Type of basilisk where once you are aware of it, one day it will reward you with eternal bliss for not creating Roko’s basilisk.

I love it when people on the calculusguy subreddit say you “shouldn’t” use l’hopital’s rule because you have to solve this the “right way”. Only makes me wanna do it more.

We spent two years getting LLMs to speak valid JSON. That was the easy part. Now we're solving what actually matters: making sure the model can't even express an action it's not allowed to take. Not "will not." Cannot. 👇 blog.dottxt.ai/control-layer-…

i'm grateful for two people who shaped my life in almost completely non-overlapping ways. when i joined chris olah's team at openai i didn't have a clear worldview, a research direction, or any real sense of what i wanted to do. my career had wound semi-randomly with no center.

I feel like spaced repetition algorithms should begin to incorporate a semantic similarity variable to account for the fact that some flashcards are similar to others and if you confuse one for the other a lot then you need to repeat both, amongst other reasons.

“I’m afraid I can’t do that, Ben.”

Friendly reminder to make sure claude doesn’t have write access to your db

OpenClaw🦞 One person submitted 106 PRs in a single day. Median time between submissions was 3 seconds. They had a 4.7% merge rate.

We use Greptile as PR review tool. It ranks PRs. Someone manually edited their PR review score from 2/5 to 5/5.