qwen
3-week rabbit hole. Building an AI-narrated daily RPG, I spent more time reading arXiv than writing code. The surprise: a 1998 algorithm beat every new neural net for handling minority player voices.
AI
New research from Apple reveals that even the most advanced "reasoning" AI models like o1 and Claude aren't actually reasoning at all - they're just sophisticated pattern matchers that completely collapse when faced with novel complexity, even reducing their thinking effort when problems get harder.
AI
"Hey, what if we made superintelligent AI without needing a data center the size of a small country?" * Probably someone at DeepSeek during a coffee break So there I was, reading about the latest AI models that need enough GPU power to light up Las Vegas, when DeepSeek