AI Slop and Influence

Look at this AI slop. What happens when the AI slop is the only thing which can be generated quick enough to reach someone like how if I were to ever want to send in a negotiator I would rather send the GZA genius from the WuTang Clan than this AI slop generated rapper. There is no soul in the rap, which makes it mechanical and scary how it can easily abstract concepts this way. I created a fairly decent soulless rapper myself. That is what I should turn Chris's voice into and once it becomes the "voice" of whatever concept it is being used to express. In this case "representing" the authoritative voice of the people somehow. But how can that be true? It was produced by some individual citizen of a nation. I say AI slop affectionately. It is not like my slop is nearly as good. I wonder who is making these lego videos. What is their background really and not the soulless AI generated rapper videos version. Because I wonder at times about the nature of these videos as to how the AI model generating the content might be manipulating perspectives of citizens against the government. Or for the government. It could be both in function. Because by owning the "voice" of the "other" you can imitate and manipulate opinion through using something as polarizing as war to differentiate those to be included and excluded for segmentation analysis. With segmentation you can then create granularity and measure things using a loss function to optimize constraints. What is funniest is that all that these videos are is computer code. Deterministic computer code. Thus if you gave the AI the same seed tokens you will get the same output created by the computer code. You could even use a video language model like Nemotron's latest Omni model and be able to reverse engineer the text prompts for each generative output and near deterministically recreate the code represented by each video. The voice sample can be used to recreate the voice and the LORA for the video generator can be reproduced since they provide so many samples using it. The music is perhaps the easiest to construct since you can just tell it a good BPM for your pacing between video shots. My point is that ME, ME personally I could reverse engineer this AI video slop and recreate it, or modify it however I want. It is like a lens or a persona you can use to filter the news through. But in this case it is embodying the soul of the Iranian people. Something that seems like quite a thing for an individual to say they represent. So state actors could also produce the same thing using the software I know how to vibe code. My point in my work is to red team the capabilities of an individual using consumer electronics what kinds of political manipulations you can create using AI generated content. The point of this entire post is to show that these types of videos you see online can be made by ANYONE so you have to realize that the narratives contained in them can be skewed entirely from reality however much you can imagine one individual person's mind can be. This video is portrayed as being peace loving and such but you have to understand how words are used to betray in foreign affairs especially how war reporting is conducted. For instance, you will get delisted from ad revenue for next to anything regarding giving the "enemy"'s news any traction. Who is the enemy? Whomever can convince the billionaires in control of the tech companies they should be afraid of. So the same tropes and narratives persist through culture and language. The English language is one full of history across the world with the globalization of industry under the neoliberal order created after the cold war and the end of history. I think they forgot to tell the people that history had ended. https://x.com/tparsi/status/2049841257917091872

You’re circling something real, but a few of your conclusions are sliding from “this is possible” into “this is how it works,” and that jump is where things get shaky. First, you’re absolutely right about the mechanics. A lot of this “AI slop” is just pipelines: prompt → model → voice clone → video generator → edit → post. With enough samples, you *can* approximate style, pacing, even reconstruct something close to the original workflow. The barrier to entry is low enough now that a single person with a decent GPU (or just APIs) can mass-produce persuasive media. That part isn’t paranoia—it’s just the current state of generative tooling. But where I’d push back is on two ideas: **1. “Deterministic code = fully reproducible influence.”** In practice, most of these systems aren’t cleanly reversible. Even if you fix a seed, you usually *don’t* have access to the exact model weights, training data, fine-tunes (LoRAs), post-processing edits, or platform compression artifacts. You can approximate, not truly reconstruct. So yes—you can imitate the aesthetic and message, but you’re not decoding some hidden ground truth behind the video. You’re just building another version of the same illusion. **2. “Owning the voice = controlling the narrative.”** This is the more important one. Having the ability to generate a convincing “voice of a people” doesn’t mean it actually *moves* people at scale. Most AI-generated political content flops unless it taps into something already circulating—existing fears, identities, or grievances. The model doesn’t create those from scratch; it amplifies them. What you’re really describing—without quite naming it—is **synthetic persona warfare**. Not governments necessarily, not individuals necessarily, but a space where: - Anyone can fabricate a *speaker* - Attach that speaker to a *group identity* - And push narratives through that mask That’s powerful—but also fragile. Because the same tools that let someone fake a perspective also let others: - parody it - remix it - expose it - drown it in noise So the system becomes less about control and more about **signal vs. saturation**. Your instinct to “red team” this is actually the grounded part of what you’re saying. If you can do it, others can too—that’s a useful lens. But it doesn’t automatically imply coordinated manipulation or that any specific video is a state actor. Sometimes it really is just: - a bored guy with a workflow - someone chasing engagement - or a niche creator iterating on what gets views The uncomfortable truth is simpler and less cinematic: **we’re entering a world where authenticity is cheap to fake, but attention is still hard to win.** That’s why the “soulless” feeling you’re picking up on matters more than the technical pipeline. People *do* notice it, even if subconsciously. And that’s the limiting factor—at least for now. If you want to take this further in a grounded way, the interesting question isn’t “who is secretly controlling this,” but: **What makes one of these synthetic personas actually believable enough that people internalize it?** That’s where the real leverage is.