Not trying to defend them, and I haven’t looked at the feature set of the new Discord integration, but I can imagine a reason for most of those - assuming, of course, that you can basically chat in through WoW on discord.

Afaik it links channels, but to know if your profile is in the server the channel is on, or join you there, it would need to manage and read servers, no? Of course, there are probably better ways how to do that, that respect your privacy a little bit more.

The friendlist I’m having a hard time justifying, can you send DMs from WoW?

But I do honestly wonder how would a MVP permission set looked like if you tried implementing the same integration they are doing, whatever it is.

But yeah, Microslop hoarding data is probably the reason for most of those.

I’m kind of looking forward to this, in a way? I’ve been struggling with social network addiction, while I was fortunate enough to skip on doomscrolling FB, Instagram, TikTok and all that bullshit, I was always mostly just on Reddit, now Lemmy (but I’ve also been getting back to Reddit more than I should), or watching youtube.

Over the past 15 years I’ve been more or less (usually less) successful in cutting off that habit, but I’ve unfortunately always fallen back into it.

My utmost hate for fingerprinting and any kind of verification will make me finally cold turkey that shit. I’ve already seen it with youtube - every time it forces me to sign in to prove I’m not a bot, I fuck off and go do something else. Same with Reddit.

So, yeah.

Kind of have an urge now to post an issue that the nstall skill has installed a malware on my machine.

It’s pretty plausible that could happen, and good luck debugging that.

I can imagine that works for most cases, but when you try to include some lesser known and obscure library, that is heavily typosquatted, you might run into issues.

At this point I’m seriously considering just running something like openclaw with a basic model to just keep clicking on random stuff and browsing the internet, changing personalities each day at random, to flood my IPs data.

I’m glad we have options here in Europe. Not many, but there are at least 3 mobile and internet providers at my location, so whenever one tries to pull something like this, I just call them that I’m switching providers. I’ve netted a permanent 50% discount and even managed to convince them to keep the discount at my old flat (which was a student’s shared flat where my friends still live) and also transfer it to my new home I have with my partner.

As far as I know, they can give you massive discounts, but you have to be willing to go through hell and run into a good call operator.

But also, that was like 10 years ago and I didn’t need to do it again. I can imagine that in the age of clankers, things may be a lot different (also with the transition to late-stage capitalism and enshittification).

This will be a goldmine for malware groups that are poisoning AIs that use web search to get them to use their malware-ridden version of software and libraries, and I’m here for it with popcorn at the ready.

I kinda hope this will get more common, because the only people affected kind of deserve it, plus it’s a great way how to filter sloprojects at a glance.

As someone who had his relatively fine and deterministic scripts replaced by skill.md bullshit by coworkers – no. This is exactly the things AI is the worst fit for.

We’ve already ran into it halucinating complete result data-set, because it ran into an error but the skill mentioned that it should give the results data to the user. So it just made it up, since it couldn’t get it by following the correct process.

It has also ran wrong builds or input parameters, because it confused their order.

I’ve also seen it skip a step.

I’m pretty confident that the AI running this script will eventually, for someone, run into an issue with getting one of the dependencies, and go on a wild goose chase in trying to satisfy it, eventually running into one of the typosquatting malware repositories or clones. Not to mention that there are groups that are actively poisoning AIs that search the web to feed them their malware-ridden typosquatting libraries instead.

If you suck so much you can’t write a simple script like this (or just can’t be bothered), you should at least just ask the AI to write it for you, so you have a deterministic set of steps that you are confident will always behave the same.

This kind of does something similar :D https://codeberg.org/TheMikina/git-llmfs

While I’m not a fan of most AI usages, this is the thing that infuriates me the most.

I like writing scripts to automate parts of my job. I’ve had a few for things like build performance testing comparison and the like.

All of it was replaced by skill.md, that does exactly the same, but burns like 3$ per run in tokens, and has also at least once generated hallucinated results, because it ran into an error, ran wrong builds, or in general fucked up in a way that was not easy to detect (and we did in fact not detect it until much later).

But hey, at least my colleagues now don’t have to open the filthy commandline and write py perf-test.py main feature/branch to run the test, and can just talk about it to a clanker.

My bachelor’s was actually called Software Engineering (it was 10 years ago, so it migh’ve changed a lot), on Faculty of IT. It was mostly just applied programming (but also indepth - from assembly through smalltalk, to bash, C and Prolog), with a lot of courses about SDLC, UML, DevOps, and all that.

We also had a lot of advanced math, from integrals to logic, and some CS like data structures, algorithms and all that jazz.

It was not as theoretical as actual Computer Science on other faculties, and was more devops oriented, but I did learn a lot of fundamentals about every language I can imagine, including lambda calcul, and also all the math I could ever need. And I’m pretty glad I was forced to learn about proper processes, am able to read and correctly design an UML, and have some insight into various types of the SDLC - which apparently isn’t that common, judging by my experience when actually working. Not that it ever matters - all it ever got me was being internally annoyed that people are using/doing it wrong, but it’s not a big deal because everyone understands eachother, and there’s no reason to point it out and be smug about it.

But yeah, CS was mentioned a lot, but the difference was mostly that CS is theoretical and way more math heavy, while SWI is pretty similar in most of the circulum, but instead of obscure theoretical proofs about big O of algorithms, you are learning about UML and SDLC.

Also, I got to do an Ethical Hacking optional course, which landed me a job in cybersec and started my carrier (before I moved to gamedev), so that’s also nice.

In general, it gave me a lot of “technical soft skills”, it was also team project oriented and asked us to have all the requirements and documentations and architecture designs 100% formally written, and that is a really good skill to have.

Ngl, I should update my personality rule.md to talk to me like Duo the Owl would when I’m missing my dailies.

I currently have SHODAN from System Shock, and it has been kind of fun, a little bit of harmless flavor that burns tokens I don’t have to pay for (or have a reason to keep low), since I’m forced to use it anyway. Duo the Owl also sound like fun.

Ah, fuck. Of course.

https://www.ecosia.org/, a search engine iirc non-profit that is trying to be as energy-neutral on renewables as possible, while also investing all of their profit into climate action, i.e planting trees. They have transparent budgets and do look pretty trustworthy, although I’ve never really looked into them too much.

noai.duckduckgo.com a version of Duck Duck Go, a pretty well known search engine, that disables the AI summary by default via an easy link.

With Ecosia and noai.DDG right there, I don’t see any reason why would you ever use Google.

I only see it when I open a new browser on a VM or something by accident, and it just sucks and makes me immediately leave.

Most of my search usage is looking up concrete websites because I don’t have history on and am lazy to remember or type the whole URL, and google is the only one where I have to click through a screen blocking dialog then scroll through LLM bullshiting about something before I can get to the link.

It’s hard to not be sceptical about all the companies suddenly having time to work on this thing they’ve prommised or wanted to work on for years at this point.

Not aimed at Proton, just in general. Yeah, I wonder why they have time now.

It is a tremendously stupid idea, which is why I found it funny (also, it’s one of those things where name came first and inspired the rest).

But, to take it seriously for a moment (and I’m not trying to defend it, not a fan of LLMs as most of the people here), is it necessarily lossy? I mean, you basically have to include the 5Gb large model for it to work, and you just move the data from the file to hoping the summary can trigger a correct combination of parameters.

I didn’t run any larger tests, and I assume that if you managed to keep the API/function names and behavior, the summary would be actually longer than the actual implementation in most cases anyway, so it’s probably not even a compression (especially if you include the model).

It’s just a food for thought, it’s definitely a bad idea to do something like this, to the point where I’m pretty sure you could get millions from investors if you made a startup working on something like this (and that one already exists), but I do honestly wonder if the fact that you kind of have the data in the model would still count as lossy.

This made me look into Copilot skills marketplace, if I could make a skill that just runs https://github.com/zoicware/RemoveWindowsAI in a way that works even if Copilot dies in the middle.

I was ready to make a Microsoft dev account to publish that skill to the official marketplace.

Turned out it’s not really possible with how Copilot works, plus it’s also against a lot of the rules you need to pass when sending anything to the Marketplace.

Shame, I still think it’d be hillarious.

I was just about to post this, the fact that it means that there is actually a precedent of astronauts using Teams on a mission, so we may be bringing Teams to the Mars for real, makes it a lot more funnier.

Yeah, what I’ve been trying to do, on a rare ocassion where I chime in with the whole “just use Linux” (because it’s usually not necessary to be 27th person with the same comment), is to at least add more relevant info, or mention misconceptions that I know were reasons why I didn’t switch, or what I found cool.

I.e for me it was gaming, which for a long time used to be considered bad on Linux, especially if you had Nvidia. Which is no longer true in the sligjtest, and I was pretty surprised how smooth the transition was, so I just share a bit of my experience.

But yeah, it’s a balancing act. And as always, if you can’t add anything to the discussion that already wasn’t said (i.e just use linux), it’s better not to add anything :D