sigmar

lol. Solid idea. Going to add an email signature with "Emailing me is billed at the following rates: $20k/M token input, $100k/M token output"

That video is insane. Is the guy so arrogant that he couldn't conceive of being wrong? How could he possibly continue to write a ticket after this interaction?

>But lately I’ve been thinking if it is just a class issue? This cohort of people likely have a cushion that softens the concussive blows they are doling out right now. They perhaps have the luxury of a somewhat functioning government and a social safety net that they are witness to in all walks of life. Over half the world does not. Science and technology, I feel, has always had a certain apathy towards the plight the people at the bottom rungs.

In the data I've seen, the US and European countries have a more negative view of AI than China and developing countries. Doesn't that fly in the face of the premise here that only people that have economic security support AI?

https://www.visualcapitalist.com/survey-how-21-countries-vie...

https://www.ipsos.com/en/conflicting-global-perceptions-arou...

https://www.mexc.com/news/161986

It was Mythos

>Our engineers, working together with Mythos Preview, built a working exploit in five days.

https://news.ycombinator.com/item?id=48139219

Lots of interesting data in this. I'm surprised China is at 16.4%, I thought that would be higher

UK's 'indefinite leave to remain' (closest thing to a green card) requires that you apply from within the UK. https://www.gov.uk/long-residence/apply-to-settle

>it has to do with being battle tested over time. If a team of humans had rewritten it in a week, I wouldn't trust or use it either.

"it was made in a week" gets repeated a lot on HN, but the PR wasn't a release. They've been working on the rust rewrite for more than a month and it hasn't shipped.

Plateaued? Lol. Based on what? Pg 18 and 45 on that link are not showing a plateau.

"available to qualifying customers’ security teams on request." Seems they're already expanding access.

oooph, been a while and I don't remember any of those details. Good excuse for me to re-read it!

this is cool. My nit pick- aren't the petrova lines curving along the wrong plane? For example, in our solar system the line should start at the sun and should be pointed at where Venus used to be, but then curve towards where Venus is now (until gets to venus). Since the astrophage will course-correct over their journey and stay in the same plane as Venus' orbit.

The blog post links a pdf that OpenAI put together of nine mathematicians that commented on the proof. Each is quite brief and written in accessible terms (or more accessible terms, at least). https://cdn.openai.com/pdf/74c24085-19b0-4534-9c90-465b8e29a...

clearly the employees that tweet the most must also have the best contributions to R&D... </sarcasm>

>OpenAI’s hiring recently has been much stronger, the guys have actual taste.

Can you cite specifics? "I won't speak bad about someone, but also won't speak good about others" resulted in a comment that seems to contribute nothing

uhhhh isn't the most logical explanation that the "exceeding" number is intentionally vague? Just like literally every other number in the CFO's declaration ("spent over $10 billion"... "raise more than $60 billion").

Helps to look at source documents. Declarations like this are written to be literally true, but not precise enough to reveal more information than they have to.

I always assume that people making those comments on HN are trying to convince others to switch to their model. Surely no one actually believes their friend circle is a representative sample of the hundreds of millions of people that use these LLMs?

posting my read (since it differs so much from the others')- there's a 'holy war' being waged by people that think LLMs shouldn't do full rewrites of software. There are various reasons people think this (think LLMs are parrots that make slop and are incapable of writing good code, have environmental concerns, or are angry that software licenses can be circumvented). I call it a 'holy war' because I think most see our current trajectory as a bit inevitable and have a strong urge to proselytize their views and chide maintainers that use LLMs in ways they don't like.

Very similar angry comments happened with the discussions of the Chardet rewrite, next.js/vinext, and JSONata/gnata if you want to look at this in context.

Really? Maybe my brain is broken but I saw lots of forced analogies and LLM habits. Like where it does an em dash and a short list:

"I was getting steak ads everywhere—Instagram, news sites, even in places where ads should not be allowed to have that much audacity." ... "everything simply—salt and pepper, no fancy rubs, no sauce safety net." ... "with twine—very farmer’s market cosplay, very “trust me, I’m artisanal.”"

ebay started attempting consignment more than ten years ago, but I think lately they only do it for luxury items. Which makes sense to me as a lot of people would just send in junk. https://pages.ebay.com/ebay-consignment/

Neat research. I find figure 11 interesting. The models behave so differently there.

imo the benchmark should be named Can_It_Pull_a_CharDet_Bench

Scrambling? Seems to me xAI built too much capacity (for what they can use in 2026). Does that mean OpenAI built the right amount? I don't see how this proves that just because we see one AI company willing to sell compute. We don't even know the terms/pricing.

I wonder if they hadn't decided to call it a drafter when they named the files and were using assistant internally? google being google...

for anyone wanting a glossary to explain the naming scheme here:

E4B = 4B effective parameters (using per-layer embeddings)

E2B = 2B (like above)

it = instruction tuned (rlhf and all that jazz)

assistant = Multi-token drafters (the new 2x speed up)

>try them directly on Google AI Edge Gallery for Android or iOS.

I'm not seeing any update to the app on my android phone... maybe later today?

>We’ve published an in-depth technical explainer

I was expected a pdf link, but this goes to a brief article on twitter/X. lol, okay...

Moves like this make me wonder- What chance is there that these models are nationalized in the near future? What will happen to the investors/economy in such a scenario?

I found it here: https://capulcu.noblogs.org/english/

What specifically is the goal of the pre-release review? Just to patch government systems first? Seems like the government was banning internal use of anthropic's models 2 months ago and now wants exclusive access for some amount of time. Clown show...

Study design, data collection, analysis, and peer review take time. O1 came out a little over 1.5 years ago

>The Guardian needs to raise their bar on what to report and how to give readers full context

Should they not report on peer reviewed articles published in Science? or only report published articles that fit your priors?

>We contend that the amount of time remaining before the arrival of CRQCs still exceeds the amount of time needed to migrate public blockchains to PQC, though the margin for error is increasingly narrow. Therefore, we have offered updated resource estimates for quantum attacks on blockchain cryptography together with an analysis of vulnerabilities and mitigations in order to urge all vulnerable cryptocurrency communities to begin PQC transition immediately while its timely completion is still the likely prospect.

they really couldn't be shouting "mitigate now or never" any louder. I'm curious how they arrived at the efficiency improvements, but perhaps any mention of that would be similar to releasing the circuit.