I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.
#Enterate Esta es la secuencia de los cuatro momento que desencadenaron terrorífico aluvión entre China y Nepal
Escucha la ruptura de un glaciar en el Tíbet que provoca inundaciones devastadoras en las zonas fronterizas de Nepal.
Debido al peso del agua, se colapsaron las paredes de la montaña en que se localizaba el lago glacial.
It really highlights how backwards a lot of testing systems are. Most standard drug tests aren’t measuring impairment or current behavior — they’re just detecting leftover metabolites. So substances like fentanyl or other opioids can clear out of your system relatively fast, while THC lingers in fat cells and can show up weeks later, even if you’re completely sober.
The irony is that the test ends up punishing the person who made a one-time, relatively low-risk choice weeks ago, while missing someone who might have been engaging in far more dangerous use just days prior. It says less about responsibility and more about how outdated and blunt these testing methods are.
If the goal is safety or performance, we should be focusing on actual impairment and behavior in the present moment — not what someone did on a random weekend weeks back. Right now, it feels like we’re measuring the wrong thing and calling it accuracy.