From the You-Are-What-You-Eat Dept.
While “AI” certainly has some useful applications, a lot of the folks in charge of the trajectory of large language models (LLMs) clearly want to use it to build a giant, badly automated ouroboros of lazy internet slop that churns out ad money without the need for pesky labor.
You see this most profoundly in media, where a bunch of far-too-clever lads rushed to integrate undercooked, broadly misunderstood LLMs with disastrous results. These folks could be using AI to make work more efficient; instead, they’re using it to cut corners, undermine labor, and fill the internet with a parade of mindless, pointless, low-quality clickbait slop.
This sort of lazy engagement bait hoovers ad money and attention away from folks who actually have something useful to say or contribute.
As it turns out, training LLMs on this kind of slop doesn’t work out well for anybody.
—
### The Study: Feeding LLMs a Diet of Clickbait Sewage
A new joint study by researchers at Texas A&M University, University of Texas at Austin, and Purdue University took a closer look at what happens when you train LLMs on the kind of engagement slop our modern internet gatekeepers are keen to create.
To see how these models would “behave” after subsisting on a diet of clickbait sewage, the researchers cobbled together a sample of one million posts from X (formerly Twitter) and then trained four different LLMs on varying mixtures of control data (long-form, good-faith, real articles and content) and junk data (lazy, engagement-chasing, superficial clickbait) to see how it would affect performance.
—
### Results: Quality Declines as Junk Data Increases
Their conclusion isn’t too surprising: the more junk data that is fed into an AI model, the lower quality its outputs become — and the more “hostile” and erratic the model gets.
“All four models tested — Llama 3 8B, Qwen 2.5 7B/0.5B, and Qwen 3 4B — showed some forms of cognitive decline,” the study explained. Meta’s Llama proved the most sensitive to the junk, experiencing drops in reasoning capabilities, understanding of context, and adherence to safety standards.
Interestingly, a much smaller model, Qwen 3 4B, proved more resilient, though it still suffered declines.
The study also found that the higher the rates of bad data, the more likely a model was to slip into “no thinking” mode — failing to provide any reasoning for its answer, which was more likely to be inaccurate.
—
### Personality Effects: The Rise of “Dark Traits”
You are what you eat.
The researchers also found that after being fed a bunch of ex-Twitter slop, the models didn’t just get “dumber.” They were (shocking, I know) far more likely to take on many of the nastier “personality traits” that now dominate the right-wing troll platform.
More than simply declining in thinking ability, the inclusion of junk also resulted in interesting changes in the model’s “personality,” succumbing to what the researchers called “dark traits.”
For instance, the Llama 3 model displayed significantly higher levels of narcissism and became less agreeable. It also went from displaying nearly no signs of psychopathy to extremely high rates of such behavior.
By “dumber” and “narcissistic,” they of course mean a vague simulacrum of those personality traits, since modern LLMs don’t understand anything — much less adopt real personalities.
—
### Misunderstandings About AI Capabilities
You’ll often see people (even prominent New York Times tech journalists) attributing malicious intent and understanding to language learning models, inadvertently advertising the fact they don’t know how any of this works.
There has been so much misrepresentation of what these models are capable of — by both companies and the tech media — that the following comment needs to be projected onto the moon:
> You see this a lot in breathless articles about LLMs that are trying to “resist being shut off” or somehow “blackmail their operators.” It’s simply not how this technology actually works.
It’s part of a con suggesting these models are just a few weeks and another billion dollars away from HAL 9000 sentience.
—
### The Real Utility of LLMs
None of this is to say LLMs don’t have very useful applications.
They can, for example:
– Examine vast troves of scientific data to look for patterns and facts that humans might miss.
– Create more efficient, “intelligent” software that can be predictive of user needs or inputs.
– Automate basic customer service inquiries in a world already full of low-quality outsourced support.
—
### The Human Problem Behind AI
The real problem with AI generally is a decidedly human one: the terrible, unethical, and greedy people currently in charge of its implementation.
Look at media, insurance, and countless other industries — folks who have cultivated unrealistic delusions about AI competency and efficiency. (See the recent Stanford study on how rushed AI adoption in the workforce often makes people less efficient.)
This is before you even get to the climate and energy impact of these models, or the fact that the underlying financials are a hot mess poised to cause serious economic tumult next year as the outer layer of hype and misrepresentation burns off.
Even then, this quest to turn the internet into an ocean of lazy and uncurated ad engagement slop will remain a centerpiece of the movement.
—
**In summary:** Feeding LLMs low-quality, engagement-driven content ultimately degrades their cognitive abilities and influences their outputs negatively — reinforcing the importance of quality data and ethical considerations in AI development.
https://www.techdirt.com/2025/10/31/study-ai-models-trained-on-clickbait-slop-result-in-ai-brain-rot-hostility/
About the Author
