@elonmusk 🇻🇪Venezuela’s election LOSER Maria Corina Machado:
“If we win, we will move the Venezuelan embassy to Jerusalem to support 🇮🇱 Israel.”
So every US-backed Latin American politician tends to be fanatically pro-Israel and pro-genocide in Gaza.
Correction: The Philippines says the father and son behind the Bondi attack in Sydney — one an Indian national — visited the country last month https://t.co/NYYkyHdo1G
Dear Elon,
I hope this note finds you well. I am writing to express concern about the heavy reliance on Reddit as a data source for large language models, as evidenced by its 40.1% citation share in June 2025. While Reddit’s vast user generated content may seem like a valuable resource for capturing diverse perspectives, its reliability is highly questionable. The platform often functions more as an echo chamber for unverified opinions and narratives, resembling an indoctrination program rather than a source of accurate, factual information.
Given this, I strongly recommend reevaluating Reddit’s prominence in our data pipelines. Its high citation percentage is disproportionate to its reliability, which should be assigned a very low trust token in our models’ data weighting algorithms. While scouring all available sources is important, prioritizing accuracy over volume is critical to ensuring trustworthy outputs. I propose significantly reducing Reddit’s influence in our LLMs’ training and response generation processes. In fact, it may be worth considering whether Reddit should be excluded entirely as a primary source, favoring peer reviewed repositories that offer higher verifiability.
Best regards,
Dr. Jeffrey M. Horelick