Outnumbered

Are moral beliefs still our own if algorithms shape them?

This year, for the first time in the internet’s history, more of the web’s traffic came from machines than from people. Cloudflare, a company hosting roughly a fifth of the internet’s websites, reported that bots account for roughly 57.5% of all requests to web content. Its own CEO had predicted that the crossover wouldn’t arrive until 2027, but began walking the estimate closer as use of agentic AI began to explode. The threshold quietly crept closer, receiving little to no media coverage or concern. The shift in internet usage raises the question: If the medium through which a growing share of humanity now encounters ideas and moral claims is one in which we are no longer majority participants, how do we form convictions inside a conversation in which we are increasingly outnumbered?

The question may have teeth; however, it assumes a tidy boundary between beliefs we author and beliefs we absorb, a boundary that was never more than a mirage. Long before the existence of the recommendation engine, our convictions did not spring fully formed from autonomous reason. We inherit them from our environment: parents, scriptures, and the accident of where we were born. The Athenian who thought slavery to be natural and the Quaker who found it monstrous were shaped by forces they did not choose. They did not have control of the morals held by their society, nor did they have control over the morals held by their religions. So long as man is not separable from his environment, his moral convictions remain inseparable from his conditioning. If external influence dissolves ownership, no one has ever owned a belief, and the question answers itself trivially.

What makes this question interesting is comparative: how does algorithmic shaping differ in form from other kinds of influence we’ve tolerated? I think there are a few distinct differences.

The first and foremost is opacity. When my grandmother shaped my views, I could see her doing it. I could weigh her views, authoritative as they were, against my own and choose to rebel if I wished. The biggest problem with algorithms is not that they curate what you see, but that they curate what you don’t see. An algorithm that hides one side of a moral argument from you severely restricts your ability to weigh your beliefs against anything. In the simplest terms, you cannot resist an argument you were never shown.

The second is optimization towards ends that don’t serve our interests. A parent influencing your morals, whether reinforcing corrupt societal narratives or not, often has your best interests in mind. An algorithm shaped for your engagement often thrives off fear, outrage, or controversy. It does not care about the resulting morals you may develop; it is millions of lines of code designed to keep you engaged.

It’s important to now take a step back and revisit the initial reason behind this questioning, the fact that bots now constitute the majority of internet traffic. While this seems troubling even at first, the sophisticated reading may be more unsettling than the first one. Most of that internet traffic isn’t propaganda lying around forums or AI-written news articles; it’s crawlers and agents fetching pages for real people. By Cloudflare’s own count, a person researching a purchase might visit five websites, while an agent completing the same task visits five thousand. The issue isn’t that bots are persuading us, it’s that a layer of non-human mediation sits between you and every idea you meet, deciding what gets surfaced before any human judgment enters.

So, to give a definitive answer to whether or not our beliefs remain ours is seemingly impossible. On the one hand, it is undeniable that the filtration of ideas that has intensified as a result of our reliance on AI is clouding our ability to judge things for ourselves. However, it is also true that the things we learn to believe are still coming from the avenues which we would otherwise receive them from. The internet forums resemble the Ecclesia of ancient Greece, and the religious gatherings of the Quakers; the difference now is that Artificial Intelligence guides you to which system you agree with. In a prior section of this essay, I made the claim that the randomness of your birthplace contributes to the randomness of what you believe, but what AI does is take away this randomness. It shifts you to where it wants you, and in doing so is shifting your beliefs in some way. So to answer the question, while AI does have a strong effect on our beliefs, it is not in any way total. Nevertheless, I don’t think the correct conclusion to make from this is that we do not have control over our ideas. In fact, ownership of a belief was never about the purity of its origin; it was about your relationship to the belief once you adopted it. It is not impossible to examine, defend, revise, or even be held responsible for acting on a belief influenced by an algorithm; however, it is much harder to do so when what influences your belief is only as tangible as the data center that holds it. That question is where the moral complexity lives.