Even if the model was open source, nobody would still know how the generated code works. You wouldn't even really know in want interesting sense of the word why the model produced the output you got.
Open source really should mean open training dataset, not merely open weights. There are some truly open source models and you can reproduce their training process if you have the compute. I can't think of a reason why those kinds of models would be incompatible with Stallmanism. Presumably they will get better and more widespread over time, like free software
LLMs are even more proprietary than regular proprietary software, I think you've got it right here. They are destroying the free software movement and it seems like most people don't even realize what's happening.
I definitely agree on this. You don't program an LLM and thus come to an understanding of what it does in the same way you do for a compiler, an operating system, a game engine, a CRUD app, or a web front-end.
Something that's the output of mass-effort machine learning algorithms is categorically different to human-made-and-understandable software. Do we have terms for these different categories?
I used a Mac for a long time, but since Polish uses diacritics heavily, the default position of Option was just terrible for touch typing. I have always simply swapped Option and Command because of that. That was always there in the system's keyboard settings, so no extra tools were needed.
It was only when I moved to overriding CapsLock that I installed Karabiner.
I distinctly remember using Microsoft Keyboard Layout Creator way back to replicate Mac way of entering some typographically correct punctuation on top of Polish layout. You could make your own custom layout as well.
However, when I still used Windows 11, I remember there being a variant of US keyboard layout built into the system that included Mac-like way of entering international diacritics. I no longer have access to a machine with Windows, but a quick Google search suggests United States – International might be what you're looking for.
It sounds to me like you want to censor perfectly fine phrasing just because AI rolled the dice to use it. Let's not forget that AI learned to "speak" from people in the first place.
Not to mention that as people shift their phrasing to avoid being accused of using AI, the AI will keep evolving, training on the new phrasing and then adopting it, too.
I understand the language evolves, but evolving language just to sound different from AI feels like wasting effort better spent elsewhere. On actual substance of the message for example.
You can believe what you want. I don't censor bad grammar, but it's not good writing, and it will have negative effects on someone's writing and who will read it. You can have your fingers in your ears or you can adapt.
We're not talking about bad grammar, but sentences that are correct and sensible, and if you wrote them before AI, nobody would bat an eye. But because they are used by AI, these days they make people dismiss the whole piece.
It's not really different from dismissing people because they happened to use the same phrasing as Stephen King. You can't even tell if it's because they read Stephen King that much, Stephen King wrote it for them, and they are now trying to pass it as their own, or just coincidentally arrived at the same literary construction.
Whenever I see those threads, they usually don't discuss the merits of the writing. They claim that AI must have been used, so it's not even worth reading.
> I think the real problem is Google Play Services. It's far, far too tightly integrated in a way that it just seems like Android was designed to be unusable without.
Wasn't the point of Play services to enable keeping OS "updated" regardless of the manufacturer schedule? Which is to say: they are tightly integrated because they were designed to be part of an operating system that Google can keep updated through Play Store even if the device OEM is slacking on porting upstream updates to the users.
I think it was one of those cases where "the road to hell was paved with good intentions".
I'm old enough and was very pro-android in the early days when it first released. I absolutely remember the frustrations around having vendors just REFUSE to update the OS on devices. Even things that were brand spanking new often got left to rot, and the OS updates in those days were genuinely meaningful software updates - Lots of new capabilities.
So I was actually ok with Google going down the route of moving critical software into an installable package that they controlled, that would provide a way to get new capabilities onto devices that manufacturers didn't give a shit about anymore.
---
But the problem is that "Google" today sure as hell isn't the google that I liked back then (and even that google was starting to shift tone).
And now this package is a backdoor that they can use like a beat-stick to force vendors to stay on their version of Android, and it functions to make "Android" (the distro that users actually use) functionally closed-source and non-permissive.
---
I still strongly believe that the appropriate regulation for this situation (across both iOS and Android) is that the company that makes the OS should be prohibited from running any sort of app store, and app stores must be selectable at device initialization (ex - the Internet Explorer ruling, but for app stores).
Somewhat, but only relevant when it comes to things like WebView but not other things. If there is a vulnerability elsewhere, your device on an older OS that the manufacturer has stopped supporting will not get any patches.
Google may think it's important to keep WebView always up to date with Play services. I couldn't care less. I still have to buy a new phone to get all updates.
reply