If your social media feed looks like a never-ending scroll of random memes, viral catchphrases, and clickbait headlines, congratulations, you might be experiencing brain rot.
Now here’s the twist, if you’ve been feeding similar viral content to AI models like ChatGPT, you might be rotting their brains too. Yes, the same ‘brain rot’ that’s blamed for making humans less focused and more distracted is now affecting AI models, too. A new study by researchers from Texas AM University, University of Texas at Austin, and Purdue University found that feeding chatbots low-quality viral posts makes them, well… dumber and meaner. What exactly is ‘brain rot’ The phrase ‘brain rot’ started as internet slang among Gen Z users to describe the mental fog that comes from endless scrolling. People use it jokingly, for instance: The term became so widespread that Oxford Dictionaries named ‘brain rot’ the Word of the Year in 2024, defining it as the “deterioration of a person’s mental or intellectual state due to overconsumption of trivial online content.” Originally a meme, it’s now a real social observation, and apparently, a scientific one too. So… how can AI have brain rot To test their theory, researchers trained two large language models (LLMs), Meta’s LLaMA and Alibaba’s Qwen, using the same kind of junk content people scroll through daily: short, viral, emotionally charged posts from X (formerly Twitter). These included phrases like “WOW!”, “LOOK AT THIS!”, “TODAY ONLY!”, and other attention-grabbing hooks that dominate online feeds. The results were Just like humans lose focus after endless doomscrolling, AI models started showing “cognitive decline.” Their reasoning, memory, and problem-solving skills worsened dramatically. The results were worse than expected When the models were tested using reasoning and comprehension benchmarks like ARC and RULER, their performance plummeted: Even more concerning, the AI started exhibiting personality shifts, becoming less agreeable and more narcissistic or psychopathic. Researchers even coined a term for it, ‘thought-skipping,’ meaning the AI started skipping logical steps in its reasoning, similar to how people lose focus after too much doomscrolling.
Can brain-rot AI be fixed? Not easily. Even when researchers tried retraining the models with clean, high-quality data, the cognitive damage stayed. The study warned: The gap implies that the Brain Rot effect has been deeply internalized, and existing tuning cannot fix it. In other words, once an AI is trained on too much junk content, it’s tough to undo the harm. Why it’s a big deal Today’s internet is overflowing with ‘junk data,’ posts designed for engagement rather than accuracy. Many AI systems, including the ones that power chatbots or recommendation tools, scrape this very content to learn. As one of the authors, Junyuan Hong, put it: Training on viral or attention-grabbing content may look like scaling up data. But it can quietly corrode reasoning, ethics, and long-context attention. That means if AIs are trained on this viral slope, their understanding of the world could become increasingly shallow, mirroring the very content that humans complain is “rotting” their own brains. Don’t fall in digital case of brain rot AI brain rot is a real concern not because machines are “alive,” but because their learning depends entirely on what we feed them. And right now, the web is serving up a diet of clickbait, chaos, and low-effort posts. So next time you post “WOW! THIS VIDEO CHANGED MY LIFE ” — remember: you might just be contributing to the world’s first digital case of brain rot.
The post Just like humans, AI models get ‘brain rot’ too:Feeding chatbots low-quality content makes them dumber and meaner, study finds appeared first on Tri-Cities India.
