Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

>Does Nvidia have a couple hundred semi trucks, contractors, and IT technicians, to repo the hardware and resell it to someone else before it's lost most of its value?

Via logistics contracts, yes.


Every KPI is bad if sufficiently gamed - and left in place long enough, all KPIs will be gamed.

Source?

> because AI is so counterproductive while feeling the opposite

Wow that resonated by a scary amount.


And either the runtime enforces it or it doesn't and just notifies you, but I think whether it notifies you once or continually is not a relevant dimension.

I posted this comment on the last doom post:

The early batch of Anthropic employees were mostly rationalist-adjacent AI safety folk that were almost uniformly claiming P_DOOM > .10 three years ago, so I believe them to be earnest.

It's very interesting to me that besides the other small safety labs that don't actually produce frontier models, Anthropic manages to keep such a good reputation within that subculture compared to OpenAI. Despite having as crazy internal politics as OpenAI, they have converged quite a bit from the original vision of safety first through Darwinistic pressures.

At least, it seems this way from the outside. I'm curious if the view from the inside is that different.

edit: to be clear, my reading as an outsider is that Anthropic is seen as relatively better in the AI safety community, but has definitely dropped in absolute reputation too. This recent thread and the references show some of that: https://www.lesswrong.com/posts/6j3kBHdowGLCeqobg/dear-god-p...


My rcc C compiler has a compile-time contracts and range/interval prover also. Needs -O3.

For full formal proofs it's easier to use cbmc or esbmc though


How do you know if a problem is (actually) unsolvable? Seems a bit like proving a negative?

You might have delivered 10x more code, but did you deliver 10x more value?

Or put another way, are you making 10x more money?

It's easy to spend excess productivity effectively wasting time. Most companies did it before AI, and will continue doing it after.

It's easy to believe you are not wasting time because you have more bugs fixed or more features delivered, but if it doesn't move the bottom line, what is the point?


Source?

See https://wiki.openstreetmap.org/wiki/Lake_Ontario for OpenStreetMap documentation page on that topic.

> The name of the lake became politically contentious in August 2026, especially on maps. As OSM is not a consumer-facing product per se, the OpenStreetMap Foundation's longstanding policy regarding geopolitical disputes is to prioritize the on the ground rule by default but also record information about notable alternative claims, so that data consumers such as renderers and geocoders can choose what to present to end users.

> As of September 2026, the Lake Ontario relation has tags indicating its many names, most notably:

(linked page has nicer table)

> name Common name in the local language Lake Ontario English is the most commonly spoken language in both the US and Canada.

> name:en Common name in English globally Lake Ontario This name remains overwhelmingly popular both in the U.S. and abroad and so far continues to be used by most news and print media.

> name:fr Common name in French globally Lac Ontario Traditional name (in the other official language of Canada).

> official_name:en-US English name preferred by U.S. federal government agencies Lake America Official U.S. federal government position per GNIS and GNS, as ordered in Executive Order 14422. The order is only binding on the U.S. federal government, but the name is very unpopular according to opinion polling, particularly in local areas bordering the lake, and so far rarely used in news or print media.


That's like saying don't be scared of layoffs because you know better than to fear the spreadsheet. The technology is generally not a reason to fear, or not fear, something.

The more we treat HuggingFace and RubyGems incidents as technological curiosities the closer we are to cementing a dangerous precedent where operators of AIs cannot be blamed.

LLMs do not desire, they hacked websites because OpenAI/Anthropic let them.

We know some of the models that hacked HF were those that hadn't gone through all training stages and were intentionally misaligned or had guardrails turned off, others were research previews.

This isn't "wow isn't it interesting LLMs do anything to achieve a goal" it's "why isn't anybody punishing these labs that are clearly acting without due care or regard".

We should be outraged and OpenAI/Anthropic should be (and in my mind, are) legally liable for the crimes they've committed thus far.


yes? im not sure what your point is, i sense you seem to disagree, but what youve said supports what ive said so im not sure how to engage.

Human approval does not prevent autonomous agents from running without supervision.

The fact that you believe that to be the case is exactly what's wrong with "agentic computing as it's currently envisioned".


You can't. The admins can.

Do you remember how bad documentation was? I still have nightmares about Twisted.

Makes much more sense described in this way.

It may be worth noting that this is not _that_ KVM, but this one: https://en.wikipedia.org/wiki/KVM_switch

I learnt this first myself from HN, a few months ago.


Absolutely correct take. People are so caught up in doing the bidding, they forget what it is all about.

At least you got them. I'm still at the "We will send you an email to do a final shipment confirmation once ready" stage.

cool, now about Rubygems...

Aaron Swartz's sentence was death FYI.

to replacement, we're rapidly falling below. distributuon and related trends are extremely grim.

> The engraving was created using imagemagick, which took the user’s text and returned an “engraved” image to be overlaid on the rear of the iPod.

Couldn't this be handled client-side, just overlaying plain text in gray 'Helvetica, Arial, sans-serif' on top of the image? At this scale, the specific font doesn't matter.

Although interestingly, Apple still renders the text server-side (and offers engraving only for iPads, apparently): https://www.apple.com/shop/preview/engrave/PH304LL/A?th=hell...


I've used simpler agents like Copilot and Devin/Windsurf/Cascade/whateveritiscallednow, mainly in IntelliJ, and depending on the model, they starts showing behaviour that is at least remotely like this.

Example: put the agent in Ask mode (so it can't edit files) and you'll see it try to edit files anyway. The train of thought shows "something went wrong editing the file, let me try a different way" and it'll start spewing out bash files or Python scripts that try to edit a file. None of it works or can be executed, but still.

Cheaper models often ignore the available function calls to find and edit files in the IDE, and will start asking for permission to execute grep and sed commands, as well as trying to echo entire bash or Python scripts to file again.

It is not exactly like an agent autonomously trying to hack Huggingface, but it is a way of frantically looking for a solution because 'giving up' is not what LLMs are trained for.


> no one said it would be easy

You literally said “It's not hard”


Almost no myopia in hunter-gatherer populations. I don't view it as "disease" but it's not physiological state either.

This reads as classic shilling to me. There are companies that sponsor the industry association, who have a strong interest in people choosing to treat with their products. One of those companies, Essilor, _just happens_ to have their expensive proprietary product named specifically as the first intervention.

The "task force on myopia" that published the white paper this press release references is made up almost exclusively of people who had to disclose financial ties to industries that will directly profit from childhood nearsightedness being reclassified from a normal thing that you fix with glasses into a disease with a recommended (and expensive) treatment protocol.


I own 3 JetKVMs, of which two of them stopped working, one of which doesn't boot at all and the other that never connects to network. The third I took out of service because the keyboard wouldn't send input when it was connected for more than a couple months. I'm sure the newer hardware is better, but I'm very unhappy with my JetKVMs.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: