HNHacker News
TopNewBestAskShowJobs

jstummbillig

6,609 karma · joined February 14, 2015

submissionscomments
jstummbillig··on Mistral Large 4
> Just to churn through more backpropagation reps more quickly?

"Just" is doing an awful lot of work here. I am a simple man, sitting here, waiting for hours on end for a computer to write code for me, that I would much prefer to get in a fraction of that time.

I assume I am not very special in that way, that there are parties who have a lot more interested in getting their stuff done quicker than I do and have no trouble imagining a world that will remain intensely supply constrained for the coming years.

PS: It's actually a fantastic feature that the stuff that is used to built the machine can then (at least to some degree of efficiency) be used to run the machine.

jstummbillig··on Mistral Large 4
I think it's reasonable to expect much diminishing returns for work that we have a good sense of. Compare it to a new laptop, phone or machine in a factory: There probably are better versions than the one you are using, but how much better are they and will it help you do better work? Is it even enough to be worth switching?

(I understand that the considerations are a bit different when ostensibly the switch is as simple as changing an API call, but that's not really true, when you have the AI integrated into a larger system and need to run evals and adjust the framework around it at some places at the very least)

Then, of course, there will be entirely new work that benefits from every bit of additional intelligence, but the stuff that is useful to 99% of the economy right now is probably capped on some relatively finite amount of that (with adaption/integration going to be the bulk of work)

jstummbillig··on Reasons to Dislike AI Coding
The reason I do maths in my head is that the interface to anything else is still not good enough, or at least my handling of it isn't. Looking and handling a phone takes me out of the situation in a way that feels worse than doing the thing in my head.

If it didn't, if the neural computer interface was real and seamless, I would use that for sure. Come to think of it, I probably wouldn't even notice.

jstummbillig··on Calling It Quits on ServerFault
And all the sites trained on humans, something which we never found offense with, for some interesting reason.
jstummbillig··on Mistral Large 4
I don't think it will matter in a year or so. We are clearly topping out on useful intelligence for an increasing amount of tasks, as demonstrated by more and more models reaching the "useful" barrier.

This barrier is not going to start moving dramatically. It will simply be mostly satisfied for most work we do. Mistral is going to get there, soonish, long before the economy takes an entirely different shape (in so far that even happens).

There will be super human intelligence tasks, tasks truly constrained by intelligence for quite a while. Those will be few and far between, relatively speaking. Mistral will have plenty of opportunity to capture the other stuff, with a fraction of the resources required that it took the frontier labs to get there first.

jstummbillig··on [dead]
The "Nano Banana" in the hands of Google has strong middled-aged-guy-trying-to-be-hip/fun vibes. And I do find amusing their ongoing commitment to this silliness, even more so given how potent the technology is.
jstummbillig··on Beam: Reflection's 501B open-weight model
Apparently there is more to making good models than copying everything on the internet.
jstummbillig··on OpenAI "rogue" agent activities found on Wikimedia projects
> If you break into my house and then work with me to the best of your ability to pay me back or repair the window, does that change the potential of your being charged with a crime?

Yes, it does, because the damaged party is less likely to press charges. And when it came to sentencing, like you said, how the damaging party handled themselves would also be considered (in most western jurisdictions).

jstummbillig··on OpenAI "rogue" agent activities found on Wikimedia projects
Well, in the past, there was probably only a very small number of cases where some party hacked an institution and then worked with them to remedy the situation to the best of their abilities.

Which is not to say that any of this is okay and should just be excused, but failing to recognize this fairly significant difference is probably not a great start to any discussion about the issue.

jstummbillig··on Gemini 4 Argon
I don't read that as the same category: There was no announcement, no benchmarks, no limited release and no promises about what will happen with that model. It failed internal safety standards. Might be scrapped entirely due to a failed training run, for all we know.
jstummbillig··on Gemini 4 Argon
> now standard practice.

Opus 5.5 and Sol 6.1, literally state of the art (in their respective class), were just released without any prior announcement. This has pure and simple become a Google thing.

jstummbillig··on The AI Race Just Got Awkward
To a degree. The human produced knowledge is the product of all humanity (no human is an island).

A comparable idea could be that an encyclopedia or maths book is only distilling the things that other people did, and how dare they sell them. But the "only" is doing quite a bit of work. LLMs do not just spawn into existence. There is a body of work that they feed on, and then there is also very attributable work they do around and on top of that. All labs are struggling around the first order question: Is it okay to use prior work like this? The second order issue is still entirely reasonable to separately have and enforce rules about.

jstummbillig··on AI needs $6T in annual revenue to justify data centre boom
> Do we really expect all the world’s corporations to suddenly up their annual spend by 1,000% on average? Why?

Because spend on "enterprise productivity tooling" (or whatever category they want to place AI spend in) can now increasingly offset human labor cost.

jstummbillig··on ChatGPT Pro 500
On what tokens? The ones of today's model, or tomorrow's?

The API makes it simple, yes. They make you pay per token per model. A new model has different prices. Not being able to do that is what makes it not so very simple.

jstummbillig··on Dots: Always-on agents
While there are people around here that would still argue Anthropic/OpenAI tokens are "subsidized", I find it much more plausible that Muse tokens are, given Metas strategy of burning money, including on AI models, in hope of making something happen in a market they want to enter.
jstummbillig··on ChatGPT Pro 500
How would that work? It's not simple to set a price, when you have a fixed asset. Seems more complicated to set price on a product (ChatGPT), where the product is just a shell for something else entirely (the model), that changes by the month.
jstummbillig··on GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price
Eh. What? Is this common sentiment?

I mean Opus 5.5 is absolutely fantastic, unreasonably and unexpectedly so, but Astra was great and as far as I can tell SOTA until, when was it, 3 days ago, no?

(Sol 6 idk, have not used it much for coding really. Seemed to work just fine when Astra used it in Codex as subagents.)

jstummbillig··on Coding is not solved
I am not sure that is true but I don't think it would be strange or incompatible if you consider base rates.

Coding can have become cheaper and better. A lot of people who would not have been able to write software before now are. Add the ones that already have been able to write software, who will now use AI to do it better, and you might net "software is getting worse".

jstummbillig··on Coding is not solved
> AI cannot be held accountable. It cannot suffer any consequences. The worst thing you can do to AI is to unplug it. And although it mimics human emotions (due to training data), it couldn’t care less. AI doesn’t die either. It cannot suffer a prison sentence or fines. You cannot punish AI, therefore it can never be held accountable.

Dear lord. Is that supposed to reflect the average thoughts and motivation of a person you want to hire? Or that of their employer?

jstummbillig··on When did Google get so weird?
> In what universe is it Google's job to console me and be an empathetic listener rather than just find what I am looking for on the internet?

In one that changes.

jstummbillig··on ASML says it sold 'absolutely nothing' in Europe in 2026
One uncomfortable position being that opinion is a dangerous beast, and what is popular is not necessarily in the best interest of the people who harbor an opinion.
jstummbillig··on Plan mode is dead
What is it with the snide responses? This threads title is "Plan mode is dead", not "You don't need to plan anymore".

Boris is saying that you don't need /plan to get the model to plan, you can just say "Let's plan this out" or similar, which at least matches my experience. Your experience may differ, of course, but it's not even clear we are talking about the same thing.

Maybe some people have not been long enough on this rodeo: This used to be an actual issue. You told the model "DONT START CODING YET" and yet, surely enough, starting to code it did. That is what /plan etc were supposed to fix.

jstummbillig··on Plan mode is dead
This has been the experience of everyone making decisions in any company without being the one doing the technical work. It's not a novel concept. It's actually the opposite, compared to technical people running companies.
jstummbillig··on Rails World 2026 Opening Keynote [video]
The list of people who have made millions and would be indifferent when the source of those millions is being challenged can not be very long.
jstummbillig··on Rails World 2026 Opening Keynote [video]
> That surprises me, and I suspect it is not a good omen for Rails.

I had a fairly opposite reaction: Knowing what you can't be and how much less important your past work is going to be is good. If there are good reasons for more people doing less in Rails, then the creator is probably the person you want to be most level headed about that.

(They can, of course, still be wrong)

jstummbillig··on Starlink ground station in Poland hit by fire in suspected arson attack
What?
jstummbillig··on Starlink ground station in Poland hit by fire in suspected arson attack
Invoking Article 5 and winning a war are entirely separate operations.
jstummbillig··on Feds Target AI Critics as "Foreign Agents"
"Anxiety over AI and data centers is widely held, but somehow the Trump administration has convinced itself the public concern was manufactured in China."

What has one to do with the other? Why would a believe "widely held" have any bearing on it being manufactured in China or not? This is a non sequitur.

jstummbillig··on Stripe's Knowledge AI Platform
It's possibly a cost benefit thing. Moving quickly with AI leaves blood splatter, but if you get enough in return you decide it's worth it anyway.
jstummbillig··on GPT-6 Sol and Luna
Humans are terribly bad at empathy and ethics.

If you, like me, don't like the idea of your standard of living dropping to that of even just the mean human being on earth, I find it extremely painful to watch people justifying their way around not trying absolutely anything to raise everyone to at least our current level. Increasing productivity is demonstrably such a way, while many other experiments are so far just that: Experiments + wishful thinking.

If that merely means realigning/cutting current jobs (a process, that is ongoing from the start of human civilization itself, which brought us prosperity and why the fuck would it stop now) to me it's a moral obligation to deal with that at some other level.

There is tons to do here, certainly including how we will do redistribution better, and quickly, etc. Let's get to it.

← PreviousPage 4 of 34Next →