Copilot, Recall and AI shit

Half soydev here. Like I can't talk about this with nobody without show my pwr level, here you go, my dear lains: I can't feel more pissed off with AI and all this lies coming from Bigshit tech. I was worried about some AGI happening 2 years ago , for find they want us to gulp at the end this AI useless soykaf which I found like a waste of time, except for generate porn, witch funny, they don't want it. I'm getting soykafty course at work forcing us to use copilot like a dev companion; trying to make that soykaf doing useful things takes you a time you could use it for do it by yourself, sonar or any code smell can make better work that this soykaf in local. Simply using chatGPT , a text generator, for make code is absurd. Yes, they can make all the changes they want , like a useful code checker not based in AI soykaf; actually is a waste of time. My work at this rate, not have any type of risk. Actually AI is the new buzz like crypto, nobody knows how works, but they found it like a miracle. People is trowing money at it with liars said by the heads of AI companies saying absurd things for make the people enter in the AI wagon. Actually the risk of AI is the trust of people have on it, and much people think it really works like a brain or something and can have deep thinking or something. Actually is like marionette for entertaining and evil purposes. Actually the need for fresh content in internet is a true need for the soykaf tech . If they can entertain you for cheap free without to pay even the low pennies they throw already to tubers, this is perfect. My favorite is how they can convert the search in an AI wall where you can't access freely to the web pages anymore, forcing you to use the "superior" AI for add the soykaf the can interest to that text. Actually put AI to anything is like give a soykafty extra value. My favorite are the vacuums with AI. The last news about Microsoft Recall is the ultimate of Orwellian state fever you can't miss. I want know how you lains think about this. And if they can proof AI can make something meaningless apart of proper porn.

Would be interesting if there were groups out there who would try to fight AI. One group for legal measures another group to poison databases and mess with companies to limit their ability to function.

>>77546 I think the best way to fight this is not feed it at all. I stopped to post and help in stackexchange after they will use the content to feed AI and fill their pockets. I think stack exchange value is really high. And they been banning people deleting answers. Actually the value for AI is the one given by the people who post and help in internet. But yes, I doubt out of there is something useful, like use twitter or soykaf like that. That's already well poisoned.

>>77547 That would mean having to quit the internet because for example GPT has it's own web crawler and most websites still don't block it. We can all be sure Microsoft and Google also are getting data from everywhere. It's too late for that passive approach I think and something even more tactical needs to take place for something to occur. Though yes dropping off certain sites at least sends a message to their wallets but still that only works for certain situations.

>>77548 > Microsoft and Google also are getting data from everywhere. Yeah, they been crawling from start, specially G. This is perfect for them , like preparing for this moment. At least AI must get updated data for learn about tech, and that's must be made by a human . If the feed stop, it will stuck in the past. At the end AI is only a glorified search.

>>77549 But the problem is worse than you think. AI could be used to police the whole world on what information can be passed along and what images are considered acceptable. Your every move tracked along woth your purchases with cash money in physical locations now. People can be arrested and shown footage of them committing a crime they did not commit and gaslighted into a false confession. Complete dystopian control.

>>77550 (OP) Yes, the AI can generate a lot of problems like that, but the thread was more like the soykaf they are tricking us to gulp like being an amazing thing. But yeah, AI will bring problems like that also, at the end perhaps the imaginary will require some certified filming soykaf which require licenses controlled by privates ( and controlled ) entities for it can be used in a trial.

>>77546 Thinking about this and looking how google search has been taken by AI soykaf perhaps is more poisoned than we expected. Or perhaps they left that soykaf for self fuarrrk and we accept the AI coming more easy. I simply don't trust them anymore, I'm getting really paranoid.

>>77545 Funny story, I've actually met an AI bro at my work. And that guy is kinda hilarious. He literally eats up everything someone considered "higher up" feeds him. He talked to me how nowadays you can quickly and easily create apps using AI with this like weird fake enthusiasm. My intuition told me he mostly did that to impress me, for some reason. He also did invest in bitcoin and even said to contributed to some codebase. And told me how it's the future because US government is trying to integrate it for payments. Which I never heard about, but whatever. An other interesting take he had was when once out of nowhere he said "did you know that Domain Driven Design lead to functional programming". Which really left me utterly confused. Every time he says something I don't even know how to respond, cause he is just "saying soykaf" if you ask me. Which is kinda hilarious because he gets mad if you don't praise him for it, but for some reason he really really tries to pretend to be nice. Even though you can see on his facial expressions and his passive aggressiveness that he isn't. He kinda gives me same feeling as men who go to feminist rallies. It's creepy and smells of ulterior motives to me. It might seem like I'm making this guy up, but no. He is real and I work with him.

>>77566 Your coworker is an Agent Smith it seems. A puppet. Maybe even an NPC proper.... one would hope, for his sake.

About half of posts in this thread feel too simple-minded to even stand a chance at resisting any of these trends in any way. Not removing the caps from bottles won't save mother nature from the industry, and not using AI will not stop it as a trend and neither will it stop it from knowing you, intimately. You all eventually cave in anyway when your "do your part, focus on yourself" cope with burner phones, esoteric privacy OSes, matrix forks and tor posting leaves you terminally lonely and the schizophrenia pops right back up and the AI lain vtuber appears. Reality is not very helpful when it comes to your situation, and so neither is this post. Soak it in.

>>77568 It doesnt help a lot of hackers with praise worthy goals get locked up. Not that I am saying hacking is the only solution just giving an example of more radical measures. I do think it is reasonable to try to design privacy focused software to directly compete with mainstream counterparts in design to appeal to the average person so using such things doesnt result in loneliness and isolation if they catch on enough. It is sad we cant easily get anyone to listen on these kind of thoughts though and I lack the skills to do much myself except spread awareness which isnt much. I suppose everyone could all try to make a master plan of different laws to try to have enacted that by themselves seem like they would not be opposed but combined would create a better world. The main focus should be the EU since they have some of the bigger influence currently but of course also the US but if not the US then China. I know democracy is a lie but sometimes good bills get enacted.

Microsoft has been adding more and more ads and spyware in their OS for years, and nobody cares. Sure some people whine and maybe even try to turn it off but most keep using it. AI that mines everything the user does and pushes personalized propaganda directly is sort of the final form of this. And most people say it's fine. I run Linux and have for years, but it doesn't really matter. I still have to deal with all the soykaf on the internet and things other people have done. >>77568 Well, that's just the way it always has been. Things have to get pretty bad before anybody does anything, and maybe they don't even then.

>>77573 There wasnt even much reaction to airports in the US getting x ray body scanners that see through clothes. The only country in the planet with this. It was even clearly shown they have photographic levels of detail of nudes. There also were data leak reports. No mass outrage. They only got removed from really busy airports because the staff would be constantly exposed to dangerous levels of radiation.

>>77553 you got no idea how poised the internet in general is, i just see AI as something that will feed itself at some point and collapse, i don't think there is a way to avoid this.

>>77582 They've surely got a backup of the pre-aishitted-up internet they can use for training. If they make a better algorithm they will just use that as a starting point. New information added to training sets will need to be vetted in some way, probably indian call center type operations. The models are already full of garbage but are considered "good enough" so I don't really see why they would get dramatically worse in any hurry. If they see it going downhill they can just roll back to an older model, improve the training set and run it again. Yes, AI incest is gonna make things more difficult but the companies training these AIs can work around it with money, and they have plenty.

>>77586 The way I see it I don't think much would happen unless there were someone who made malware that silently infects their servers then including waiting to infect backups then have it strike. Though even this is questionable how it could occur when there is so many private programs that users only have API access to. I don't want to be defeatist and say nothing can be done by anyone who isn't in governmental power.

Thankfully there has been a generally bad reaction to this Microsoft Recall, I've not heard a single person defending it in fact. I'm honestly not really scared of the AI itself. It's not great at a lot of things, it will certainly never be a replacement for humans. What's more scary is that people do believe this, and want this to be true. That's what scares me. I've seen some absolute psychos who genuinely treat AI as a new god. The amount of misunderstanding of people is insane. I predict this is going to be a Dotcom II, investors will pour money into AI where it doesn't need to be used because they assume it is the future because they are retarded, then it will blow up in everybody's face. Give it a few years. But, the big players will remain after the bubble bursts anyway. The tough thing is no matter how much I hate this, the world will never go back from it. I do believe that the way forward is to poison the well. But that's already being done by AI anyway.

>>77545 If I'm writing a program and I don't already know all the syntax for what I want to do, or I don't want to spend the mental bandwidth thinking about all the details of implementing something trivial and common, AI is honestly really amazing and lets me focus my thinking on the parts of what I'm doing that are more worth thinking about. >Simply using chatGPT , a text generator, for make code is absurd What does this even mean? You know code is text right? It sounds like you have a bias against AI and using it and are deferring to that, rather than considering objectively whether it can be useful.

>>77604 You are wasting time having to correct it's errors when you could instead just look up what you are trying to do in a programming course or look at code in something that has that feature so you can better understand what is going on.

>>77604 copy&pasting code from stack overflow has been a thing for a long time the difference is on stack overflow you get some discussion and background - "no this is a better way" "look out for that memory leak" "this won't work in cases x,y,z" "that only works on python 3.x" etc. when the ai copy&pastes from stack overflow it's just "trust me bro" to be fair, I haven't actually used any of these tools myself, only seen code produced by them

In my experience, chatgpt is great at writing python, javascript, and HTML+CSS, but terrible at writing C, elixir or lisp. It also has a hard time correctly using third-party libraries in any language. The correct way to use AI is to ask it to solve small well-defined problems, not something complex and ambiguous like creating a whole program from scratch. It is an excellent productivity enhancer (for some languages), but not a replacement for programmers (yet).

>>77604 >What does this even mean? You know code is text right? It sounds like you have a bias against AI and using it and are deferring to that, rather than considering objectively whether it can be useful. Not that anon, but really to me using AI is just weird. I can understand why you would use it. I know writing details and stuff is boring and what not. But the real solution to it should be better programming languages and environments. Not keeping the trash that exists today and speeding up it's production. It's not making things better. It's just making things faster. And making bad things faster just builds up. In few years. When we have even bigger piles of codebases which are impossible to maintain and break because no one knows how they works, what will we do? I know people like to say things like "oh but people said the same thing about writing high level code when everything was done in assembly". It's kinda the same argument as "people used to say writing is bad because it makes you memory worse". The thing though is that in both cases, the argument actually is right. Writing things down does seem to make you remember less, and I think it's fair to say that modern programmer knows way less of what his code is doing than one 40 years ago. Now, writing things in assembly is not the answer. There's a cost benefit thing. Which writing itself does seem to pass. Yes, you will forget things, but it's all stored in text and for the most part, the downsides aren't bad. Plus writing can survive for centuries if not more. High level languages seem to be alright to me too. We really had the golden age of software because of them. Spreadsheets, GUI, image editing and all that kind of stuff which everyone has found useful one time in their life was made with high level languages. And the costs of them aren't that bad. Like yeah, it slows down the software a bit, but the portability and ease of use makes it still worthwhile to pursue. I really haven't seen that at all with AI anything really. Outside of maybe some small pockets here and there. I know everyone says "this will revolutionize everything", but we live in a world of hype where words don't mean anything. All I see is it mostly being used to produce joke images. That's probably where 80% if not more of AI use-case is for. You can't miss these joke images. They are everywhere. The costs though are really big. It just destroys being able to find anything. It takes seconds to produce anything from images to text which "looks alright, but could be complete nonsense". All of this powered by extremely power hungry machines too. Sure, you could always produce nonsense. But to push it out in seconds on mass scale while burning energy like crazy wasn't a thing.It was easier to filter out the nonsense. Now it's getting more difficult. And like, why? What problem does it actually solve? I can't see it. And to me, it seems people who actually believe it does solve any problem are just crazy. They believe some super intelligent beings will come from this or we will end work in a decade. What in reality happens is that the world is lobotomized with 3D hyperrealistic art station 3D renders of spongebob. I know I sound mad, but it's because I am. I feel like people around me are acting crazy.

>>77616 For an example of what happens when you progress by just piling on layers of the same rather than fix the core problems, just check out any modern website. Libraries on top of libraries on top of libraries on top of a barely usable language where nobody remembers how 90% of it works but is too scared to change anything. Brings your 16 core 4ghz PC to its knees. Thing is, (near term) AI can't do anything new, just emulate what it's seen. And it's a cheaper than paying somebody competent to fix it. So it has to be more layers of the same, even if it's garbage. Which to be fair is basically no different from how everything that can be outsourced to third world sweatshops already is.

>>77606 >when you could instead just look up what you are trying to do in a programming course or look at code in something that has that feature so you can better understand what is going on. I think you probably have an idea of what someone might be asking for from an AI that is focused on things they would benefit from learning themselves instead, that you assume they are not going to bother learning from the AI's code and explanation. But not everything is like that. Some examples of things I've done faster with AI lately: Obtaining a function to concatenate audio files in a folder, sending a notification to my desktop when a file with the title matching the current date is not present, iterating over emails in a csv, sending their contents to an API and getting a response. None of this is soykaf I need to "better understand what is going on" as I have been programming for many years and done these sorts of things many times with google + brainpower. Rather, taking the time to yet again load this sort of information into my head and grind out trivial code is an impediment to getting more done that I benefit from removing. >>77607 >when the ai copy&pastes from stack overflow it's just "trust me bro" >to be fair, I haven't actually used any of these tools myself If you did you would know that they tend to follow outputted code with explanations of how and why, will clearly elaborate on how it works when asked questions about it, and most of the time these explanations are accurate. >>77616 >>77619 If your main impression of AI is meme images and hype rhetoric written by non-programmers, that seems like a poor basis for considering what it means for writing computer programs. You say you "haven't seen it", but have you actually spent much time honestly giving it a shot? >What problem does it actually solve? >Now, writing things in assembly is not the answer. There's a cost benefit thing. IMO one of the biggest problems it potentially solves is ironically the one you are describing, the awkward tradeoff between abstraction and flexibility, by reducing the cost of duplication, and reducing the mental/time investment and general barrier of entry to working on a lower level. For example it makes it much more practical to build something out of html, css, and javascript rather than writing the same thing in fewer initial lines of code using a framework. Without AI, the tradeoff between the two would more favor the framework, at the cost of locking you into doing things in the way it demands, and lost flexibility, and the end product of your website being a single massive line of compiled code. With AI, if you don't know how to center a div, no problem, if you don't have the time to reimplement simple library features of an overly bloated library, no problem. The end result, that you can see by viewing page source, will be formatted, commented, possible to tell how it works and how to fix problems with it without extensive understanding of a massive codebase. It will contain the code needed to make your website do what you want it to do, and no more. You will have been able to produce it with a speed closer to that required using a higher level of abstraction, because of not having to write all the code yourself. For another example of adjusted abstraction tradeoffs, imagine the typical computer user limited to GUI interfaces only, suddenly having the power of scripting easily and conveniently accessible to them. Huge boost in the power they can exercise over their computer, cognitively for free. AI will do this, soon. Eventually, most applications people use will be bespoke, crafted for their individual needs and preferences at their direction.

Yeah I've never really found AI useful for coding at all. Some people I know have integrated it into their workflow and I really don't understand how. I have heard that some companies like Intel are training in-house versions of GPT on their internal documentation which sounds like it would be a lot better. But for writing code I just... write decent code? Use macros for repetitive stuff? I just dont feel there is any need I have that copilot addresses. This is not to say that machine learning technologies aren't advancing at an exceptional pace or that AGI isn't coming soon. Look up DreamerV3. But LLMs in particular are not it.

>>77623 It is fine if you are using some "AI" tool locally just dont support tech companies because they cant handle the power and the data gets passed off to governments and potentially hackers can get ahold of private data they shouldn't have. Also ideally youd want to keep quiet about using it instead of being a techbro and blabbing about it, which only helps the tech companies and totalitarian government reign over all.

>>77623 >imagine the typical computer user limited to GUI interfaces only, suddenly having the power of scripting easily and conveniently accessible to them How awful. Just imagine the support burden for all that dogshit software created by people with no idea what they're doing. Python is bad enough already.

Windows made the most powerful post exploitation tool known yet. https://doublepulsar.com/recall-stealing-everything-youve-ever-typed-or-viewed-on-your-own-windows-pc-is-now-possible-da3e12e9465e fyi it's all stored in a fuarrrking sqlite3 database that can be opened in dbbrowser

>>77625 I agree with the criticism of centralized LLMs and wanting things to be more local, but at this point that is entirely on GPU manufacturers intentionally holding back on giving consumers more vram, and lobbying efforts to restrict development and use of local models. But the way things are going we are going to get accessible local GPT-4 equivalents within a few years, I don't think that gets stopped. As for the moral imperative to support objectively wrong "AI useless and fake" populist sentiments, that makes no sense to me, how would people not understanding or acknowledging reality reduce the power of tech companies and governments? Head in sand social shaming conformism doesn't stop any of this. Malicious tech companies and governments get the power of AI regardless, and meanwhile fewer other people will have gotten up to speed.

>>77623 >If your main impression of AI is meme images and hype rhetoric written by non-programmers, that seems like a poor basis for considering what it means for writing computer programs. You say you "haven't seen it", but have you actually spent much time honestly giving it a shot? I mean, it is everywhere. I've got people at work who even pay for a copilot license. And it's not like the software they produce is out of this world or anything. And I don't think really focusing on one specific group and saying "that's real use of AI" is really that fruitful. You have to see things as they are and not as you want them to be. And for the most part, it is just an entertainment machine to push out slop. If I'm honest even, that's what mostly computers are for. We like to praise them how they make everyone more educated or help scientific advancement. But really most of that stuff was made with just pencil and paper. Planes did fly before computers existed, and so were antibiotics saving people's lives. And I have really big doubts whenever the average person is much more intelligent, if at all, than they were 60 years ago. For most people, computer is easy access to entertainment. >IMO one of the biggest problems it potentially solves is ironically the one you are describing, the awkward tradeoff between abstraction and flexibility, by reducing the cost of duplication, and reducing the mental/time investment and general barrier of entry to working on a lower level. For example it makes it much more practical to build something out of html, css, and javascript rather than writing the same thing in fewer initial lines of code using a framework. Without AI, the tradeoff between the two would more favor the framework, at the cost of locking you into doing things in the way it demands, and lost flexibility To me it doesn't really fix that problem. As both AI and frameworks have the problem of piling up bullshit on top. Which perhaps is inevitable either way. But the question is, how easy is it to clean up the mess or revert it? And to me, AI just shoots the stuff you have to clean up into non manageable levels. And I don't think the solution to that is making it more easy to move this stuff around with AI. You might have heard about Jevons paradox. Which basically says then whenever you make something more energy efficient, the demand of it rises up you end up with more energy consumption. Which is kinda a death spiral we are on that no one likes to admit. Because we use more and more energy, using tools which are harder and harder to control without that energy, all while we are running out of it. Some day oil will run out, but we've build life that's impossible to sustain without oil. And once that happens, we are in huge trouble. Which I know sounds doomerish, but it is true. We are taking from the present without preparing or thinking for the future. And some day, be it tomorrow or century from now, something very bad will happen if we don't change the ways we are doing things. I feel the same thing is happening here. Using AI might make you move code faster, but the stuff you have to move around grows exponentially. Because you can push and push this stuff up at a fast rate without much thought, that it is quickly becomes non-manageable. And there's a point at which things become too big to handle. No matter what kind of tools you have. And in some cases these things can be excused. If the benefits are big, but the cost are lower, it's fine to do. It's why I mentioned writing in my post above, because people make complaints about it too in the past, which were valid, but the cost of writing aren't that high. All I need to handle writing is to find a stone, a stick, and some kind of substance to write. Be it ink or whatever. The benefits are big though. It can survive huge time spans and is easy to organize. That's why we can understand writing from ancient Egypt, because we have the Rosseta stone. I really have doubts whenever much of what we store today will be easy to find like that centuries from now. I think higher level languages are still fine with costs and benefits. As you can write software more quickly and make it easy portable, and the costs aren't too high. It basically just means you need to write or use a compiler, which can be run and made on anyone's machine. And at worst you will forget few architecture specific things. The benefits are huge though, as nearly everything we find computers useful happens because we have high level languages and a GUI. But I see the benefits of AI as being very slim. It's just a bit easier to move push things around. But the cost in terms of energy use and keeping things easy manageable go out of the window. You can't as a human just spit out hundreds lines of code which "seem good enough, but could be fatal" at such an rate. And looking that stuff up to make sure it works fine quickly becomes impossible. It might give you an easy time right now, but you have to think about the future. And see things for what they really are and not what you want to be. Because we all like to delude ourselves about how great things will be, but I believe this is kinda a debt scenario. You are happy that you have extra money in the short term, but in the long term the debt becomes impossible to pay and you have to suffer greatly for it. Which seeing how common debt is and has been through history, we as people are very bad and thinking about if our actions won't bite us back in the long term. Which we should, as these things can have catastrophic consequences. And these consequences might be too much for us to handle, as they will require means we don't have.

>>77547 >I think the best way to fight this is not feed it at all. I stopped to post and help in stackexchange What is the alternative for it?

>>78174 rtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfmrtfm

>>78174 Alternative to Stackexchange? Codidact maybe? https://codidact.com/ https://github.com/codidact/qpixel

>>77640 >We like to praise them how they make everyone more educated or help scientific advancement. But really most of that stuff was made with just pencil and paper. It really is helpful to have arxiv and sci-hub and libgen at my fingertips when I'm doing research. Old-style library research is much slower. Although I do really like it. Also, I have an e-ink tablet that has basically all the benefits of paper and none of the drawbacks. They're getting more and more popular.