benterix

OTOH... We know these plans are heavily subsidized, and we're happy to use them to the limits. At some point they will run out of investor money and will have to readjust. Do you think people using the existing plans will switch to per-API-call payments? Or, how many of them will?

Given enormous competition and good progress on the part of the Chinese and open models, I'd say let's enjoy the situation while it lasts. It's like chap Uber drives a decade ago subsidized by Saudi money. I have no sympathy neither for sama/amodei nor people who invest in their companies, and happy to use their resources in this crazy show where everybody pretends something and pushes their own agenda.

Yeah LLMs/GenAI is the current hype but it's just a tiny fragment of the general landscape.

> the US and its allies

Which allies?

Well, you can start with this:

https://www.theguardian.com/us-news/2026/may/11/musk-v-opena...

I'd say it's almost impossible at that point. Specifically, Altman said so many lies in the past that people stopped believing anything he says.

I think the core of this distrust is the fact that these companies positioned themselves against humanity from the start by saying people will lose most jobs etc. Not only it didn't happen, but many people feel several aspects of their lives got worse because of LLMs, in spite of obvious advantages. So the distrust and reluctance are real.

> People think otherwise with AI partly because Anthropic kept telling us that they didn't have to write code or review code any more for most of their work. Their agent swarms just comb through their github, slack and wikis to figure out what to do next, and another swarm of agents just review, test, merge, deploy, A/B test, and revert the code. Boris alone merged nearly 300 PRs in the past week (or two?).

Apart from many other issues with this, heavily subsidized subscription plans won't last forever, and if you start burning your own money on tokens in this way, you'll soon realize it's terribly inefficient.

While you're probably right, reading your comment from a consumer perspective it makes me think how much we've normalized bait & switch.

As it happens in large orgs, with mixed results. The biggest irony being the whole Transformer architecture being actually conceived at Google, only to be implemented as a product/service by another company.

Well, many people don't have very warm feelings for American LLM providers so they don't care. (Which matters because, at least anecdotally, they do care when buying a new car.)

But you're mentioning several things that predate the current LLM craze and belong to the ML domain. These mostly benefit from GPUs but often have much lower hardware requirements. I'm talking specifically about the moat of LLM providers.

> The AI companies will want to control what's possible and find new things to do that "need" their services.

That's correct. The problem is they have smart people, tons of money, and several years to figure that out, and the best thing they can come up is a coding agent.

I wouldn't say "completely implode", too much money was poured int it, but it's clear we're heading in that direction. You get a model that is "good enough", plus privacy, plus savings in the long term.

Paradoxically, the better results we get from general harness of coding agents, the less moat Claude and co. get. It's unbelievably how fast some open models outpaced frontier models of just a few months ago.

I'm torn these days. The subject matter is interesting, but the LLM writing cliches just make it unreadable, I gave up after half a minute.

> There will be businesses no human can comprehend or manage.

I believe we're getting close to that in pure math.

> You can’t imagine writing software without Claude Code anymore.

I know it's a hyperbole but repeat a lie often enough...

I believe you are explaining very basic things to an LLM.

I believe by now we know exactly what it's good at and what it's terrible at.

The problem is that our CEO's fear of the future that pushes them to peculiar decisions that objectively make no sense (cf the infamous discussion of the Microsoft employee on Github that couldn't force its agent to do the proper thing).

It's not the first time I witness this kind of discrepancy and probably not the last, I just learned to adapt to it.

Thanks, that's very useful. I find people's small individual tests more important than the usual benchmarks that tend to be gamed by every single lab.

I'm very happy to pay high taxes in Europe, and enjoy everything that they bring.

My first reaction normally would be "I'm sorry for all involved".

But... On second thought, I'd like say, "Good luck and I hope this ship drowns down".

Call me cynical but for me these are mostly pricing changes, the change in quality is imperceptible. I believe after a few iterations we will be closer to the real cost.

> 01. Removable Webcam With its easy-to-disconnect magnetic connector, you can simply unplug the webcam whenever you want to ensure that no one can access it.

What about the microphone though?

The camera issue has been solved years ago by a simple analog hack of physically obscuring its field of view, with some business units having a physical switch built-in.

The same is much more difficult for a microphone, hence the appeals of privacy-conscious folks about it, mostly unanswered.

But if Republicans are Zionists, and Democrats are too, what hope is there for peace in the Middle East?

I can understand that Israel's long-time strategy is to keep all their neighborhood in a state of permanent mess so that nobody is strong enough to be an existential threat. But after almost a century, it's clear this is not working.

Paradoxically they are so stupid they thought they are actually helping Orban not realizing Europeans consider them idiots.

People who say there is no free speech in Europe have never lived in an authoritarian country.

Same with me. I found a new hobby: reading pre-LLM HN. It turns out I missed so many interesting projects and discussions. Some are a bit funny in hindsight, some are inspiring.

At the same time, the current version of HN is still usable, you just need to mentally filter LLM-related stuff. It was similar with cryptocurrencies TBH.

OK so I see someone at Microsoft decided that their strategic decisions do have impact on their income so they are implementing a crisis management strategy.

The way I read this, it sounds like "We're not giving up on disabling local accounts, telemetry, ads in the OS and other things most people complain. However, we're willing to make some concessions in how aggressively the updates are pushed on users and a few minor points or things we planned to implement at some point anyway but didn't consider that important."

Prove me wrong.

It's not exactly rosy as you say, but the leaders are increasingly open with Trump as to what they actually think (see Merz's latest comments) and he uses this as an excuse to realize what they planned a few years ago. There is zero chance that any Europen state starts bombing Iran.

OTOH, while refusing to do wrong is already something, there are limits to what they can actively start doing - and concrete consequences if they start doing something Trump doesn't like. He's highly unstable and can decide to take a decision that can be perceived as crazy, and harmful to the USA, but at the same time harmful to the ex-ally who decided to stand up to him. So the EU leaders basically do what they can to just wait the crazy guy out.

A specification, whether formal or less formal, is very different from the source code.

The word "expected" is a weasel word in this context, especially given how muck backlash MS has received. I'd expect a link to a study where users say: "I'd like to have an LLM integrated with my operating system and my browser" and how it changes over time. Then you can seriously argue for "increasingly expected".