When you do a search on Google, scroll through your social media feed, or even shop online, the information you see isn’t random. It’s selected by algorithms. That is automated systems designed to decide what content you see and in what order.
These systems can be incredibly helpful, saving time and tailoring results to your preferences. But they’re not perfect. In fact, algorithms and the data they use can carry hidden biases that affect what we see, how we think, and even the decisions we make.
One of the favorite classes I teach is Ethics and Technology. The biases we all have can easily spread online. Some of relatively inane… like which baseball team is the best, or who the GOAT basketball player is… other’s can cause major issues effecting people’s lives, freedoms, and more.
In this section, we’ll explore what algorithmic and data bias is, why it matters, and how to spot it in everyday technology use.
What Is an Algorithm?
An algorithm is just a set of rules or steps a computer follows to solve a problem or perform a task.
As a Computer Science professor, I talk a lot about algorithms in my classes. However, you don’t have to know what an algorithm is at that level, to understand how algorithms control what you see, and who you see, online.
Consider a search you might do: “best laptop for students.” An algorithm decides which results to show you based on things like:
- Your location
- Past searches
- What other users clicked on
- Page popularity
- Paid ads
- Who, and how many people, have linked to a page/site
- Site authority
- and much, much more…
That algorithm is designed to be efficient and helpful, but it’s not neutral. It makes decisions based on patterns in data, and those patterns come from people.
If you follow people on social media, you might notice certain people showing up in your feeds over and over, while you rarely see others. This is because social media is designed to show you people you interact with more. This is true on sites like YouTube, Facebook, Instagram, and well… all of them.
This unfortunately means you might see content from the “loudest” social media users and websites, but not necessarily the best, as the search engine can’t necessarily determine the “best” on it’s own. It needs feedback from users… And people are not neutral.
Where Bias Comes From
Bias in algorithms usually doesn’t come from bad intentions. Most of the time, it comes from one of these sources:
1. Biased Data
If you train an algorithm using biased or incomplete data, the algorithm will reflect those flaws. For example, if a hiring tool is trained using resumes from mostly male applicants, it might learn to favor male candidates, even if that wasn’t the goal. This has actually happened – https://www.theregister.com/2020/12/08/texas_compsci_phd_ai/ .
2. Human Assumptions
Every algorithm is designed by humans. That means the decisions about what to include, what to ignore, and how to rank or weigh information all come from human judgment.
For example, someone has to decide what counts as “relevant” in a search engine. Do they value popularity? Recency? Location? Expert opinion? These choices affect what results you see.
3. Feedback Loops
Algorithms learn from user behavior. If a lot of people click on a particular video or news article, the algorithm assumes it’s high quality and shows it to more people. That can create a cycle: popular content gets more exposure and becomes even more popular, even if it’s not accurate or fair.
This happens a lot with social media and video platforms. Controversial or extreme content often gets more clicks, so the algorithm promotes it, even if it spreads misinformation or reinforces stereotypes.
YouTube was even known for weighting dislikes the same as likes, because it showed user engagement, and what should have been a sign to not promote a video, instead caused some videos to get even more popular.
Examples of Algorithmic Bias
Let’s take a look at some real-world examples to show how this works.
Example 1: Job Listings
A few years ago, researchers found that job ads for high-paying tech jobs were being shown more often to men than to women, even though the women had similar qualifications and search histories. The algorithm had learned (from biased data) that men were more likely to click on those ads, so it showed them the jobs more often, reinforcing inequality without anyone intending to.
Example 2: Facial Recognition
Several studies have shown that facial recognition systems are much more accurate for white male faces than for people with darker skin or female features. Why? Because many of these systems were trained on datasets that mostly included white male faces. The result is a tool that works well for some people and poorly, or even dangerously, for others.
Example 3: Social Media Feeds
Your TikTok or Instagram feed doesn’t show you every post from every account you follow. Instead, an algorithm decides what it thinks you’ll like based on past behavior. But this can create a “filter bubble”, a situation where you only see content that confirms your interests or beliefs and miss out on diverse perspectives or new ideas.
Why This Matters to You
You may be thinking, “OK, but I’m not designing algorithms. Why do I need to know this?” Good question.
Even if you’re not a programmer, you use systems driven by algorithms every day. If you don’t understand how those systems work, or how they can be biased, you’re more likely to:
- Be misled by search results or social media content
- Miss out on important opportunities or information
- Share or believe false or one-sided stories
- Make decisions based on flawed data
In school, at work, and in everyday life, being aware of bias helps you stay informed and think critically about what you’re seeing and sharing.
How to Spot Algorithmic Bias
Here are a few questions to ask when using technology that gives you filtered or personalized information:
1. What’s Missing?
Are you seeing just one side of a topic? Try searching for the opposite perspective and see what comes up. This is especially important for news, political issues, or controversial topics.
2. Who Benefits?
Ask yourself who might benefit from the way the content is ranked or presented. Is it a company trying to sell something? A political group trying to promote a message? A social media platform trying to keep you engaged?
3. Is It Repeating My Own Views?
If your news feed always agrees with you, that might feel good, but it’s not necessarily healthy. It could mean the algorithm is trapping you in a bubble where you’re only hearing opinions you already believe.
4. Are Certain Groups Left Out?
Look at the images, voices, or stories being shown. Are people from all backgrounds represented? Or does it seem like the content is mostly designed for one kind of person? This can help you spot underlying assumptions or gaps in representation.
What You Can Do
You can’t control how every algorithm works, but you can take steps to be more aware and intentional about your digital life.
Use Diverse Sources
Don’t rely on a single app, website, or platform for your information. Look for multiple sources, especially ones with different viewpoints. This helps you balance out any single algorithm’s influence.
Clear Your History Occasionally
Your past behavior influences your future results. Clearing your search history or using a private/incognito browser can help reduce personalized bias, especially when doing research or making important decisions.
Seek Out Counterpoints
Actively search for different perspectives on important topics. This not only improves your understanding but also keeps you from being manipulated by narrow viewpoints.
Talk About It
The more people understand algorithmic bias, the more we can push for transparency, fairness, and ethical design in the tech world. Whether you’re in a classroom discussion or chatting with friends, bringing these ideas into the conversation makes a difference.
Looking Ahead
As more decisions in life, from college admissions to loan approvals to dating matches, are influenced by data and algorithms, understanding how bias sneaks into these systems will become more and more important. It’s not just a tech issue. It’s a fairness issue, a privacy issue, and a critical thinking issue.
You don’t need to be a coder to ask good questions, spot red flags, and make informed decisions.
That’s what it means to be a smart, ethical technology user.
Recognizing Biases in Algorithms and Data was originally found on Access 2 Learn
One Comment
Comments are closed.