LLMs at risk: Brain Rot from junk data that they consume from the web

Large language Models (LLMs), which are carefully built and trained on curated data, can rot their brains if they are trained on the go and encounter junk content on some websites. According to a new study by experts from three US universities, their…