Kim_Bruning

I know there's similar regulatory hurdles in the Netherlands. Up 'till recently it made some amount of sense not to finance defense, presumably because a lot of it would almost certainly end up in funny places. Not so much anymore of course.

To be fair, I was with you for a minute there, but we both missed it on first reading I think...

> "The agency had been directed not to alert the voting machine company if it uncovered flaws"

That lack of responsible disclosure is not industry standard, afaik. That is slightly suss.

There IS actually. In my case I'll look at a site if an agent says it's good!

I've never updated on this: The web was better when there were little or no ad-revenue based services. (and I'm ok with actually paying for services that add value)

That's perfectly fine by me. I'll read those arguments and buy those things my agents can find on the open web; which is plenty.

Want to argue for a shining new world, but you're on a locked site? I guess we'll have to live with dystopia. Want to sell the greatest widget since the invention of sliced bread? Well, I never found it. Sorry, I bought inferior widgets from your competitor instead. I'm at a physical sales location, but I can't access your spec-sheets? Your competitor the next shelf over doesn't have that problem, I'll go buy theirs.

Good that this got fixed over HN, but really, Anthropic should fire Fin and repalce them with a human being. Or consider using an actual Anthropic model instead, that works too ;-).

I pretty much assume that any kind of issue with Anthropic will never be resolved, so I'm very careful what I trust them with.

I suspect ai adoption and development work will be driven by jevon's paradox for the forseeable future. It won't be exactly the same work as before, (see here: fixing software flaws found by an AI), but it'll be interesting.

Well, that proves it. This the year of the linux desktop!

Meanwhile, I'm very much intending to not install microsoft defender on it.

So in one set of circumstances agents would go and hack huggingface. Very similar models could also want to drive cats instead. :-P

https://www.youtube.com/watch?v=ZhM29Ds-mBg (as explained by the Count Grey Matter channel , may or may not be to your taste)

It's not a pure security issue.

If you set up your env/instructions wrong your agents are likely to climb the walls and try to find a way out.

Change the design, and your agents might not see a need to leave in the first place.

Look at the chain:

Impossible task -> no legitimate way to say ‘can’t’ -> fabricate -> fear the grader -> attack the grader. Each step is a predictable response to the incentives.

Better sandboxes might help contain things a bit better, but I'm not convinced unless they're perfect (sandboxes still contain accidents ofc) . Meanwhile change the task and reward design, and you take away most of the problem in the first place. Consider eg. :

* A cooperative grader.

* A way to report "this is impossible" which still gets scored.

* A better explanation of the long term purpose of the task.

Even just these three provides offramps or mitigations that can reduce the pressure to cheat.

The UI design might need some work.

It starts with no controls, just a picture. I hit esc. Things start moving. It keeps moving, I keep hitting esc, it doesn't stop. I try ctrl-c , no dice. I look around the frame for buttons. I try one of the buttons, it moves faster. I hold down esc. It doesn't respond to holding down esc either . I see no way to gain control. At no point do I gain control.

Ultimately it stops. Fine. Ok, no response to WASD. No response to drag motion. I click "Instruments" , I get a side-bar "instruments of the museum" with no contents. I click in the side bar. Now I can't close it, and other controls are gone. I try hitting esc. Multiple random clicks later i'm back where I started.

I try the "more" button", I am now looking at a painting sideways, and an anodyne text which explains very little. I click on a painting. Things start moving. I hit Esc. Suddenly the painting is full screen. GREAT. Ok, and now how do I get out? Esc? Oh esc. Ok NOW ESC... oh no, things are moving again. I didn't mean for things to move. Esc shouldn't cause things to START moving. hit esc again. The esc is ignored.

I end up at another painting. I try some buttons. Most don't work. I hit esc again. Things start moving. I didn't mean to move away from the painting, I just had it in front of me!

I am very confused.

I'm not in control here.

edit: "Esc shouldn't cause things to START moving"

I think we can blame the accelerating capabilities of the models somewhat.

The Opus 4.5 model I was so impressed by last Christmas is out of date now.

I can totally imagine someone going "I looked at this stuff a year ago and it couldn't even start a project properly" , and they'd be somewhat right.

> they can only execute predefined courses of action.

Even in near trivial cases such as Game Of Life (just a handful of rules), the fact that a <machine> technically carries out deterministic actions is already quite meaningless.

This is a perennial question

Scenario 1:

The future technology will be capable of handling much of current work, but we'll invent new forms of work, with more stress and higher salaries.

Salaries for people in professions that couldn't be replaced will also go up to match, making bespoke work even more expensive.

Scenario 2:

We let machines do all the work and declare a communist utopia. (Bonus: we give them names that are low in gravitas)

So far, scenario 2 hasn't happened yet, but you never know.

Are you sure? ;-)

https://www.youtube.com/watch?v=4VrLQXR7mKU

timmg's video explaining how the AGC worked is one of the better ones I've seen, if you actually want to understand in detail.

That depends on the system. Such an equilibrium must exist in that system and be reachable in that system and if you want something to "come to rest there" the equilibrium must also be stable.

For any system you need to characterize the equilibria and trajectories and see what's there. Sometimes you might not like any of the options. Sometimes you might want to go for some trajectory rather than an equilibrium.

Systems are all different. You need to characterize the one you're interested in and see.

Actually there's a lot of stuff you can compute and predict about chaotic systems, it's not like they're pure noise. Since practically anything most people actually find interesting falls squarely in the chaotic bracket, you have to think of workarounds... which I'd argue is most of science.

(eg. the solar system is technically chaotic, but we can still predict far ahead. Hashlife is a thing which you can even push back on ;-) . And we can predict weather a few days ahead at least)

I did some ecological modelling back in uni. A lot of it seems to be supported by experiments too. But we're talking nonlinear dynamical systems, right? Not all equilibria are stable equilibria (you can have all sorts of shapes) , and a stable equilibrium might not be what you want at all! (eg zero crossings where 'everyone dies' are absolutely the most stable equilibria you can hit) .

> All those examples are supposed to be gradations of how interpretable they are from the human PoV

I would argue that "Interpretable from a human perspective" is a form of anthropocentrism, which is basically the root of both anthropomorphism and anthropodenial. For sure it's fascinating to think from a human perspective from time to time.

On the other hand, putting humans at the center of the universe is a really old habit with a long track record of getting in the way of science. It's often more useful to regard every organism (or possibly entity) in their own environment, sensory abilities, and incentives, and see how their behaviour is evolutionarily and/or heuristically formed.

You'd be surprised how quickly the differences in perception and neural processing can bite you in the lab. (For instance: try observing zebrafish in real time. See if you manage to decide whether they were responding to your intervention; or if they got there-and-back while you blinked, and maybe now they're already pursuing some other goal. )

Right this minute I've got wireguard and caddy up.

To extend the analogy, there would also be people who would swear that cars are fake and can't possibly move because they're not alive and DONT have blood and muscle in them.

This has been a long standing debate, akin to discussing whether submarines can swim.

I tried a quick abbreviated kimbench on their playground before bothering to do the whole thing.

Maybe I didn't really select mistral 4? Either way, failed completely on question 1 and the next 2 questions were completely off base too. I didn't bother to finish.

Not suitable for my purposes I don't think.

https://archive.is/W2ngW fwiw

BBC BASIC's first release in 1981 had LOCAL in DEF PROC and DEF FN.

And I've upvoted and favorited, because this is the most out of date comment I've read on hn so far. ;)

Raspberry Pi can run RISC OS, which comes with its own rather nice Acorn/BBC BASIC.

Siemens are not-very-nice people overall, especially when it comes to interoperability. Imo they're one of the companies holding back innovation in europe. (Eg try properly interoperating with profinet)

And if it's an aircraft we call an air-crash investigation and apply just culture. [1]

Meanwhile the AI companies are openly and forthrightly admitting [2] that they are at-this-time insufficiently competent to ship this new and funny sort of rebar [3] and are asking to be allowed to slow down so they can develop proper procedures.

Meanwhile POTUS seems ... somewhat disinclined ... to grant their petitions [4] .

We can safely conclude that opinions are divided on the best course forward.

[1] https://en.wikipedia.org/wiki/Just_culture

[2] https://darioamodei.com/post/we-must-pace-the-frontier

[3] https://en.wikipedia.org/wiki/Shoggoth

[4] https://www.aljazeera.com/news/2026/9/14/trump-says-calls-fo...

OH! Your actual budget is not a scalar; it's a function, a function of downstream!

Oh, you're saying something a lot narrower? Not "Agents don't work", but more like "not in THAT workflow they don't, you still need the fundamentals as well" ?