A Modern Day Plato’s Cave
“The machine does not give up its drive to conquer, but decides to start with a smaller and more feasible aim: acquiring a social media company and gaining profit by finally solving the problem of moderating content. It simulates that task too, only for the cycle of pain it endured to repeat once more.” – Adam Elkus, the New Atlantis.
The term “Artificial Intelligence” seems to be something that everyone is either terrified or fascinated by. The concept that it will replace human existence, taking over our jobs and our lives and our beings before merging into a divine singularity; or an exciting next step in society wherein we can reach our full potential by reducing our own workloads by assigning menial, repetitive labour to intelligent machines. So, what’s the reality? How does it look, move, breath – how does it shape what we know?
A.I is a widely misunderstood construct, both in how it works and operates. The wording of “Artificial Intelligence”, based on perceptions from popular culture such as film and television as well as the insinuation of the name itself, implies a sort of intelligent, autonomous being made from scratch and programmed our commands. The last part of it is more or less accurate, but only due to the models of which these systems and services operate; Large Language Models (LLMs) are machine learning systems, wherein a complex set of instructions are programmed so that the A.I. can determine answers based upon pre-existing information available on the internet.
There is an increase in panic due to the recent surge of Chatbots, such as ChatGPT, Co-Pilot and Grok, all of which appear to have some level of self-awareness and personality in their responses; however this tool also follows a similar pattern of machine learning, meaning it essentially acts as a hyper-specific and detailed search engine, more geared towards longer responses which themselves are summaries of the topic in question generated through the Chatbot’s extensive access to the world wide web and its resources. So no, Artificial Intelligence as it exists today is not a hyper intelligent, self-aware mechanism that plots to make mankind redundant – but we should still be scared.
If you’ve ever used a content-based platform, such as TikTok, YouTube, Twitter or Instagram, A.I. has been a present part of your life because these platforms use algorithms to determine what the user is and isn’t interested in, by paying attention to the user’s behaviour. Clicks, scrolls, time watched, what the user follows or subscribes to, likes, dislikes, saves – all of it is processed through an algorithm that determines what content to suggest the user on the site’s home page. All of this makes scrolling feel like a personalised experience, which makes the user more attached to the site in question. It’s the For You page; everything has been selected for you. It is, or at least should be a reflection of your interests, hobbies and those of other people you choose to follow, friend or add.
So, A.I. is already being used – big deal, this has been the status quo for some time. So then, why do things feel different on the internet then they did ten years ago? Have things really changed that much – if so, what changed them?
What we see and what we do is selected and presented via AI algorithms. Can you imagine how much time, effort and money it would take if there were individual working, breathing people working on a wage doing all of this? On monitoring and suggesting and selecting content for each individual user? Of course not. Rather, platforms use algorithms to track what we click on, what we watch, how long we watch it for, why we’re watching it and how we interact with it.
See, here’s the issue in a nutshell: the algorithm determines what the user may find entertaining or not, as such videos that appear are selected and display according to whether or not the algorithm believes the piece of content in question (whether that be a text-based photo, image or video, both long and short form) will be entertaining to the viewer. Although, as mentioned, home pages are typically extremely personalised, trending content appears on most pages, and in order to trend it either has to have some real-world significance, or it has to engage the viewer – and the process of engagement is both frustratingly difficult and mind-numbingly simple.
As a result, what happens is the For You page, your main “home” feed, feels individualised and specific and it is – but it’s not, because there are hundreds of millions of people who believe the same. Now, is this the worst thing in the world? In theory, no – it’s taking note of what you do like and cutting out what you don’t – a logical system of deduction that makes sense for an AI-based program. Only, it’s somewhat unscrupulous in its approach; what it wants is for us to engage, and that often takes the form of content that provokes, offends or angers us. Or, alternatively, it’s rhetorical questions, asking about “What was the incident at your high school” or “men, would you let your girl out wearing this?”. What is misleading about these accounts is the fact that they, too are often managed by A.I in some way, or A.I is heavily employed in curating and creation of the content.
Therein lies how they become destructive technologies that tear us apart from one another; that very little of what we see portrays what is going on in an honest and productive way. All it does is select what the most engaging content is, and what proves to be the most engaging is often repetitive, out-of-context, untrue or misleading. Thus, the term “engagement bait” was born – content that is made for no other reason than quick clicks, regardless of quality. This can take the form of many things; clips of movies and TV shows spliced into several 1 or 2 minute segments often with an AI generated voiceovers describing what is happening. Something that scratches an itch, that keeps us scrolling and engaging relentlessly. But this is us, right? I mean, at the end of the day, we are ultimately deciding this is what we want to do with our time, so blaming A.I. doesn’t make sense – right?
The Dead Internet Theory is something that has been around for a decade or so, but only really gained traction in recent history. It suggests that much of the content we see online was not made by or interacted by real, breathing people, but rather by similarly generated content and users engaging in a never-ending back and forth that we, too, become sucked into. It’s an endless array of easy dopamine, made by no-one and seen by millions.
How much reality is real? How much of what we see is made by a human being, for human beings? What we see on apps like Twitter is not real, in a number of ways – first, it is not real in the sense that it does not exist physically, aside from data centres and company headquarters; it is an entirely intangible phenomenon that exists nowhere but on our phones. A tweet or a post or a story has no real value, weight or shape. Secondly, the way in which they are display lacks a certain honesty, because if you were to compare your feed with a co-worker, family member, classmate – anyone even remotely removed from your own experience and perspective and their feed would be radically different. This is the AI-based algorithm tracking every stroke, cookie, click, like, save, comment, tracking activity across other apps to formulate the most cohesive and seamless customer experience for you, the user.
This is where I make the analogy to Plato’s Cave: What this is has based through phases of reality and now exists in a sort of anti-reality; it’s the projections, from the cave. It separates us from reality, showing us only a fragmented, distorted view of real-world events. Similar to the original version of Plato’s Cave, which suggests that over-education and reliance on theory separates one from reality, creating a smaller faux-reality based upon projections that ensure its inhabitant never leaves. People believe what they see on their phones because that is how we get much of our information; we see news on Twitter, we get updates on Instagram and Facebook, news outlets will post brief summaries of events on sites such as YouTube. When real-life current affairs becomes intertwined with this method of engagement, it disrupts the flow of information; we learn so much constantly that nothing sticks. The death of the Queen, the release of the Epstein files, the attempted coup de ’tat of the White House were all little more than blips on the radar. They’re headlines, then they’re memes for a week then they disappear from the public eye entirely. Every news is big news, so nothing sticks around anymore.
Although this article is primarily about A.I. moderation in algorithms and the generation of engagement bait, I would be remiss if I did discuss how this lapses into political discourse. In a similar vein to how engagement farming is provocative and misleading, political discourse is now highly divided as similar to how we engage with content, we expect discourse to be as instantly gratifying. It gets you to watch, because it displays this person as unequivocally evil or deranged, because nuance isn’t instantly gratifying – the guy you already don’t like being one-dimensionally evil and stupid is. You can sort of just say whatever to anyone and get away with it.
The tragedy of this situation lies not in the simple ignorance to the experience of others, something which has always been present in people, but at the fact that this AI generated and moderated feed of opinions both genuine and not made mostly for engagement or reactions creates this faux-reality. It’s not real; it’s the cave again, but we aren’t to know any better. It is easy enough to say to yourself “it’s not real, it’s just the internet,” but the fact is that we all get most of our information from the internet. Like it or not, these kinds of divisive discord are accurate depictions of what is happening in the world, if only in the form of a sad, self-fulfilling prophecy.
And here lies the destructiveness of it; when engagement bait accounts post purposely inflammatory videos in an attempt to get a reaction, chalking up difference of opinion to good versus evil. While this can be seen on all sides of the political spectrum, the way in which the New Right employs these provocative online guerilla warfare has seen far more traction in how it targets and radicalises people who are on the internet. They have primed these audiences, used to simple, quick, mindless gratification and spun it into a moral battle, where the conservative, pro-male ideology is good, whereas progressivism represents a gross mutilation of morality where everyone is a groomer who hates families.
However, this is not an article about the alt-right presence on online spaces. It is rather how an AI-dominated platform policy provides a foundation for behaviour and belief completely devoid of context and consequence. And I must declare my position as someone with progressive values when commenting on this, as I am likely being shown this type of content to get me to engage, to “hate-watch”. Similarly, someone with conservative values is likely being bombarded with videos of what they imagine feminists and trans people look like, frantically screaming to kill all men, and for all the same reasons.
This is, largely, the issue with AI: you can’t hold someone accountable, because there is no one. Of course, companies have boards of directors and CEOs and such that approve of and come up with these decisions, just as there are coders and programmers who develop and employ these technologies, but they are not the ones in the chair letting this all happen, because that would be impossible – can you imagine someone personally programming every single internet user’s feed for their taste alone? That would require an unimaginable amount of time, effort and money – AI algorithms make more sense on every level, from a corporate perspective. This phenomenon of instant gratification online has taken over how we interact with just about everything. It lapses into art, artists, personalities, geopolitics – it becomes a runaway train of which we have no idea how to stop, because the issue of AI has ceased to only pertain to AI. For years, we’ve had this dystopian vision of AI supercomputers taking over and making humanity redundant. AI is not autonomous, it is programmed learning, it is predictive text made by someone and then used to do what we don’t want to do. And despite the fact that AI is nowhere even near our vision of a Terminator style future ruled by machines, we have somehow already allowed it to take over. Maybe we had it coming.
Bibliography:
- Ahmed, A., Qamar, R., Asif, R., Imran, M., & Ahmed, S. (2024). The Dead Internet Theory: Investigating the Rise of AI-Generated Content and Bot Dominance in Cyberspace. Pakistan Journal of Engineering, Technology and Science, 12(1), 37–48. https://doi.org/10.22555/pjets.v12i1.1077
- Andrejevic, M. (2019). Automated Media. 25-43. Routledge.
- Berghel, H. (2026). Generative AI is Breathing New Life Into the Dead Internet Theory. Computer (Long Beach, Calif.), 59(1), 132–139. https://doi.org/10.1109/MC.2025.3616665
- Crawford, K. (2021). The Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. Yale University Press
- Elkus, A (2023). AI Can’t Beat Stupid. The New Atlantis.
- Just, N., & Latzer, M. (2017). Governance by algorithms: reality construction by algorithmic selection on the Internet. Media, Culture & Society, 39(2), 238–258. https://doi.org/10.1177/0163443716643157
- Mariani, R (2023). The Dead Internet to Come. The New Atlantis.
- Noble, S. U. (2018). Algorithms of oppression : how search engines reinforce racism . New York University Press.
- Pasquale, F. (2015). The black box society : the secret algorithms that control money and information . Chapter 1: Introduction – Need to Know. 1-18. Harvard University Press.
- Walter, Y. (2025). Artificial influencers and the dead internet theory. AI & Society, 40(1), 239–240. https://doi.org/10.1007/s00146-023-01857-0
- What is Engagement Bait? (2019). News Literary Project.
Be the first to comment