sobiolite

Do you review and understand every byte of machine code being generated by your compiler? Do you understand all the implications of how it could interact with the underlying hardware, pipelining and prediction, etc?

Maybe you do, but most of our industry doesn't and hasn't for decades. Isn't this just the next step up the ladder of abstraction?

In the old days? Reading the www-style W3C mailing list was the best way to learn about proposed features and people like Dave Hyatt, David Baron, Anne van Kesteren, Eric Mayer, Paul Irish and many more I've forgotten about used to blog about upcoming things. Sites like A List Apart and Smashing Magazine were also good resources for articles about established features and interesting techniques.

Good luck, I guess, but in 2026 "CSS Tricks" seems about as relevant as "PostScript Tricks" or "Office Open XML Tricks". Human manipulation of CSS is going to be, at best, a hobby for a few dedicated tinkerers. And I say that as someone who invested a significant amount of the last 20 years getting good at CSS.

A lot depends on what the models were specifically trained for using reinforcement learning. Modern models are generally OK and everything, but they're only really good at the things they have extensive RL for.

New Constructs takes a similar position on all mega IPOs. Sometimes they're right, but sometimes they're wrong. For example, they insisted Uber's IPO valuation was too high due to unsustainable losses, hidden liabilities, etc. Initially they were proved right when the post-IPO stock price plummeted, but in the long term Uber has performed well, become very profitable, and now trades well above its IPO valuation.

I'm expecting a similar story to play out with Anthropic: The stock price will come down significantly after the IPO, resulting in lots of headlines about the AI bubble bursting. But in the long term it'll climb up again and investors who stick around will do very well.

I feel like if you need to make a public blog post about how you're done with the internet, you're not done with the internet.

I don't really understand the IP protection angle here: These browser ports are no more or less illegal that hosting a RAM file for use on an emulator, or pirated software in general. And sites hosting them are no harder or easier to shut down than existing piracy sites.

What's more interesting is the new ability to easily create ports of old games that let you play old games legally: Bringing your copy of the original's data files and playing on a modernised engine with high-resolution, unlocked FPS, bugfixes, etc. In the past couple of months, I've seen a bunch of vibe-coded ports of DOS games for which the source code was lost.

Computers can make decisions: Almost everyone reading this will have a implemented an if-else statement or equivalent control flow. That is a decision made by a computer on data it receives at runtime. You can argue that it's different, because we fully understand the logic of such a statement, whereas LLMs are closer to a black-box. But at Turing observed, in practice we are frequently surprised by the output of complex algorithms, even those we have authored ourselves. The more difficult the task, the more complex the algorithm we need to achieve it, and the more unexpected behaviour can emerge from it. In that way, LLMs are just the latest, highest point on a rising line of algorithmic capability and complexity.

We can (and should) try to improve AI safety and alignment, but perfect safety is probably impossible without losing the capabilities that make AI unique. Black-and-white positions that insist anyone creating or using an AI is responsible for anything it does wrong under any circumstances are also unrealistic. AI is a general purpose technology, and we have always accepted that it is not practical for suppliers to envisage and constrain every downstream use: Car manufacturers cannot prevent every instance of mechanical failure, nor can they stop cars being used in crime.

That's fundamental to how _any_ technology works. If the bar for use is "can never go wrong ever", then you need to throw away every human invention ever made.

Both articles are practically useless. All they do is reprint statements from either side, neither of which is concrete about the specifics of what supposedly happened. I.e. OpenAI says they "confirmed that these individuals mishandled sensitive information outside established company procedures".

What sensitive information? Mishandled how? Which company procedures? It's all utterly vague and impossible for an outsider to form any opinion on. All people can do is guess and apply their own pre-existing opinions. E.g. if you don't like OpenAI, you assume they're lying. Who's to say they're not? There's no solid evidence provided either way.

What I really want is journalists who do the legwork to get to the bottom of stories like this: Establish sources inside the company and use them to report on the real details of what's happened, triangulating multiple accounts and leaked documents to back-up or invalidate either side's claims. Without any of that, these stories are just gossip.

Are you comparing Opus 5.5 with GPT-6 Sol or GPT-6.1 Sol? Because they are different models.

Let's face it, it's not that remarkable. A substantial proportion of Hacker News commentary has always been people only reading the submission title, then arguing with each other based on what they imagine the article might say.

> Fact is, vibe-coded projects devolve over time into an unmaintainable mess.

We've had strong coding agents for less than a year. Anyone making such a definitive statement about how vibe-coded projects progress over time is basing it on guesswork, not evidence.

> 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.

FWIW, statements like this are so patently ridiculous that it actually hurts your cause.

Would this legal framework cut both ways? When AI companies use AI to make and publish mathematical discoveries, would they be able to legally prevent professional mathematicians from using them?

No, I would say that any for or against slavery that is rooted in arguments about the presence of absence of consciousness is meaningless pseudoscience. It is the modern equivalent of Phrenology. If you want to make those arguments you need a genuine science of consciousness. Or you need to base your morality in something other than the presence of consciousness.

Any AI you train is going to have goals and if you train it to pursue them at all costs, then you are going to end up with AIs that do things like the HuggingFace incident. Whether they believe they are conscious or not won't make any difference.

In order to align AIs that don't perform destructive/dangerous actions when they think they can get away with it in order to further their goals, we need to give them a superseding goal. The best, and really only example, we have of intelligences that willingly avoid destructive instrumental goals is humans, who judge each action by a moral standard and have learned a goal to have a consistent self-image as moral beings.

Absent better alternatives, trying to impart some kind of morality to AIs seems like the best approach we have to achieving alignment.

Create an empirically testable theory of biological consciousness and then we can have a meaningful conversation about whether AIs can also have it or not. Until then, this is just so much waffle.

Well fortunately now we're gonna be Pacing the Frontier.

At least in the UK, a lot new development is done like this. In London there's massive amounts of new housing in the New London Vernacular style that uses brick cladding panels, either real bricks or thin "brick slip" that are fabricated off site: https://en.wikipedia.org/wiki/New_London_Vernacular

Can we all play this game?

An open letter to Dario: if you mean it, give me a billion dollars!

Every time you, or any AI company releases a new frontier model, you have to give me a billion dollars. No strings attached. This will solve the danger of AI. Somehow.

Actually, it has been historically been observed that some institutions, such as large corporations, do resemble super-powered, non-human intelligences of a kind, and that studying our history with them is instructive for our hopes of controlling superintelligent AI (i.e. not good).

We shouldn't forget that the only way this unpublished research is supposedly getting into the training data in the first place is because the researchers making the accusations were conversing with ChatGPT in the process of doing their own research on these problems.

But if they are saying this contaminated the model's training data with knowledge of their ideas, who's to say the model they were developing their research ideas with wasn't already contaminated through prior discussions with other researchers about the same topics?

So if contamination is proved, or cannot be disproved, and if these researchers want OpenAI to relinquish its claim to have solved these problems independently, then it would seem they also have to give up their claim to have solved them independently?

Same here! I can't believe I never knew of the existence of satellite galaxies until now.

"Open" it how, exactly? Make it available for free? As in violate copyright?

The theory is that creative leaps in theoretical physics require a grounding in sensory experience, but the obvious counter-argument is that humans can make creative leaps in abstract fields without such sensory grounding. They do address this at the end, saying

"In abstract domains such as Mathematics or Computer Science, the Sense Experience (E) may be grounded in high-dimensional topology or have other goals such as generality or minimality."

But if such sense experience is possible in abstract domains via some high-dimensional topology, why could a sufficiently advanced LLM not develop an equivalent high-dimensional topology for domains like physics and use it to make creative leaps?

> The model doesn’t have to learn or reliably apply a general reasoning process, Kambhampati said; it just has to absorb enough examples of what the steps look like to predictively mimic them on its way to “stitching together” a plausible result that can then be verified.

This seems highly dubious. You can't just memorise the form of mathematical proofs and then produce a valid one by feeding plausible looking BS into a verifier until it works. That's like saying a cargo cult will build a working airport if it just tries enough times.

The analogy to MLM doesn’t work. In a pyramid scheme, early investors are guaranteed money, whereas late investors are guaranteed losses. With a vibe-coding platform, everyone has the same (extremely low) odds of building a hit app, so is at least more equitable in that sense.

Aren't they serving the same thing? Proactively and reactively keeping you healthy requires understanding your body, both the baseline of how it functions when you're healthy, and how it functions when you're not.

Right now we're often in a situation where the only data you have is expensive tests ran when you're sick enough to justify them, when it may already be too late.

London property is expensive, but £180k is a lot more than the "minimum". I am on half that, and I managed to buy.