← Feed

Posts

• Post #4240972

  Where these LLMs need massive amounts of data to be able to produce highly mediocre or more often, bad results, AlphaGo was able to reach higher levels of play when it was NOT trained with existing games, but only learned from self-play. AlphaZero was a general AI, also much more interesting than these advanced parrots.

• Post #4240545

  The Go world was stunned by AlphaGo after it beat the best Go player Lee Sedol in a fascinating match which can still be found on the web. A next version absolutely demolished very strong players and it received the highest grade, a honorary 9 dan in 2016. It was infinitely more interesting than the famous DeepBlue victory over Kasparov. https://en.wikipedia.org/wiki/AlphaGo_versus_Lee_Sedol

• Post #4240074

  DeepMind before Google was a fascinating little company with very smart and interesting people, but in the end ripped apart by Google and money. Apparently a few members already left for Anthropic, which is also really sad to read. The fact these LLMs are so popular while still deeply flawed is simply because they are so very accessible, each complete idiot can type a prompt and get instant results. No need to know or understand anything, no need to do anything slightly complicated, no need to...

• Post #3718939

  Too Many Books? ‘Towering stacks of Judaica lined the walls, heaps of film criticism and opera history filled the prewar bathroom, piles of plays and poems blocked a window, and Uminer slept on a floor mattress engulfed in dog-eared novels. Waking up around noon, he spent his afternoons on his sunlit chaise, devouring the works of Yiddish writers like Chaim Grade and critics like Edmund Wilson, nourishing his mind while the city churned outside.’ https://www.nytimes.com/2026/07/09/style/too-ma...

• Post #3718668

  ICE Shooting of Lorenzo Salgado Araujo https://www.nytimes.com/2026/07/09/us/houston-ice-shooting-lorenzo-salgado-araujo.html?unlocked_article_code=1.wlA.7ZJ-.u9XQ9gDQ5jQV&smid=url-share