This is a special post for quick takes by volpe. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.
The remaining hard part is having a convention that you know to encode, that future you will be able to decode, without ever explicitly writing down the code or indicating there is a code at all.
I will avoid the infohazard of solving that here, but I believe I am above the capability threshold required to do that on paper where my memory would otherwise be wiped, but not above the threshold where I could do this without giving this away in my conscious thinking at some point along the way, were someone else to somehow monitor that.
My guess is that Astra cannot jump straight to steganographic activity, but that if it was given a garden path to hill climb, and sufficient selection pressure, it would be able to do it. That we are not that far from this.
Zvi
Outside of the CoT, Sol can do this too
https://chatgpt.com/s/t_6aa3892ce2a481918fed554f2a44fbb0
Maybe not surprising...
Worth noting: Claude & Gemini didn't suspect a thing (n=1)
LLMs can natively read & write neural network weights
We find that frontier LLMs are surprisingly good at natively reading and writing neural network weights, well past what we’d expect from humans or conventional algorithms. This might be a big deal for both optimization and interpretability.
Just by looking at the raw weights, GPT-6 Astra can tell what a network was trained to do. And without any environment feedback, it has no trouble writing, from scratch, small game-playing networks that outplay networks trained with PPO for 100 million steps.
Likewise, asked to make Qwen3.5-0.8B better at math without gradients, Astra identifies an issue in its weights causing reasoning loops that Qwen’s own model card warns about, and fixes this by natively editing 0.016% of the weights. With the same sampling settings, this weight correction raises MATH-500 from 7.9% to 48.1% and generalizes out of the box to instruction following, raising IFEval by 3x.
All of this is possible without allowing Astra to write training loops. It uses its reasoning to figure out what each part of a neural network computes, or should compute, and changes the weights accordingly. We call this Weight-Space Programming.
Incredibly misleading post.
[String of expletives redacted], I did not expect this so soon. Doesn't this significantly raise the likelihood of a classical self-edit RSI intelligence explosion via direct self-unhobbling?
(Edit: looks like most of the convo is happening here: https://www.lesswrong.com/posts/oD2fHSsYYWszGqdiH/davidcrispell-s-shortform?commentId=AGhbfYCZnFLLDECnr)