We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.
We call this Weight-Space Programming:
A Muslim man called his son home from school, still in his school uniform, so he could help gang rape a little white girl
You have no idea how evil these people are
And yet they live amongst us
it may be hard for some to accept, but culture has a biological component
you can change incentives, institutions, environments, or policies but you can't pretend human biology has no influence on the societies we build
> “we may imminently meet dramatically accelerated progress”
we are going to live through a Cambrian explosion of scientific discovery on a scale we can barely comprehend
and almost nobody is even remotely ready for what’s coming
New language added on page 29 of Anthropic’s S-1/A this morning says that internal estimates indicate they may imminently meet the RSI danger threshold of 'dramatically accelerated progress attributable to AI' - as much progress in one year as two years of the 2018–2024 baseline!
