest to the largest selection of its users.
Another way the algorithms can be harmful is if they are designed to learn from other people’s data as well. This can allow misinformation or harmful content to spread if it becomes popular.
The less considered issue is what is not seen. Content may be hidden if it is not deemed interesting to a user. This means that contrasting views and different opinions may not be seen and could be hidden, resulting in the user being caught in an echo chamber. This can allow for unbalanced opinions to be shared and viewed by people. This can cause influence and change in many areas, such as politics.
What can I do to avoid them?
While algorithms are unavoidable in modern computing and online activities, it is possible to prevent them from skewing your actions.
The first is always to view news and information from multiple sources in separate places. If you read something on social media that appears to have strong opinions or suggest controversial theories, it is always worth further researching alternative sources and trusted sources.
The second is to only use trusted services online. Software vendors can process the data you input how they wish including for training algorithms. It is important to understand the services that you use and the data that you provide to them.
To conclude, algorithms are an integral part of everyday life and online activity. They provide many useful benefits to industry and individuals and the internet would be unrecognisable without them. However, the importance of finding balanced opinions in order to form your own has never been more important.