This from the company that says developers need to be “verified” to allow them to share apps even outside the play store to prevent malicious, dangerous, or negligent app installs…
3 Likes
Quitting Claude | How we are preparing for an Anthropic exodus.
Ideological LLM avoiders will have to find a new OS: Phoronix - Linux Dropping Assisted-By Requirement
Currently there is a policy requirement for Linux kernel patches where AI tools contributed, they should be documented with an “Assisted-by:” tag to make clear AI was used in the creation of said patch. But Linux kernel developers are now looking at dropping this requirement amid the rampant use of AI/LLMs.
It’s not exactly news that Torvalds is a wanker or that the Linux Foundation (and many other projects that describe themselves as ‘open source’ instead of ‘free software’) is compromised by big tech.
3 Likes
Not at all; GNU is the OS and Linux one of its kernels, hence GNU/Linux. I admit Hurd is not as mature as Linux, but if the slop hits the fan too hard either fork or switch the kernel over are there ultimately for you. The beauty of Free software is that it gives everyone that agree or not agree to the copyleft license the four freedoms. And why you should honour the (Free software) work of others’ by not stealing it from all of us. But if you can avoid that I see no problem with local LLMs.
2 Likes
I just used ClaudeAI to solve a notifications problem with bluetooth. 5 questions I asked and 5 solid replies I got with no errors. Problem solved.
All code given back to me in my .cpp and .h files, is all human readable, as is the corrected parts of my QML. I couldn’t be happier, pretty much the end of an 8 month journey, fixing the notification was just the icing on my cake.
7 Likes
This is what I see too, AI is a very helpful tool by developers knowing how to use it correctly, it is a dangerous tool in the hands of “vibecoders” who cannot understand what the AI is doing.
3 Likes
Everday a new ai vs dev development discussion. Mostly like black and white and without grey. Everybody have to make the choice whats more important to them… May I use app xyz because its nativ but build with ai or not? Answer yourself and all is good. When AI finds 15 year old bugs in openbsd it can be a good tool in the right hands and the opposit in the wrong.
Beside that we have bigger problems than that with ai… environment, society, employment and so on
And I think we should leave this “sfos ai app development” microcosm go over to how we handle the implications with that technology on another level…
So yes, please not antoher telegram client but also please not another microcosm discussion about that topic everyday…
5 Likes
GNU is not the OS, it is merely, commonly, one of several main components of the OS. Isn’t it?
I think AI can be compared to nuclear weapons. The people who control them tend to have great power and prestige, but they can be very dangerous in the wrong hands and generally have only negative consequences for ordinary people.
2 Likes
Transparency would be cool so you can see if an App just is build with AI assistant, or if it’s mainly build from AI.
1 Like
Could be some kind of point system. From AI-assisted 1 to fully vibe coded 10. Then we could have a filter in the app stores that defaults to 0 to protect normal users and advanced users can then set the slop-scale filter to their liking.
2 Likes
Positive? 
If we see a crash like that, millions upon millions may die. The higher they climb before the jump, the greater the damage. We are pretty high up now.
I do genuinely like the Chum approach that ‘borrows’ from something familiar such as nutri-score. That being said, I think that it’s familiarity may be it’s biggest downside: Even after changing the color-scheme, nutri-score is specifically designed to communicate, that one end of the spectrum is good whereas the other end is bad.
Keeping that in mind, I do quite like the llmgd approach since it only communicates the information without attaching visual meaning to it (ai-declaration.md does use colors from green → red → purple).
Nutriscore is a good example of an imbalanced scoring system. Cheese gets a D because it contains fat? Uhm, yeah,.. that’s how it’s supposed to be. If cheese with some cancerous (but not yet detected) replicate will get a better score, how is that fair and how does it make sense?
1 Like
This should be way less controversial with AI, green for human coded and red for totally vibe coded with all the shades in between looks like a perfect fit.
Absolutely. And I think it’s worth the effort, if you use AI, to at least nag it enough to follow the design palette of the system you’re developing for.
4 Likes
It depends on the dev which can prompt AI to more follow Silica UI.
2 Likes