TOMDM

You’re unable to view this Post because this account owner limits who can view their Posts.

12gpm

I've seen a lot of people claiming that these recent examples are just trivial combinations of existing prior work.

I'm yet to see anyone cite the prior work. If it's true I think some credit is due for the authors Fable/Sol are branching from.

You're not running an Opus 4.8/GPT 5.6 tier model on $5k of hardware at a useful Tok/s

I can delete cookies to obscure usage history

No, the people growing their food built modern civilization.

(Or millions of disconnected stakeholders with different incentives collectively built modern civilization, but who wants to put that on a bumper sticker)

PostgreSQL WAL-RUS, no relation to PostgreSQL WALRUS https://github.com/supabase/walrus

My intuition is that there would be a fairly stable base load, but doing something like switching on a new training run of a frontier model would be incredibly spiky, thousands of GPUs going from somewhat idle to 100% in seconds.

Would demanding that large spikey users of energy like data centers implement some sort of demand ramping/isolation from the grid in the form of a massive capacitor bank or flywheel generator between them and the grid help reduce the risk here?

The article caters to the majority of hn users that use a browser that automatically translates the article. Chrome on Android certainly did.

You are the minority demanding to be catered to.

AI startups taking unprofitable risky ventures in search of growth opportunity and future returns makes sense to me.

Maybe most of them or all of them lose on their bets, but there's potential for a future where revenue grows beyond the immense capex and research investments.

Oracle though... Immensely risky capex to service a startup industry with what will soon be a commodity...

From the article

As we noted above, the bottleneck in fixing bugs like these is the human capacity to triage, report, and design and deploy patches for them.

...

To begin, we’ve released Claude Security in public beta for Claude Enterprise customers. It’s a tool that helps teams scan their codebases for vulnerabilities, and which can generate proposed fixes for them. In the three weeks since launch, Claude Opus 4.7 has been used to patch over 2,100 vulnerabilities. (This is faster than the open-source patching described above in large part because enterprises are fixing their own code, whereas open-source fixes usually require volunteer maintainers who work through coordinated disclosure.)

Your critique of the article would likely land much better if you engaged with it.

AI is polling as one of the most disliked topics in the USA right now. More hated than ICE.

I don't think your source substantiates that.

From your source:

ICE

Somewhat negative: 9%

Very negative: 47%

AI

Somewhat negative: 24%

Very negative: 22%

As long as you're not bound on parallelism or bandwidth then it's "free", but if you're constrained on either resource then your lighter predictor model just needs to save you more cycles than it congests on average.

Are the human mathematicians who wanted to solve this problem just too stupid to brute force for 80 minutes?

Waymo has cars that drive themselves and are dramatically safer than people in most conditions and yet they're only in select cities.

Do you just think Google hates money, or does this only work for hover cars

I have the sudden urge to frame some wallpaper.

Secretary of War thank you, no dead naming glorious leaders very fine people

The irony in your comment is astounding.

The department of education for one.

It's been crippled essentially without anything being passed by Congress

The issue isn't that they can't be stopped, it's that they don't fear that they will be.

If you or I knowingly and flagrantly break the law, our understanding isn't that we'll be stopped and nothing else, it's that there will be punishment, that justice will be done.

How are we at a point now where we all know the administration is breaking the law. That they know they are breaking the law. That they can be stopped months or years later, but there is no justice. Not even the hint or thought of justice.

These people should fear the consequences of their actions, but at the moment there are none.

It's mandated by law how tariffs may be levied and how states may police themselves and how they may independently conduct their elections.

If you think law will stop this administration then you're not paying enough attention.

We're likely to see Ukraine some of their own domestic anti drone capability for Gulf State Patriots.

Gulf States trading dollar for dollar with Ukraine on Patriots and anti drone capability is likely to leave both parties very happy.

How reachable are the agents with this exposure?

I wonder if some of these agents could patch the exposure themselves if notified.

Why just make up a definition?

From the oxford dictionary:

Noun: speak to or treat with disrespect or scornful abuse.

Verb: a disrespectful or scornfully abusive remark or act.

Note the lack of personhood in those definitions. I can insult an object, event, person, corporation or even an idea.

Why get yourself twisted like this?

They can do a bad thing, and then you can make fun of it with an insult.

Own it, the insult is warranted, why hide and pretend it's not an insult.

If Microsoft is consistently shipping slop, then they deserve insults over it; not every "bad" thing is always unwarranted. Locking someone in a box is "bad", prison is a necessary thing that benefits society. Insults are "bad" and sometimes warranted.

You can argue that banning insults is a bad look, bad move, that the insult is warranted or whatever, but are you really going to die on the hill that calling the company Microslop isn't insulting?

Then you know and understand nothing.

A company being asked to violate their virtues refuses, and then communicates that to reestablish their commitment to said virtues?

Tell me more about what they should do if a virtue signal in such a situation is a nothing statement.

Anthropic doesn't get to make that call though, if they tried the result would actually be:

8 AIs running on 8 machines each with 10 million GPUs

AND

2 million AIs running on 2 million machines, each with 10 GPU's

If every lab joined them, we can get to a distributed scenario, but it's a coordination problem where if you take a principled stance without actually forcing the coordination you end up in the worst of both worlds, not closer to the better one.