2.1 Biases in Search Engines and AI Algorithms
If you use social media, you’re exposed to algorithmic bias every time you open an app. Algorithms drive most of our online interactions, and because they’re designed to be automated, we often assume that they’re a neutral force. This is not the case. Algorithms are created and used by people, and so they are inevitably influenced by the inherent biases that we all carry (Gillespie, 2016).
Platforms like Instagram and TikTok, for example, have been accused of suppressing content and shadow-banning creators, often in ways that disproportionately affect marginalized communities. These biases can prevent content featuring Black, Indigenous, and disabled people from reaching our feeds (Brown, 2021).
Examples of Algorithmic Bias from Social Media
Throughout this resource, we’ve intentionally included examples from social media. These platforms are embedded in our daily lives and can be powerful tools for sharing lived experience and perspectives.
The following TikTok videos offer two examples of algorithmic bias. As you watch, consider: Who and what has been missing from your own social media feeds? Whose voices are still missing from the videos we’ve included?
If the search algorithm were truly neutral, we could expect search results to show a variety of styles for different hair types, as well as a full spectrum of autism representations. But assumed norms of race and gender have been embedded into algorithms (Noble, 2018), so harmful biases are replicated and reinforced in our online environment. This remains the case even when some tools, like Bing in the above video, say:
I can’t create images that represent medical or neurological conditions like autism, especially when tide to a specific identity or appearance. That’s because autism isn’t something that looks a certain way—it’s a neurodevelopmental condition that can affect how people communicate, socialized, an experience the world, but it manifests uniquely in each individual. There’s no one visual representation of a quote autistic person and quote and reducing it to a single image risks reinforcing stereotypes.
We recognize that algorithms are constantly evolving and that platforms have taken steps to address bias in recent years (Google AI, 2025). As a result, the specific examples we’ve provided above may no longer fully apply today. However, algorithmic bias is ongoing, and it’s important that we remain attentive to how marginalized communities are being affected, even as the platforms and patterns of bias shift.
To learn more about algorithmic bias and its continuing impact, check out the scholarship of researchers like Dr. Safiya Umoja Noble: https://safiyaunoble.com/ 
.
The systematic biases that are embedded in algorithms.
The practice of partially blocking, limiting, downranking, or reducing the visibility of content without the awareness of creator or users.
