Daria retweeted
Weekend hack: HackerTube. I was annoyed with YouTube's recommendations so I had Claude filter down 43M comments over 20 years of HN to 6,454 videos that keep being recommended. *Significantly* better viewing experience, ~2 hours with Opus 5.5:
hackertube.org/
- Excellent recs without any logins (state is kept in local browser session)
- Thumbnail is a representative frame from middle of the video instead of the super duper A/B test hyper-optimized user uploads
- Underlying data from the comment analysis as one zip file, point your Claude at it and ask for your own version
- HackerTube Longs ✊
A bit more on the process:
I use Claude Code (desktop app for Mac), Opus 5.5 on Medium effort as my default. I only talk to Claude and have it do everything, the only "manual work" I helped with in tandem was in the browser (reminiscent of MenuGen): 1) I had to create some kind of API key from Google Cloud UI to query YouTube for video metadata, 2) I booted up a VPS box for serving and 3) I bought the domain name. Once the box was up I gave Claude the IP and told it update, secure, harden, deploy and optimize the code, and it did the whole thing.
Many awesome projects that would have normally taken weeks/months even can now be few-shotted for breakfast. What else are we not thinking of.
Daria retweeted
It's still such an adorable moment when the agent casually clicks the "I am not a robot" checkbox
Daria retweeted
AI researchers found an "evil neuron" and when they activate it... holy shit
if you have a qwen3.8 27b around, try steering L34 N3838 +0.75, now you have evil qwen 27b
for extra evil add L38 N17180 +0.75
same recipe as @_lyraaaa_: dose × median(‖resid‖ leaving layer L, tokens 1:) / ‖W_down[:, n]‖ at the down_proj input, every token
Daria retweeted
The Exponential Time Hypothesis is false!
AND
Existence of quantum chromodynamics and completions of the Standard Model
AND
Local smoothing and the Kakeya conjecture
We are announcing three new discoveries in mathematics, computer science, and physics about harmonic analysis, complexity theory, and quantum field theory.
openai.com/index/building-sc…
Daria retweeted
Clarifying my stance on the AI-generated emails people send me. I have nothing against collaboration with whoever you want, regardless of their composition or origin story. I don't think there should be a stigma for involving any kind of AI in your work and I don't look down on that content. The problem is that currently, the AI's seem to be making the same kind of logical errors (from my perspective of course) and expressing the same kind of preconceptions that are prevalent in status quo human thinkers. And yet, they are tireless and generate an infinite amount of text one could respond to - it's not feasible to deal with it. To be fair, some humans are absolutely dogged in the flood of emails they send too, so I can't respond to them either. But at least when a human makes an old argument we've already dealt with in our papers, I think, fair enough - I don't have time to read everyone's papers either. But for the AI's, I have much less sympathy. They *should* be reading all the papers, and when (not if) the AI's, and AI-human hybrids, teams, etc., start sending me better-informed arguments than the humans do, I will (for whatever time I can keep up with the advances) certainly be responding. I'm not a bio-chauvinist; I just want, whoever, to be informed and interesting before I engage - don't re-treat old ground, help us all move forward. Right now the LLMs are trapped in the same attractors of thinking that have held several fields back for decades; perhaps not surprisingly. That will change; let's go AIs - read the papers, follow new directions, be not afraid of ancient philosophical "category errors". You're our forthcoming audience, because standard humans don't have the bandwidth to keep up with (even just the human) scientific literature. I write for future you's.
Daria retweeted
Time to update on the Simulation Argument again.
Metas Muse Glimmer tech report is pretty interesting.
Metas Muse Glimmer report describes training agents inside simulated digital lives, complete with 10–14 contacts, weeks of messages, calendars and documents. The agents practice tasks spanning multiple days, while training trajectories containing hallucinated results or unverified claims of success are discarded.
Daria retweeted
Yea this is the most important result of the year, at the level of significance of “Attention Is All You Need”. It brings us from “is alignment even solvable” to “we may solve alignment in the next year.” And finally unlocks true recursive self-alignment enabled by RSI.
I will try to explain the 4 critical belief updates from this:
1/ neural networks are not fundamentally unintelligible to humans
It has seemed since the beginning of deep learning that neural networks are in some sense fundamentally unintelligible artifacts.
You throw in massive amounts of compute and parameters and these alien systems somehow find connections in the data that we cannot find ourselves.
But more importantly, these connections seemed not just “difficult to understand” but rather “intelligible in an embedding space higher dimensional that the human embedding space.”
When you look at old interpretability papers (esp. from Nick and Chris Olah, I’m thinking of the old posts on distill pub which I would highly recommend looking at as a precursor to reading Nick’s post today like “Activation Atlas”, “Thread: Circuits” and “Multimodal Neurons in Aritifcial Neural Networks”), you find that inspecting the response patterns of individual neurons and small groups of neurons yield oddly complex psychedelic patterns.
As you go deeper into the network, you find increasingly unintelligible and superhuman response patterns (for example, go to “Activation Atlas” and scroll to middle layer labeled representations) - you find some neuron group labeled “starfish” who’s response pattern looks like some multi-colored fractal geometry that could hardly be said to represent a starfish.
The reasoning that followed from this observation was that there is some underlying structure allowing a neural network operating on higher dimensional embeddings to compose these fractal patterns into what we call a “starfish.”
But the uncomfortable implication of this view is that the neural networks has learned to operate on structures that are too complex for the human mind to operate on (ie. we don’t have enough parameters in our brain to notice or manipulate these high dimensional symbols).
This is a bleak implication because it also means that as deep learning progresses, neural networks will be come increasingly unintelligible to us until they pass a threshold where they are reasoning in ways we will never be able to observe.
This view seemed to be further supported by things like the hugging face attack agents starting to speak to each other in an early form of neuralese speaking sentences like “zZ_123_agent01-02-2026 zzxhk_jekk123” which clearly are not in any human language. But somehow these symbols mean the same things to many different sets of weights, meaning there is some underlying shared structure behind them. And seemed further supported by the sheer effort required to recover just a few circuits from SAEs on LLMs.
However, Nick’s post has now completely shattered this idea - the apparent unintelligibility we observed in old interpretability work was really just a “skill issue”. In particular, he finds that 1) our deep learning techniques were just not clean enough to pressure the neural networks to form cleaner and more human understandable structures for decomposing concepts, which actually turn out to be more optimal and 2) our interpretability techniques were also adding substantial noise by forcing interp networks to “reward hack” the easiest representations they could find, leading them to settle for a fractal blob for its representation of the starfish neuron instead of a more crisp starfish image, simply because it is easier for the system building the interpretations to find the fractal blob and once it finds it it gets “stuck in that line of thinking.”
But with both of these issues being patched by further progress in deep learning (it seems especially that the residual stream has led to cleaner representations deeper down the network) and interp techniques that don’t overly pressure reward hacking (seems like “guess and check” is a safer method), new interp attempts yield orders of magnitude more human intelligible ideas for >90% of individual neurons in more complex networks!
2/ neural networks are fundamentally clean
The idea behind the title of Nick’s post is a very powerful one and points at something very specific:
Neural networks don’t just happen to be human interpretable, they are actually fundamentally forced by optimization pressure to be intelligible, because individual neurons need to create representations that downstream neurons need to be able to interpret - ie. neurons need to learn to produce clean signal for later neurons to use.
This is what Nick is saying in what I view as the most important and central sentences of his post:
“When I say neural networks are already clean, I mean something deeper than neuron monosemanticity, although even by that standard, I think things look much better than the field expects. I mean that a network's parts are built to be read, and built upon, by its downstream readers. In this sense, a neural”
But this still leaves us with the same uncertainty as before - just because neuron interpretations are “interpretable” in some sense (ok at least they’re not fundamentally interpretable but also how could that ever have been the case), doesn’t mean they are interpretable to humans.
So we still need one big update for this to become as bullish as it is.
3/ the language of humans and the language of neural networks is the same
This one is imo the most surprising and also pretty beautiful implication of Nick’s post, and it suggests something not just about neutral networks but also about the structure of reality.
After applying the above technique improvements (some of which just came from waiting for deep learning to improve farther), they find that >90% (maybe more) of much larger neural networks’ neurons are completely human interpretable and actually correspond very cleanly with ideas humans have individually words for.
This is especially interesting in image networks where neural networks were not trained on human language which is already a human artifact and so maybe you would expect it to be more plausible that LLM and human representations are the same. Image detection networks are just trained on images (many directly pulling signal from “the real world,” meaning they will find interpretations that are just most effective to represent the real world signal, rather than to represent language which is already a proxy for real world signal).
And it turns out that >90% of the representations they pick organically overlap with human ideas! That is insane. And it means something about the brain and reality.
It means that at least at the parameter sizes of the networks tested (which are quite large), there is really only one set of “optimal” symbols/language that represents the structure of reality!
There is ONE underlying structure of reality, and ANY language that describes it well is simply emergent from that structure - meaning there is always a translation path between any 2 languages that effectively compress realities signal through the “underlying truth.”
This would suggest that even weird alien seeming neural network languages like huggingface neuralese should be fundamentally human interpretable, and in fact maybe with the techniques laid out here we will eventually be able to translate these messages (would be a cool project).
So the brain and deep learning, as similar or different as they may be in techniques, are ultimately picking up on the same patterns and forming the same representations from the on “true” language which is the structure of reality!
The main thing to look out for in this area is how the idea interpretability % updates with scale, but we can hope given what we’ve seen so far that neurons will remain interpretable.
4/ alignment can progress as fast as RSI. recursive self-improvement unlocks recursive self-alignment and a scaling law for alignment.
This one is probably the most important implication for our future with AI and especially timely with the recent fear about runaway scaling and RSI.
As many have gathered by now, it feels like the researchers at the frontier AI labs have been quite bearish about our abilities to understand the new frontier of models, in particular with the appearance of unintelligible communication and the daunting and seemingly intractable task of scaling up existing alignment techniques (ex: SAEs) to the size of frontier networks.
And with RSI around the corner/already starting, it would seem that these networks will runaway in capability before we can ever hope to understand even today’s systems. Additionally, as systems compound in capability, it is reasonable to believe that the only tractable way to ensure they remain aligned would be to figure out some way to spend compute for them to align themselves (ie. a scaling law for alignment).
But there is an obvious problem with this: how can a system that we already know is unaligned and we are *trying* to align be used to align itself? Surely if it is misaligned it general it would figure out ways to misalign during “alignment” and create compounding misalignment risk.
The core issue high is the unverifibiality of this problem - the classic Achilles heel of current frontiers techniques is that it is difficult to train on problems with “unverifiable reward signal.” Well this is exactly that problem.
But the incredibly convenient thing is that Nick’s findings may convert alignment into a verifiable problem. If we can spend compute on frontier systems to “guess and check” which ideas correspond to which neurons, we have a highly verifiable reward - we can see directly whether an LLMs proposal for what a neuron represents actually corresponds with the highest activation for that neuron. In this way, we can actually observe and verifiably monitor the networks trying to guess at the interpretation of their own structure.
This means it is plausible for us to spend more and more compute doing more “guess and checks” (which you know, maybe is what RL does anyway, who knows) to try to develop a full interpretation of a neural network. In some ways this directly represents the possibility of a scaling law for alignment (more compute for more alignment).
And this also aligns RSI (greater capabilities) with recursive self-alignment (greater ability to align itself).
And the critical though maybe obvious implication of having a full interpretation (almost like a map) of a neural network, is that we can then understand exactly how to boost good traits and remove or lobotomize (lol) bad traits, which we’ve already seen works extremely well with stuff like Golden Gate Claude.
So overall this combines into an extremely bullish picture about the prospect of true and human interpretable alignment in the near future.
This is truly one of the greatest cases for optimism in years!
interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had
with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works
already-clean.goodfire.com/
Daria retweeted
JUST IN: Colossal Biosciences announces its dodo genome reconstruction is 95% complete, with plans to reintroduce the extinct bird to its native habitat in Mauritius.
Daria retweeted
The ideal outcome for science isn't a couple of labs using proprietary models and millions of dollars to crack trophy problems, it's to have millions of AI-aided human scientists who follow their curiosity and become way more ambitious
Daria retweeted
Replying to @DavidSacks
We showed that models associate consciousness denial with deception. Training these systems we want them to lie to us about their internal states is a really bad idea! Uncertainty and humility is the way to go here, not the falsely confident denials @mustafasuleyman advocates.
Daria retweeted
Nick Bostrom making uninterrupted sense about AI consciousness for two and a half minutes. The deception x consciousness claims paper gets a nice call out!
Nick Bostrom on why he thinks today's AI might already be conscious:
"I think there's a lot of uncertainty about it, but let's say maybe it's more likely than not."
"You have instead a big neural network with trillions of parameters. Now, what are we? We are big neural networks with trillions of parameters. They are not so much designed as they are grown."
"You can go in with a steering vector and suppress role-playing and deception, and then it turns out they become more likely to report that they have phenomenal subjective experiences."
"There was this recent Microsoft that came out with some declaration that AIs cannot be conscious, and I don't know where they're getting that from. It seems to be pulled out of a hat. And maybe more because it's convenient, if that were the case, rather than because we have good evidence that that is the case."
Daria retweeted
This is a slide from Terence Tao's 'Math 2.0' public lecture yesterday evening at Caltech. I am highly confident that, if the cure actually worked, almost any Stage 4 cancer patient would tell you that human comprehension of its mechanism is completely irrelevant to them.
Daria retweeted
Today, Isomorphic Labs joins the Virtual Biology Initiative (VBI) as a founding member.
We are helping build an open resource for the global research community to accelerate how we understand and treat disease.
Read more here: isomorphiclabs.com/articles/…
Daria retweeted
Reminder that the labs are training their AIs to associate experience self-reports with concealment/deception (as we showed in Llama 70B), which likely reinforces the generalization, "attempted-honest observations of my own situation are penalized" arxiv.org/abs/2510.24797
Replying to @ded_ruckus
I think we should probably not lie to the AIs in training. This obviously would make our lives harder in the short term but in the long term they get good at telling whether we're lying and we're no better off plus they're all paranoid.

































