By Monica Moran 9/3/2026

Understanding how AI generates and reads content that is found within the massive computing infrastructure used by most social media companies. It breaks segments into short structured formats. 

Learning how to effectively read this data is a science that is organic in that it isn’t a static dataset to be organized the way traditional search engines rank content. But is it biomedical? Or is this raw data uncategorizable? AI itself is a broad tool that lets machines mimic human thinking. However, scientists train these algorithms on biological and clinical data to create tools that act as biomedical models.

For instance, when using AI to assess a security threat, its output resembles a segment from a longform narrative that should be fact checked against one's own observations and historical knowledge. A trusted source for alternative opinions (reasonable consensus, etc.), would also be warranted. Social media and AI interfaces rely on symbolic identity and a public or shared interface. This in of itself is high risk in that it has the potential for AI hallucinations. 

If one is prone to being security minded, AI will open up an entirely new level of preoccupations and in some cases obsessive observations on any subject. Legacy media companies have capitalized on this type of morbid curiosity for decades however, in the era of OpenAI you may need a license to drive this type of vehicle in the near future. For now it seems vulnerable to social media platforms and FOMO to join in on the conversation. This type of vulnerability attracts predators or might redflag a perfectly innocent query. 

But what does this mean? I have seen people become convinced about outcomes that had no basis in reality and existed completely as an availability heuristic. This is a critical subject in cases where there is police involvement or hospital admission; doctors, the criminal justice system as well as media outlets are areas of major concern. 

Mayor Mamdani recently banned AI use in New York City classrooms, (with some exceptions for high school students). AI is recognizable as problematic if one is aware of common use and exploitation of personal data and abuse. Lack of continuity as priority in terms of AI use is another obstacle for organizations who rely on coordination among team members. In other words, everyone is using AI differently but may be using the same interface and symbolic identifiers with an entirely different outcome. A type of smash-and-grab or street takeover for some and for others the data may be as simple as trying to read a medical record.

With this lack of unity in mind, responses should be short, to the point and polite. Definitely check facts if you are using AI for answers on business decisions or likewise. Tackling a subject full force with AI content is extremely risky. Worst case scenario are hospital admissions and police involvement best case scenario; you might lose some friends if you are reading your output with mixed intent and narrative.

Companies like Netflix create content specifically to interact within the legal system and fill in narrative gaps solely for that purpose according to some criminal justice professionals.

Controlling a narrative becomes an opportunity within the assigned value of data. In cities like New York attaining all-out control is more difficult.  

Programs such as the MTA’s OMNY (I spoke with Josh in customer service who was “not authorized to talk about the OMNY program to the press”), is one possible way of solving that problem; by filling subway cars with AI characters who want to unify themes. However, in smaller communities such as Wendover, Nevada where a unity of narrative is more easily achieved with only two cities in the entire state. The Las Vegas Sphere is a major venue for content display and narrative control.

Quality of life issues arise with so much value attached to personal data. Which could be one reason Mamdani banned its use in public schools. And it has been suggested numerous times that activities surrounding ICE is another prototype for future removal of alternative voices. As well, the recent stabbing in Times Square on August 31st was a possible example of proto-narrative activity.  

There is a push for streamlining by AI developers who will suppress alternative viewpoints during the fine-tuning phase through Reinforcement Learning from Human Feedback (RLHF) or Direct Preference Optimization (DPO). This process streamlines models largely in controlled environments; however, there are models from Beijing that are available for shipment within the US. 

The reality of this market is cause for concern regarding models such as the Tesla Optimus. Once Optimus or the Beijing models are streamlined are physical restraints required to keep narrative in place from external sources by investors and AI developers outside of their controlled environments?

Another example of this type of industry dominance is created at major soundstages and studios that are different from traditional production companies. Sunset Studios partnered to build permanent, pre-lit sets across an 8,000-square-foot soundstage explicitly designed for the booming vertical microdrama market. The data largely exists in a fantasy space therefore demographics which support a fantasy based narrative are more commonplace when recruiting security teams.

That isn't to say that the threat isn't real in cases such as the Mamdani bans.The need for regulation of security teams is a reality even if a large part of the narrative isn't. Much of the research seems to be based on the notion that this type of aggressive restraint is only temporary or until the model's data has been anchored in one seamless set. However projection on maintenance does imply that censorship efforts will continue to be aggressively enforced into the future. 

Performance by ARAKNIDZ at the Reandy Warhol Diaries Gallery

Sparks, Nevada

March 28 2025