Please excuse me while I'm having a little existential crisis, lol.
-
@nina_kali_nina I'd never want the world I knew to stay static. Change is inevitable. I don't think it's ever worth mourning the past for the past's sake.
But, you say existential crisis, is your problem because it is good or because it's bad?
I assume the latter (I still can't get AI to complete simple tasks correctly the majority of the time).
@JigenD big tech is undermining human connection and labour solidarity, while getting richer. That is nothing new, of course, but it's happening faster than ever
-
@nina_kali_nina i don't know about these projects in particular, but not all llm use is automatically bad
@alreadydeadxd that is either sophistry or depends on one's moral values, though.
-
@nina_kali_nina not sure if this helps or makes it worse but im not actually sure the blocked user trick is reliable. On some projects I got this warning and spent a few hours looking through commits without finding anything. And I did notice that it counts things as involving copilot even if it didn't do anything real. Like someone not a maintainer had opend a pull request and unintentionally(according to them) asked for a copilot Review. Which doesn't affect the project but Shows up in filters
So I don't know if I can actually trust this to find what's Ai touched and what isnt
@lunathemoongirl it's not super reliable, but it is generally reliable. Consider that Python for example has a policy that allows AI usage: https://blog.miguelgrinberg.com/post/llm-use-in-the-python-source-code
-
@nina_kali_nina not to get conspiratorial but I guess companies like Microsoft also have an incentive to make you think everything is made by Ai and to not be able to tell.
The claude user profile itself also has no commits but shows up as author in like chardet. So idk it does seem to have wonky special rules. And maybe it gets into projects through being associated with the commits in some pull request and the maintainers couldnt tell?
I mean harfbuzz and chardet are very clear but others.
@lunathemoongirl the detection is based on the commit authors. If you review a PR, you'll see who is the author. I understand that it is in the big tech interests to sneak it in as many places as possible, sure, but I've only seen "embrace, enhance, extinguish" practice and not "oh boy, that was ai? We gotta be careful in PR reviews next time"
-
@nina_kali_nina I hear you. But consider this: like any other profession that creates thing, you can divide it into industrial and artisanal. You can produce stuff, or craft stuff. For software, this has just become more real, more relevant and more visible. You and some other gifted ones will remain artisans, crafting beautiful things. Software giants will spit out crappier software.
@loredema it is a tragedy for, well, pretty much everyone. Software will be worse, end users will have poorer experience at best, rich folks will get richer while shipping systems proliferating racism, sexism, and so on. And probably worst of all, the trust in software chain will be destroyed.
-
@lispi314 I'm not sure, but all these projects have AI authored commits in master
-
@nina_kali_nina @bstacey I do think we need a "web of human trust" of some sort.
Instead of default-open systems where we ban bad actors, we will have to move to default-closed systems where we explicitly allow people trusted by current members of a given community.
-
@enigmatico "not working" is subjective in the eyes of people making decisions, unfortunately.
but yeah, it seems people making decisions about critical systems need to be far more selective in what can be blindly trusted from now on -
-
@nina_kali_nina @bstacey it's not going to be a single project. In some cases it is going to be as simple as changing settings in existing software – like flipping fedi instances from blocklists to allow-lists.
Code foundries also already have existing affordances useful for this.
This is pretty fscking sad, but we've been feeding Big Tech with our code, with our blogposts, with our shitposts, for so long and what we got back from it is *gestures at everything*…
-
-
@nina_kali_nina @bstacey I do think we need a "web of human trust" of some sort.
Instead of default-open systems where we ban bad actors, we will have to move to default-closed systems where we explicitly allow people trusted by current members of a given community.
@rysiek @nina_kali_nina @bstacey the issue with that is it doesn't really work for anything online where a lot of the time you kinda just stumble upon something by chance and get involved; I for one sure as hell didn't have anyone to introduce me into most of the communities I've been a part of over the years; I simply joined out of the blue and slowly became a part of it. -
@lispi314 @bstacey @rysiek I'm a bit privileged here, you can imagine a worse situation. but I've started my open-source journey back in 2005 from a deeply rural place; I didn't know a single person who'd code, and my best bet at learning about the community was going to libraries, sending letters (at first it was letters, not emails), visiting large cities for events (8 hours by bus one way!). It's not what we're used to, but it used to work and maybe it can work again.
-
@lispi314 @reiddragon @bstacey @rysiek well, I'm a counter example to this: https://tech.lgbt/@nina_kali_nina/116181763047346401
-
@nina_kali_nina I really, really hope there will be some push to have “AI free” software and hardware recognized. Much like we have ogm free for food. So you can know and choose accordingly.
@loredema @nina_kali_nina This would be great, but I worry that this would invite AI companies to train on anything labeled as "AI free"..
-
@lispi314 @bstacey @rysiek @nina_kali_nina if I had to rely on someone I knew irl to introduce me into online tech communities, I wouldn't have been a part of any of it.
Tech communities are few in Romania, none was around where I grew up, and most of them are very superficial, into shit like "look how the new iPhone has a sharper camera! you can see it if you zoom in until you can count the pixels!".
Personally I've only met like 3 people who were serious enough about it to do shit like running Linux, flashing their phones, getting involved in open source stuff etc. (and all of them I met years after I got into it myself online). -
Please excuse me while I'm having a little existential crisis, lol.
And if that wasn't bad enough, Mozilla has embraced AI (in its code, too), while Linux considers relaxing AI code policy and has some examples of patches co-authored by LLMs.
I am still yet to think hard about what I want to do about it. But the world I knew is no more.
@nina_kali_nina once again i am sad there isn't a kernel with the same model as linux (aka just a kernel, you can swap out everything else)
the distro model is good, actually, as it gives several pre made configurations (both meanings) that cater to different usecases, and it's a lot easier to switch distributions than to switch for example from freebsd to netbsd
-
@lispi314 @reiddragon @bstacey @rysiek well, I'm a counter example to this: https://tech.lgbt/@nina_kali_nina/116181763047346401
@nina_kali_nina @lispi314 @bstacey @rysiek I didn't even have access to a library with programming books; my local school did have a library but all the books were either literature or natural sciences, and most were older than my parents so they wouldn't lend those. Events were also just not a thing. -
@nina_kali_nina once again i am sad there isn't a kernel with the same model as linux (aka just a kernel, you can swap out everything else)
the distro model is good, actually, as it gives several pre made configurations (both meanings) that cater to different usecases, and it's a lot easier to switch distributions than to switch for example from freebsd to netbsd
@SRAZKVT
i liked the world. it was nice
@nina_kali_nina -
@rysiek @nina_kali_nina @bstacey the issue with that is it doesn't really work for anything online where a lot of the time you kinda just stumble upon something by chance and get involved; I for one sure as hell didn't have anyone to introduce me into most of the communities I've been a part of over the years; I simply joined out of the blue and slowly became a part of it.
@reiddragon yes. I am not saying I am happy about any of this.
But also, I said "web of human trust" for a reason. If my instance has your instance on an allow-list, and your instance has a third instance on the allow-list, my instance should implicitly trust the third instance.
That way we can have that chance discovery of things.
Also, posts could be public, but replies only available to instances within the "web of human trust". Same with code.
