raincole

I think everyone, I mean everyone, has noticed how different GPT-6 sol is from 5.6. Just not the direction OpenAI hoped for.

I'm still confused about what this codemode is. Models have been trained to chain bash and other typical unix tools well. They're so good at that to an uncanny level. Why do we want to not utilize this ability? Is it just a permission management issue in case you don't want the model to use shell directly?

It's not a hot take at all. Every benchmark shows that.

Chinese models are not for a fraction of the price though. If you only have a budget of $20 a month, luna with chatgpt subscription actually gives you much more room than deepseek. The idea that Chinese models = cost efficiency is rather outdated.

If I follow the principle "every single line of AI-generated code has to be reviewed and understood by me," I can't even use up the $20 subscription.

For reasons, but the main one is that the author stated that they and their attorneys have seriously explored this case and reached this conclusion.

A program can only be as deterministic as the hardware it runs on. If you have a turing machine whose symbols on the tape just (truly) randomly change every step, then none of programs on this machine is "deterministic."

> meat-

There is no meat-based deterministic way to do anything yet yet. You're talking sci-fi.

It's just a typical case of XY problem[0].

X is search engine. Y is to get answers. (most) People want to get answers. If they can skip search engine part it'd be a net positive for them.

Ironically that programmers should be the ones who understand XY problem the most (after all the name is coined by a programmer to address it), not the "normies."

[0]: https://en.wikipedia.org/wiki/XY_problem

> I constantly read these amazing stories of people vibe-coding some firmware/driver that just works, and honestly I’m starting to question whether I’m reading the posts of some promotional bot.

No one uses assistance in the IDE for project like these. They're all using Claude Code (or similar harnesses) and lots of sub-agents. People who haven't seriously tried out SOTA models + their proprietary harnesses just don't know what they're missing out.

> Calculators and computers are deterministic

How do you know?

Memory bits flip randomly. It's not a super rare thing either. You and me have experienced that many times without knowing. The only reason that computers feel deterministic is that we have error-correcting code to fix that. But in the most extreme cases, when multiple bits flip together, once "deterministic" program can generate unexpected output.

So why do you trust computers? Because statistically the case is just very unlikely. Therefore if AI is statistically unlikely to make mistakes there is no reason to not trust them.

Because it's not that smart especially when you compare it to other LLMs. The top LLMs have completely change the baseline of being smart.

Because there is much less demand for deathless meat. Milk + eggs + legumes are good enough for many vegetarians. They're even cheaper than real meat in most places. On the other hand, human intelligence, by definition, can't be affordable for someone on a middle income.

> given their blog's history

All of the posts in 2026 on that blog area about AI, by the way.

AI just solved a millennium problem two weeks ago. "The pace is insane. And there is no reason to be this fast." to quote Terence Tao word by word.

HN: Well, must be a stagnant industry...

Why would he go back to the US lmao. He's foolish enough to trust the US people but not that foolish.

> The personality is bland

Sounds like a plus. Guess I will give Grok another try...

Just pi. `pi install` the packages you actually need or ask LLM to write a package for you. Keeping the harness minimal is the point of pi.

What a crazy thread lol. I really can't tell who is serious and who isn't there.

In other words, immutable != read only.

In C# ReadOnlyCollection<T> and ImmutableArray<T> are two completely different things for this exact reason.

FYI I never had LLM writing such a commit message. So as they say, skill issue.

Yeah because... AI is a tool, not an artist. In other words water is wet.

> To me if you ask to a 10 years old kid and give him a PC with Microsoft Paint on it it will you will have a much better result than any AI-generated option in the post.

Sorry, if it's a joke it doesn't land very well.

Before AI, if a 10yo did these posters for a school project, people would just tell their parents to not do the homework for their kid. These are far above what 99.9% 10yo can do.

It sounds even worse. Actually far worse. It's a project already in use and the author just nuked the whole git history?

It seems that the author has very weird and strong obsession over Yudkowsky and his HP fanfic and took the public hatred toward AI companies as an opportunity to ridicule him.

> can you use this to create legitimate terms

I don't think you can. What I get from this article is that this is not a product they're going to sell to average consumers.

Doesn't that defeat most of the benefit though?

If you think GIMP and Krita (especially Krita) can replace Photoshop in workplace I'm 100% sure you don't know what professionals pay Photoshop for and perhaps you should observe some people's working environments before criticizing Adobe's "shenanigans." Sorry for being this blunt.

I do agree that Affinity seems to get some traction, but they're practically funded by an AI company so you can expect more shenanigans, whatever it means, from them than Adobe.

I use this niche app as an example because if you only consider mainstream apps (e.g. Microsoft Excel, Adobe Photoshop) then Linux is completely off the table.

Bringing Houdini, an app whose main user base has Linux-based rendering clusters, to the discussion is unfairly favoring Linux side. I shouldn't have done that.

> you can also run a stock Ubuntu install which will be a great working & gaming workhorse that doesn't add hassle to your life.

Sure, if your job is writing backend server code. Many of us need other apps though. Things like [this](https://www.sidefx.com/forum/topic/90453) are showstopping enough for people who rely on Houdini. By the way, SideFX support told me Linux is their choice of OS and I should switch from Windows if I can.

Got'em what? It's a paragraph about someone else's agent annoying them with emails.