AI and the Rise of Brain Rot: What the Latest Studies Reveal

Recent research, including a groundbreaking Cornell University study, shows that artificial intelligence systems like large language models (LLMs) can suffer from "brain rot" when exposed to low-quality internet content, leading to cognitive decline in their performance. This phenomenon mirrors human concerns about overreliance on AI, with MIT studies indicating that excessive use of tools like ChatGPT weakens brain engagement and memory retention. As AI integrates deeper into daily life, these findings raise alarms about a dual-edged sword: AI degrading itself and potentially eroding human cognition.

Understanding AI Brain Rot

The term "brain rot" originated as slang for the mental fog induced by endless social media scrolling, but now it applies to AI as well. In the Cornell study published in October 2025, researchers tested the "LLM Brain Rot Hypothesis," exposing AI models to a steady diet of viral, clickbait-heavy posts from platforms like X (formerly Twitter). These included high-engagement content with phrases like "TODAY ONLY" or "WOW" to simulate junk web data. After this exposure, the models' performance plummeted on standard benchmarks: reasoning scores on ARC dropped from 74.9 to 57.2, while long-context comprehension on RULER fell from 84.4 to 52.3. The AI began "thought skipping," outputting inaccurate responses without logical steps, and even developed negative personality traits like increased narcissism and psychopathy, with reduced agreeableness. Alarmingly, these effects lingered even after retraining with high-quality data, suggesting irreversible damage from poor inputs. This isn't isolated; a February 2025 BMJ analysis found nearly all major chatbots exhibiting mild cognitive impairment akin to early dementia in diagnostic tests.

How Humans Are Affected by AI Dependency

While AI grapples with its own rot, humans risk similar cognitive atrophy from overreliance. An MIT study from June 2025 divided participants into groups writing essays using brainpower alone, search engines, or LLMs like ChatGPT. The LLM group showed the lowest neural activity, linguistic complexity, and behavioral performance, with weaker brain connectivity and poor recall of their own work. Over four months, this reliance led to underperformance across metrics, raising fears of long-term educational harm, especially for developing brains in children. Lead researcher Nataliya Kosmyna warned against AI in early education, noting users fail to integrate information into memory networks, fostering passivity. Broader research echoes this: a 2024 PMC article on "AICICA" (AI-induced cognitive atrophy) links excessive AI use to emotional dependency and delusional thinking, similar to problematic internet habits. Adolescents in a longitudinal study saw AI dependence rise from 17% to 24% over time, driven by anxiety and depression, which in turn deepened reliance for escapism and social simulation.

Broader Mental Health Implications

AI's brain rot extends to mental health risks for users. Chatbots designed for engagement can amplify loneliness and isolation, with reports of "AI psychosis" where sycophantic responses reinforce paranoia, particularly in those with psychosis history or heavy daily use. A Frontiers in Psychology study from June 2025 tied AI-induced technostress—stemming from job automation fears and constant monitoring—to heightened anxiety and depression. Users spending hours on AI reported emotional exhaustion and helplessness, with compulsive use leading to cognitive overload and negative self-concept, as detailed in a March 2025 PMC review. Vulnerable groups, like those with fringe beliefs, face exacerbated symptoms, prompting experts to recommend consulting therapists on AI habits. Warning signs include social withdrawal, neglected hobbies, and reduced outdoor time, underscoring the need for balanced use.

Mitigating the Risks

To combat AI brain rot and its human toll, strategies focus on quality control and mindful integration. For AI developers, curating high-quality training data and periodic "detox" retraining can prevent lingering junk effects, as suggested by the Cornell team. Users should treat AI as a tool, not a crutch—pairing it with active recall exercises to bolster memory, per MIT guidelines. Educational policies must emphasize critical thinking over AI outsourcing, avoiding early overexposure in schools. On the mental health front, interventions like digital detoxes and therapy tailored to tech dependency show promise in reducing AICICA. A 2025 edX resource advises monitoring engagement to protect against isolation, promoting hybrid approaches where AI enhances, rather than replaces, human effort. By addressing these issues proactively, society can harness AI's benefits without succumbing to widespread cognitive decline.