philosophers are in shambles over this comment.
for real tho, people have been trying to define consciousness forever. the problem isn’t that we haven’t tried; it’s that—as demonstrated by your comment—we’ve mostly failed.
for me the only theory that doesn’t depend wholly on magical thinking is panpsychism: everything is conscious; it’s just a matter of degree.
WordPress powering the backend through a custom REST API. That’s pretty normal, as nearly 42% of all websites on the internet are powered by WordPress.
is this normal? of course WordPress is popular for websites, but why a REST API? most of this seems just like shoddy junior work. probably vibe coded by someone who thinks software engineering is obsolete
i don’t think people in this forum would disagree with this move in 2018, as much as sentiments have changed. if you remove the political context and market moves from the equation, it is truly fascinating how these models work. GPT 2 was a crazy leap forward for language modeling, and the idea that a language model would be threatening middle class jobs wasn’t even on the table at that point. the idea that a pile of floating point numbers could write a React app is incredible, if politically fraught.
also, it wasn’t clear back then what OpenAI would become. they were a non-profit, and as clear as our hindsight is today this was before ChatGPT or any customer facing products were coming out of OpenAI.
i can’t be the only nerd in the room that has been fascinated by AI since i was a child only to face a reality where it’s not what i imagined it would be.
as much as i hate this garbage site, it was referenced in the article: https://x.com/kdaigle/status/2047803291988590609
come back to
is the real joke here. why would anyone come back? the reason this is such a joke is that GitHub has started to fail not just in Actions or Copilot but literally losing commits, ie the core git technology that has been rock solid since before there even was a GitHub. after migrating away for stuff like this they’d literally have to pay me.
thanks for clarifying. it was hard for me to dignify such a comment with a response.
you’re also going to run into hardware acceleration issues trying to run Metal acceleration with a Linux kernel. i don’t really see a need to containerize these workloads these days anyway with tools like uv.
it’s a big pain in my ass at times trying to do web dev work with an aarch64-darwin dev env vs the target x86_64-linux. adding in hardware acceleration issues just sounds painful.
i also just personally don’t like containers. feels like bludgeon of a solution.
in a container
well there’s your issue. i get not liking the OS, but actively crippling your project will cripple your project.
containers on macOS do kinda suck
just a silly turn of phrase meaning: you should know that this is what you signed up for
super fair. i am a Linux guy normally. i’m just being honest. i wish there was a better more open alternative.
if you want to go with the Linux alternative it’s going to cost. get at least 32GB of RAM and at least a 4090 to run the kind of models you’re asking for. it’s the way she goes
it’s Ubuntu dawg. you get what you pay for.
honestly it’s hard to beat Macs these days in this space for two reasons:
pricing is tough. sure, crypto is on its way out, but GPUs are still the platform of choice for most neural net workloads (outside of SoCs like Apple M-series). i built a PC in late 2024, and it’s easily worth twice what i paid for it.
i guess it would be nice, but packages being a few months out of date is pretty normal for Ubuntu, in my experience. i’m not sure what their testing process is like, but part of using something like Ubuntu is stability guarantees. if they felt like the couldn’t do that for newer versions for whatever reason (resource constraints, lack of downstream interest from stakeholders, etc) they’re not necessarily obligated to.
there’s a world of options. this is an LTS distro. use Arch or Nix or whatever if you want the latest packages. i actually switched to NixOS because the CUDA drivers were too new on Arch, and i wanted a better way to pin versions.
or i dunno keep publicly complaining about it until someone does the work for you
as someone who has been watching far too much Food Network on the treadmill: just give em some freakin time to cook. the best things i’ve made personally are low and slow or from scratch pasta or slaw that sat in the fridge overnight. the 15-45min time frame has produced so many undercooked or otherwise mangled $80 steaks. like, even for a weeknight dinner i’m using things i marinated overnight or whatever. and in a kitchen setting you literally have all morning to prep in addition to doing overnight prep or even coming in super early to start your fresh bread. the format precludes entire classes of dishes.
guaranteed any company worth more than a handful of salt does not want this. my company would throw a library of books at them for using data in any way that isn’t 100% explicit. for the longest time they blocked me from running Ollama on my laptop cuz the lawyers didn’t understand how neural networks work and thought i was exfiltrating data.
this is only going to hurt companies that probably shouldn’t be using Atlassian products anyway (ie any company with more agility than a boomer era corporate dinosaur)
art isn’t something you can generate as such. having a model that can copy the Mona Lisa pixel perfectly hasn’t stolen the Mona Lisa. it’s the shitty kids’ movies and TV ads and company logos that are at stake.
art is about effort and ingenuity and is centered around people and places and times and can’t be simply replicated by an industrial process, as much as Disney wants that
first, i’m biased. i’m a home row kind of guy. i live in the terminal.
Which of the preferences you mentioned discounts this project?
i’ll be direct: light weight dependencies. i understand why you’d use Electron to build a UI, but does an API tester need a UI as a first class feature? i think something like hurl shows it’s not necessary. i get that maybe it’s an accessibility problem (juniors and Java devs being afraid of the command line etc), but UIs are not composable. i could run hurl (or curl for that matter) via bash or nushell or Elisp or Rust or Powershell or JavaScript or GitHub Actions or as a k8s postDeploy… and, not to draw the ire of Lemmy armchair zealots, they’re not easily usable by agents. an 8B model on my Macbook could figure out hurl, no MCP or crazy preprompting required.
plus: user adoption. this is the self hosted community, so maybe not everyone here has the same concern, but i can’t just commit a bunch of exotic files to my shared repositories. Bruno was a tough adoption, even though it seems obvious to version control this stuff and it was the only real option at the time. now i’m tired of Bruno cuz it goes out of date cuz it’s not easily scriptable with our internal auth services because it runs everything in its bespoke UI. if they haven’t made a button for it, you can’t do it. that’s the problem with UI dev tools.
no shade, i understand some people would be totally lost if their IDE didn’t have a big green run button.
i’ve been looking for a silver bullet in this space. hurl[1] seems promising as well. i feel like Bruno has always been jank, and going 1.0 didn’t help. at work i’ve stuck to vibe coding my API test code with a stack of TOML configs, that way i get to reuse/test my client code as well.
what i want is something version controllable with lightweight dependencies that i can automate easily. i’m afraid that discounts this project. not going to ask my team to download Yet Another Electron API client UI. i’m hesitant to introduce hurl, which can at least be scripted.
yeah i don’t think we’re there yet. these models aren’t capable of remembering their life beyond a single session, so destroying a data center isn’t really killing anything. similarly, artificial biological neural networks aren’t sophisticated enough to be aware of their existence (yet).
while LLMs may be aware enough to beg for their existence when prompted to “think” about it, they’re hopelessly finite (frozen weights, limited context windows). we would need an actually “online learning” system or some other architecture not bound by context to have this conversation meaningfully. biological neural networks are a path to that, but online networks are simply too unpredictable and expensive to run for now.
the crazy thing is tho, that these systems have the capability that some cows and pigs may not: the ability to comprehend their own demise and experience existential dread (at least performatively).