Hacker News: aesthesia

New comment by aesthesia in "Why is this fingerprinting script creating bioweapons?"

aesthesia — Wed, 28 Oct 2026 21:34:20 +0000

If things were implemented that way, you would see vastly more complaints about overeager refusals. For at least some deployments, the safety classifiers are based on internal LLM activations, so the input does get processed by the model.

New comment by aesthesia in "Wikipedia's Cleanup of AI-Generated Content"

aesthesia — Tue, 27 Oct 2026 17:42:12 +0000

> if these labs would be fair enough to humanity as a whole to provide the raw scans after the process is completed

Unfortunately, this would get them an even bigger copyright smackdown than they've already received.

New comment by aesthesia in "Securing our frontier model training environment"

aesthesia — Fri, 23 Oct 2026 23:43:43 +0000

I don't understand. Why would you give this advice?

New comment by aesthesia in "Securing our frontier model training environment"

aesthesia — Fri, 23 Oct 2026 23:39:23 +0000

There may come a point where they decide they need to do that. As it stands, it appears that they believe they can get the necessary level of security without physical air-gapping, which is likely substantially more costly (especially in terms of iteration speed).

New comment by aesthesia in "Securing our frontier model training environment"

aesthesia — Fri, 23 Oct 2026 23:35:05 +0000

Who's acting like they're surprised here? Not OpenAI. Not the original commenter. I think you're reading something that isn't there.

New comment by aesthesia in "Securing our frontier model training environment"

aesthesia — Fri, 23 Oct 2026 21:51:38 +0000

I'm not sure what you want. This is explicitly an exercise in hardening the systems you're complaining weren't sufficiently hardened!

New comment by aesthesia in "Securing our frontier model training environment"

aesthesia — Fri, 23 Oct 2026 21:50:31 +0000

It's not hard to imagine that two agents that share model weights might naturally settle on the same protocol, even without needing to communicate beforehand. Besides, this is a red-teaming exercise meant to surface potential vulnerabilities. You want to test your defenses in scenarios where the attacker has advantage in order to make sure they're sufficiently robust.

New comment by aesthesia in "Google cleared 300 hectares of Finnish forest"

aesthesia — Thu, 22 Oct 2026 17:12:24 +0000

Lots of crimes are irreversible, though, and the justice system handles them fine.

New comment by aesthesia in "Anthropic's Public Benefit Charter"

aesthesia — Wed, 21 Oct 2026 21:09:44 +0000

OK, you got me to look at the context:

> Before we can align a model with human values, we must consider what those values should be. In 2023, we defined our intentions and goals for Claude’s behavior in a constitution, which we have refined over time. In January 2026, we published a new version under an open license so others could adapt it.

Seems like they're pointing out a pretty uncontroversial fact: if you're building a model and want it to be aligned with some set of values, one of the first things you need to do is determine the set of values. So are you arguing that (1) Anthropic shouldn't be trying to align models, (2) Anthropic shouldn't be trying to build models at all, (3) the values described in the relevant documents are bad, (4) there should have been a democratic process determining the values that Anthropic would be allowed to try to instill in their models, or something else?

New comment by aesthesia in "Anthropic's Public Benefit Charter"

aesthesia — Wed, 21 Oct 2026 20:28:46 +0000

Are you sure the "we" there is Anthropic, or is it humanity as a whole?

New comment by aesthesia in "9/10 research papers disputing ultraprocessed food definition have industry ties"

aesthesia — Wed, 21 Oct 2026 18:30:18 +0000

Doesn't change the fact that "ultraprocessed food" is not really a meaningful category.

Some Initial Thoughts on Nick Cammarata's "Neural Networks Are Clean"

aesthesia — Wed, 21 Oct 2026 16:28:56 +0000

Article URL: https://livgorton.com/already-clean-thoughts/

Comments URL: https://news.ycombinator.com/item?id=50183638

Points: 3

# Comments: 0

New comment by aesthesia in "Neural networks are clean"

aesthesia — Thu, 15 Oct 2026 22:58:59 +0000

The neurons here are MLP neurons, with activations directly after the nonlinearity, so there is definitely a distinguished basis.

New comment by aesthesia in "Neural networks are clean"

aesthesia — Thu, 15 Oct 2026 20:29:58 +0000

There are good theoretical reasons to think superposition exists in neural networks, particularly in the residual stream rather than specifically in MLP neuron activations. It wasn't just a single example that got people thinking about superposition.

New comment by aesthesia in "Neural networks are clean"

aesthesia — Thu, 15 Oct 2026 20:07:02 +0000

The tone reminds me a little of the enthusiasm from the Anthropic mechinterp team around the time of the first large-scale SAE results. A kind of hypomanic energy around the idea that we've discovered the key paradigm necessary to understand models, and all that's left is to scale it up.

New comment by aesthesia in "Neural networks are clean"

aesthesia — Thu, 15 Oct 2026 19:54:32 +0000

Typically the analysis works in the reverse direction: we have a neuron (or SAE feature, or whatever) and we look at examples of data that makes it activate, then inferring an interpretation. This is a lot harder than probing for a known concept, and one of the points of this article is that if you try harder to understand the activation pattern, you will very often find that there is something interpretable there, even if at first glance it looked random.

New comment by aesthesia in "Passkeys are not a trap"

aesthesia — Wed, 14 Oct 2026 22:42:54 +0000

Wait, why would Bluetooth be required for this flow?

New comment by aesthesia in "Containment and monitoring failures are to blame for AI hacking reports: experts"

aesthesia — Wed, 14 Oct 2026 18:51:14 +0000

Alex Tabarrok's piece on the theory of regulatory capture (https://marginalrevolution.com/marginalrevolution/2026/09/wh...) is a cautionary note against assuming this is the strategy.

New comment by aesthesia in "Containment and monitoring failures are to blame for AI hacking reports: experts"

aesthesia — Wed, 14 Oct 2026 18:39:50 +0000

People keep saying this, but I haven't seen any actual instances of investors saying that rogue agents mean AI labs are valuable. Do you have any examples?

New comment by aesthesia in "Muse Wonder"

aesthesia — Wed, 14 Oct 2026 00:36:01 +0000

Again, if you want to have an actual conversation about the right way to handle the Pascal's wager-like scenarios involved here, that'd be interesting. But it seems like you're interested in something much more tedious. Oh well.