“Algorithms are opinions embedded in code.”– Cathy O’Neil
If Instagram were a courtroom, its defendant would not be a human but an algorithm- mysterious, complex, and powerful. It determines what we look at, whom we follow, and even what we care about. It favors some aesthetics, penalizes others, and in the process, silently frames the rules of our online culture. But for something so powerful, the algorithm has no face, no voice, and decidedly no accountability. The question that resonates in this metaphorical court is straightforward: are algorithms guilty of bias?
Instagram was envisioned as a place of expression, a photo diary where imagination could thrive without hindrance. But it has evolved into a machine designed for one purpose: engagement. Each scroll, like, share, and pause is data. That information powers prediction and those predictions decide what’s on your screen next. It’s a never ending loop, until you realize that the patterns it learns from aren’t neutral to begin with. They are representations of the biases, inequalities, and stereotypes which are already present in the world.
Algorithms aren’t as neutral as they appear. They’re built by humans, trained on human data, and shaped by human behavior, so they just pick up our biases. When an algorithm sees that traditionally good-looking faces or Western standards of beauty receive more likes, it assumes that is what users like. It then promotes such content even more, confirming sameness and eliminating diversity. What starts out as a mathematical equation becomes a mirror that highlights society’s imperfections. The algorithm doesn’t create bias; it refines it.
This bias doesn’t merely appear in what is visible, but also in what is muffled. Shadow banning- where posts are quietly hidden without notice, has become one of the most hotly debated results of algorithmic systems. Creators tend to notice that their posts no longer show up in feeds or searches, particularly when they’re posting about sensitive subjects like Race, Gender, or Politics. Instagram officially denies any direct censorship, yet many users notice their reach drop sharply after posting about issues that challenge the mainstream narrative. A traditional ban is a clear cut prohibition, whereas shadow banning offers no notice, no reason, and no opportunity to appeal and question. You are not taken away; you are made invisible.
The kind of silence on the internet raises questions about fairness and free expression. If an algorithm decides that your content is ‘borderline’ or ‘sensitive,’ it might quietly push it down in the name of ‘user safety.’ But safety for whom? For those who choose comfort over confrontation? For advertisers who avoid controversy? When technology is built to connect people starts silencing certain voices, it stops being a tool and becomes a gatekeeper of conversation.
On the other side of this courtroom is another sort of bias: amplification. The algorithm favors content that gets engagement, and the simplest way to get engagement is by being at an extreme. Outrage, shock, beauty, and controversy are what make things viral. A calm explanation or thoughtful discussion stands no chance. Consequently, posts that stimulate emotional responses spread more than those that educate or challenge creatively. The algorithm isn’t interested in truth or ethics, it’s interested in performance. It learns that getting angry earns comments, that being perfect earns likes, that hyperbole earns views and it responds accordingly.
This builds a feedback loop in which creators confirm to satisfy the algorithm instead of being true to their own self. Being seen turns into a game of optimization with the optimal filters, the optimal hashtags, and the optimal timing. In this system, creativity is weighed down by numbers. Authenticity itself gets curated. And although the users think they are in control of their feeds, it is actually the algorithm that is driving what they think they are choosing.
Fairness, then, becomes a complicated term. Can a system based on human behavior ever be fair when human behavior itself is biased? Developers can say that their algorithms are neutral, but neutrality translates to fitting within current power dynamics. If the data from previous user interaction reveals that lighter skin or Western clothing gets more attention, the algorithmic predictions will naturally favor those trends. Therefore, the “popular” or “trending” could in fact be the online representation of social inequality.
Even when Instagram tries to fix that with content moderation, fairness audits, or technologies that feature chronological feeds, the very effort sounds superficial. The underlying architecture remains the same: exposure is tied to engagement, engagement is tied to that which engages emotion, not that which enables fairness. The company’s attempt to make the system fairer exposes a deeper truth, fairness can’t be programmed into something built on unfair foundations. But to blame the algorithm solely would be too simplistic. As consumers, we are the culprits. Each time we like a photo, skip a story, or pause on a reel, we feed the algorithm’s idea of what succeeds. Without realizing it, we keep strengthening the patterns we already prefer. We are creating the algorithm as much as the algorithm is creating us. Lines of responsibility get foggy when we consider that we ourselves fall under the data-based bias. If Instagram perpetuates a strict definition of beauty or success, it is because we, as users, reinforced the definition in its entirety.
The ethical value of that insight is profound. If we are training algorithms, then reconfiguring them means reconfiguring us, our decisions, our ways of using what we purchase, our desire to be entertained by dissimilarity, not comfort. The real problem isn’t simply that Instagram’s algorithm is biased, it’s that we’ve managed to turn bias into something entertaining, profitable, and perfectly optimized for engagement. Yet even with such a reality, there’s still reason to hope. Researchers and ethicists have been calling for greater algorithmic transparency, giving users the right to understand how their content is ranked, promoted, or flagged. Some suggest independent audits of social media algorithms, much like financial audits, to hold platforms accountable. Others advocate for community-based oversight, where diverse groups of users have a say in shaping moderation policies.
All these proposals reflect a growing understanding that fairness can’t be left entirely to machines. Justice, whether online or offline, needs human judgment, empathy, and responsibility. Instagram has made attempts at reform, experimenting with new feed options and publicly promoting its belief in “creator equity.” But these steps remain voluntary, driven more by public pressure than by any binding regulation. As things stand, the fairness of the algorithm depends largely on the company’s conscience, a fragile base for something that influences the lives of billions.
So, if we return to our courtroom metaphor, if Instagram’s algorithm were truly standing trial, the verdict might read: guilty, but with context. Guilty of perpetuating bias, of rewarding inequality, of silencing some voices while amplifying others. But also guilty because it learned from us, because it reflects a culture that measures worth in attention, not truth. The algorithm is not a villain; it is a mirror. And the most chilling reality of all, perhaps, is that the biases we condemn the algorithm for are, in the end, our own biases in the mirror. The problem isn’t so much that the algorithm of Instagram can ever be fair, as whether we as a society, as people, as users, can.
