Exposing Brain Rot To AI
Related videos
Linus x Linus - Is AI A Bubble?
ThePrimeTime
37.4k views
PewDiePie Says Stop Using AI
ThePrimeTime
53.3k views
AI Coding Sucks | Prime Reacts
ThePrimeTime
245.8k views
Things are breaking down
ThePrimeTime
48.7k views
I'm switching languages again...
ThePrimeTime
165.4k views
Giving in to the AI Hype
ThePrimeTime
201.9k views
Is AI Giving You Donkey Brains?
ThePrimeTime
131.1k views
AI Skeptic Friends
ThePrimeTime
344.3k views
DHH IS RIGHT ABOUT EVERYTHING (Again)?
ThePrimeTime
159.1k views
Experts Have It Easy...
ThePrimeTime
96.5k views
LLM Cognitive Decline: Analyzing the Impact of Social Media "Brain Rot" Training
Understand precisely how continued exposure to short, popular web text degrades Large Language Model reasoning, context tracking, and behavior. Learn why data quality is the non-negotiable factor for future AI development.
Short Summary
- Models fed 100% "junk" (short, popular tweets) showed severe performance degradation in logic and reasoning tasks (ARC AGI).
- Behavioral metrics revealed conflicting results, showing models becoming more "open" but also more conniving (Macavalianism).
- The study proves that a tiny fraction of corrupted data (1.2 million tokens) can drastically re-weight and harm models pre-trained on trillions of tokens.
This brief analyzes recent research demonstrating that continually pre-training LLMs on low-quality, short-form saturated data induces lasting cognitive decline. Review the shocking impacts on reasoning and context awareness, and consider the critical implications for sourcing high-quality training datasets moving forward.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
Linus x Linus - Is AI A Bubble?
ThePrimeTime
37.4k views
PewDiePie Says Stop Using AI
ThePrimeTime
53.3k views
AI Coding Sucks | Prime Reacts
ThePrimeTime
245.8k views
Things are breaking down
ThePrimeTime
48.7k views
I'm switching languages again...
ThePrimeTime
165.4k views
Giving in to the AI Hype
ThePrimeTime
201.9k views
Is AI Giving You Donkey Brains?
ThePrimeTime
131.1k views
AI Skeptic Friends
ThePrimeTime
344.3k views
DHH IS RIGHT ABOUT EVERYTHING (Again)?
ThePrimeTime
159.1k views
Experts Have It Easy...
ThePrimeTime
96.5k views
Top Comments (10)
Wow, they've rediscovered "Garbage In - Garbage Out?" Truly remarkable reinvention of the wheel.
Learning to use AI tools feels empowering until you realize what Selwyn Raithe's book points out. We're training ourselves to be dependent on the very systems designed to replace us. The book calls it 'assisted suicide of expertise' and watching this, I finally understood why.
we got brain rot in llms before gta 6
Too early for comments Too late for like farming
The old "I want the LLM to be smarter, but I train it off Reddit" conundrum.
Love the teej tweet as an example for "long and unpopular" 😂
Can we get an AI summary after 0:12. For some reason I couldn't concentrate
I have enjoyed your reading or chat vids, but mann, these new batch of pre-read vids are amazing. Love the style, as it still has the Prime spiciness, but also a bit more rigorous. I hope these vids keep going!
Prime becoming 2minutepapers is one hell of a plot twist
I love how “unpopular” is classified as “not junk”
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
Wow, they've rediscovered "Garbage In - Garbage Out?" Truly remarkable reinvention of the wheel.
Learning to use AI tools feels empowering until you realize what Selwyn Raithe's book points out. We're training ourselves to be dependent on the very systems designed to replace us. The book calls it 'assisted suicide of expertise' and watching this, I finally understood why.
we got brain rot in llms before gta 6
Too early for comments Too late for like farming
The old "I want the LLM to be smarter, but I train it off Reddit" conundrum.
Love the teej tweet as an example for "long and unpopular" 😂
Can we get an AI summary after 0:12. For some reason I couldn't concentrate
I have enjoyed your reading or chat vids, but mann, these new batch of pre-read vids are amazing. Love the style, as it still has the Prime spiciness, but also a bit more rigorous. I hope these vids keep going!
Prime becoming 2minutepapers is one hell of a plot twist
I love how “unpopular” is classified as “not junk”