The question with AI is very simple and it is the same as with any technology: if we let big corporations to monopolize it, then all use cases (even strongly legitimate and ethical ones) will push people to their closed ecosystem. Occupy ALL spaces, dispute ALL attempts to enclosure users. This is the way.
I agree. The current AI landscape has a few proprietary giants but there's a ton of competing open source alternatives and I think it's generally in a pretty good spot for now.
@@noderunner_ Open models are currently kicking the proprietary models' asses. The open source community has nothing to fear. Like everything else, open source is better than the proprietary stuff, and there are healthy communities around it.
Nick, I just wanted to say that I very much appreciate the fact that your background music is barely audible and your videos are not filled with "memes" and flying useless effects and cards everywhere. I miss out on many good Linux TH-cam channels because of how distracting their videos are. Thank you for that.
I don’t know of any linux youtuber that does this. Also this kind of comment is turning into a meme of its own lol. But yeah Nick’s videos are actually great!
League of Legends requiring Vanguard is what made me switch to Linux. Gaming was my only reason to use Windows. Now that my most played game is unplayable, Linux is the superior option.
@@axeldewater9491 it was playable but a pretty bad experience imo, frequent game freezing for a couple seconds and consistent cursor bugs at the start of every match for me.
Yes! I was afraid there wouldn't be a new video. As a Proton customer - very excited about Standard Notes - that was one of my requests and I actually got an email about them addressing this. Good times and warm Spring weather. Hope you have a great weekend Nick!
With Proton aquiring Standard Notes, they almost have a have their office suite completed, with Standard Notes offering both word processing and table calculator features.
With Evernote doing what they did I am going to use Joplin ( used in the past, much better now ) Also dnld LogSec and have been using a little bit for notes. I also dnld Standard Notes but didn't use it yet, but looks very interesting. Once I saw this title, I browsed around the program and noticed it also has self hosting with Wireguard built in with very easy setup for people, I assume Netbird also. Only thing is it installed to Admin which I'm going to try to see if it'll just work with Lmt user acct with Win 11 Pro. We'll see. Just want to add markdown plugin and learn that later as I guess that can be very powerful once you figure that out. Also will work with my new Linux laptop.
I hope Proton gets something similar to Google photos. (I know there's proton drive but that's more akin to Google drive I want something more tailored to photos and videos) Once they have that they'll have an alternative to all the google services I use and I can finally break free.
On my Fedora install. I've replaced the bad espeak tts voices for Pipers TTS voices that run using pytorch and onnx and they sound better than some of the Microsoft Natural Voices on Winodws. (theres and app that's called Pied that makes this super easy to do ).So all TTS related stuff that calls on speech-dispatcher now references theses voices! There is a new Linux speech API and framework called "Spiel" in the works meant for neural voices like these Just recently, we have open-sourced LLM models nipping at the heals of GPT-4 (Mixtral 8x22b and Command R+) Fedora including Pytorch in there F40 repos is one of the best decisions because the modularity of packages in Linux is what makes this platform the best for running AI locally. But it will take time for all the proper packages to come.
@@tommy12331 Yeah, but I guess that's the price we gotta pay if we want to progress. Most of humanity's work was built upon theft, forced colonization, slavery, and destruction.
@@tommy12331 Honestly, the existence and widespread usage of Creative Commons and other forms of copyleft makes me think that, if there was a *truly* ethically built AI that could be used explicitly for non-commercial purposes, there would be individuals who would be willing to volunteer their work to aide in creating such an AI. The logistics for actually *building* that, though, is far beyond me.
You can’t train an AI on “ethically sourced data.” There’s not enough data of that type in existence. You’re talking just public domain stuff, which for an LLM would make the AI sound more like Shakespeare than a useful modern assistant. And for images, the largest source of public domain images are those that were generated by AI. The number of human produced images in the public domain is minuscule considering the length of copyrights and the relatively short period of time cameras have been available to the general public.
They don't just need to make AI ethical and open source, they need to make it optional and more importantly removable. Better yet they can not do anything with AI.
I think AI in Linux could be a good and useful tool like many of the other things in Linux however, what I'm vehemently opposed to is including AI in the operating system by default furthermore I'm opposed to AI that needs Cloud infrastructure to run. If Fedora and other distro maintainers want to include AI it should be available as a voluntary download and it should be a locally running application this way everyone remains honest, and those of us who don't want AI integration in our operating system won't have it shoved down our throat like Copilot.
Exactly This. One of the main reasons I just put Debian 12.5 on older laptop. Will take a little bit to get it fully up and running but I'm learning. Another is talk of the Explorer Patcher and similar apps getting stopped with 24.. Only reason Win 11 is bearable. With AI also getting shoved down our throats, tea leaves have been very heavy for a year.
@@speedytruck I agree speech synthesis is a important accessibility feature but my issue is when the program relies on cloud infrastructure that is to say the program sends data to a server to be processed and then sent back
I think the pascal series will be supported, my 1070 did run well under wayland if only sticking to apps designed for wayland protocol otherwise its basically not usable like steam which runs under xwayland has flickering problems and can't even be used or playing games flickering problem or the cursor is magnified 4x times or illegible text and with x11 there's the vsync issue, too much screen tearing that makes it unbearable.
Yeah, I agree. It would be great if more open source developers would self-host their forums and other stuff. It's annoying that I forced to use Discord just to communicate with the devs/community. The problem is that it requires a phone number on many servers to post.
@@wabbajocky8235 yep, it's under "Verification Level" in moderation where there are 5 options none (no restrictions) low (discord account needs email verification to chat) medium (also needs to be in the server for at least 5 mins to chat) high (now needs 10 mins to chat) highest (needs a phone number) in this case it's the "highest" option making it required
There are good open source solutions for self-hosting Forums. The most popular is probably discourse, but Matrix is also an option if you want something more Discord-like
"We can't find a way to affect the kernel modules and boot state on Linux." So what you're saying is, you don't know how to competently make a client and server-side paired anticheat without outsourcing it to some other company that also doesn't know how to do the same thing... Sounds like a skill issue imho.
It's not a skill issue, Linux just isn't ready made to be rootkitted like windows. It has actual security that has to be patched through. The entire "cheat detector" is still security through obscurity and making one for Linux would expose too much of how they work, making it even more trivial to create workarounds. The cheaters make workarounds anyway, they are always a step ahead of slow moving companies, but as consolation the game companies will now have hardware level access to collect data on your systems... so still worth for them. Idk why anyone would install this trash.. It already exists for many games and are those games are free of cheaters? Nope.
@@craigslist6988 That was my point though. They apparently can't be bothered to make better anti-cheat that doesn't rely on trapping the system. And yea, there are overhead memory access cheats that will probably never be able to be patched to work with something like sandboxed or server-side anti-cheat, but their lack of willingness to try and absolute laziness to just say "well, we know how the game works and how it would be exploited for cheating, but nah, We're just going to put a third-party scanner on people's hardware that hardly works and call it a day"-- it's pathetic laziness that these big companies can't solve. And their responses of "oh its an endless battle" as these games with kernel level anti-cheat see just as much cheaters as TF2 and it's antiquated setup is horrendous. So maybe it isn't a skill issue. They just don't even try.
I'm glad they're fixing the sync issues on Nvidia with Wayland but they're a bit late because I already got rid of my Nvidia card because of the problem.... X11 just felt horrible to use and I was getting frame flicker on any game. So I went and bought a 7800 XT and I haven't had any issues since
12:05 I agree. Hearing that they can't get Vanguard running on Linux sounds like a good thing to me, because the PC sercurity channel showed recantly why Vanguard could be spyware. Riot games being owned by tenscent is enough reason for me anyway. 14:00 Ah. Discords moderration, a wonderfull dream of nightmares again.
If Proton gets notes I'll probably buy a proton subscription. The only thing I wouldn't use from them would be their password manager (unless bitwarden goes to shit somehow).
@@noderunner_ It's nice they offer it. However, may not be the best thing to put too much in one basket. Unless find out they have 2 extra backup servers in numerous countries outside of 15 eyes safe from 500 year tsunamis and such. Just for the business customers or top tier consumer level maybe.
Ethical AI, sure, let's increase energy consumption because climate change isn't a thing. We should also run our military drones and guided missiles with FOSS, so they're ethical.
It's actually pretty amusing that in 2024, developers are still having trouble with their anti-cheat systems. They keep making them more aggressive, borderline sanctioned malware at this point, but it's not working. It's like they're doubling down on their failed strategies instead of trying something new. By 2028 you will need to run their own custom OS while having a always on webcam pointed at your mouse/keyboard to prove your not cheating ... By 2032 this will not be enough so they will remove your ability to run the game locally, while forcing folks to stream it over the internet instead (custom OS/webcam will still be mandatory obviously).
Ability to map the super key in KDE without having to run commands through the Terminal....finally! Gnome file picker using Nautilus, yes! Its the little things....lol. Great video as always Nick. thanks for the updates!
The guy who runs the PirateSoftware channel used to catch cheaters for various Blizzard games when he worked there, including world of warcraft and diablo 3, and he did it using heuristics of what people were doing in the game itself. For example, some of the shop UI in diablo 3 was loaded off screen and so invisible to the player but visible to a bot, so he caught a load of botters by building a heuristic to detect people clicking on UI elements that they wouldn't be able to see. He loathes kernel level anti-cheat because while cheat vs anti-cheat is always an arms race, detecting cheating by in game behaviour and then doing bans in waves (every 3-6 months or so) not only makes it so that the people who make the software then have to guess about how they were caught, if they charge money for their software they also have a bunch of pissed off customers wanting their money back.
Nvidia issues finally being fixed soon is wonderful, but they tool so long that I changed to AMD a week ago (7700 XT yayy) I had the stuttering in games like Yakuza 4 and 5 and kept me on X11. I could have swapped to Wayland for everything else but flipping back and forth is annoying, so now on AMD I don't have to wait for the next driver and Plasma 6.1. I can just use Wayland for everything now.
I disagree with the integrated AI idea. Many people are so put off by this integration (especially online) that they would rather switch to Linux. This should basically just be an installation option, never a kernel component that I have to compile out at the end (just to get rid of openai)....
There's one local use of LLMs on Linux freely available now, and it's great. Speech Note does speech to text, text to speech, and translation, using one of a number of models that can be downloaded. The speech to text, using the large Whisper model, is better than anything I have ever seen: it got "less well known eleventh century hagiographer" accurately on the first test we did. Haven't tried text to speech, and a quick test of translation suggested something at about the same level as Google Translate--useful, but with basic grammatical errors. The speech to text, though not real time, made me realise that there is a use for LLMs in my little world, and Speech Note handles recorded voice files, which is the obvious use for the desktop.
I want to switch away from Windows because I don't want to have AI integrated in the OS. Fedora was the one I was looking at because it has updated software and a stable release. This just cut Fedora off from what I want to install. Arch, Debian or Suse it is then.
I'm writing this on a computer with Fedora, as an experiment. It's not all I hoped for, as I was thinking of using it at work. And AI definitely made me stay wirh anything else. I'm going back to install some sort of Debian instead. AI? Really? Nope from me. Don't worry, I'll still watch your videos :-)
As I could not wait for Tuxedo adding Plasma 6 packages in Tuxedo OS (3.0), have switched to OpenSUSE Tumbleweed with Plasma 6. At first changes/improvements seemed not that big, but now using it for a while, would not go back to Plasma 5.27...
Hey, I know this is not the place I'm supposed to ask this in, but have you got any idea on how to connect to wifi on a Musl LLVM Gentoo install without wpa_supplicant ; nmcli ; iwctl/wifi or any simular commands. USB tethering doesn't work. I've tried to boot from a live USB of another system and copy both the kernel and the binary command files themselfes, but it doesn't seem to work. I've made sure my wifi itself works properly. Help would be greatly appreciated. Thanks!
That's true of most technology. The same Linux kernel the drives the open source community, coul also be used in nefariously by hackers, scammers and even by the servers that run AI algorithms
B..but without a rootkit how can they safeguard you from cheaters by looking at (and totally unrelated maybe selling) your browser history, entire file system, active user time, etc? This is the ONLY way to milk you for ca... I mean.. combat cheaters.
Nah. Ai/ML on the desktop is mostly a stupid fad with few exceptions. Just like 3D displays were in general. Most people I know are entirely uninterested in using anything that is 'AI enabled'.
I don't want Ai in my operating system unless i install it myself or something, I don't really trust the idea of the idea having an AI baked in to the os.
I'm personally pretty excited for the Standard Notes being integrated into Proton, as I tried Standard notes but disliked the paywall, yet I do pay for Proton as I use their VPN E-Mail and Cloud Storage.
It sounds like this will fix the flickering issues I've had when I tried to run Wayland before on my 2060. Excited to give it a shot in June/July when KWin gets updated!!
As someone who is somewhat interested in AI, but does NOT want to use any that is unethical or even somewhat questionable, hearing Fedora focus on helping to develop and create ethically trained AI and other safe and privacy focused AI tools that of course would be Open Source, AND they seem to be taking their time to do it right rather than just go all bandwagon on the AI trend, gives me some hope for humanity and makes me much more excited personally! (Side note: I even plan on tinkering with an open source AI source code to make an artist/user/privacy friendly generative AI tool that never connects to the web, only trains on data you explicitly give it, along with custom made public domain, royalty free data as part of it’s factory settings, and is easy to configure and above all, easy to 100% delete ALL data you give it, and reset to it’s factory settings!)
I was willing to lose the ability to play half my library when I made the move in 2019. Turned out I needn't have worried, as even back then Proton was pretty damn good.
Nobara? Why not Bazzite? I think that's better for handheld PC/Console, especially as you can still follow the SteamOS model of whole-image upgrades BUT with Bazzite's Blue Build infrastructure means you can also stick whatever else you need in the system itself (say, VPN apps).
+1, Bazzite feels like a better SteamOS, eve better than Valve's, plus it is built on an intended image-based system compared to twisting Arch to be one lol
There's Fedora Silvernobara if you wanna have image upgrades. It's just a repo you add to your Fedora Silverblue in order to get the latest improvement from Nobara into your Silverblue.
Seeing people mentioning bazzite more and more online makes me happy. Jorge Castro and his team are doing awesome work with Ublue. I'm still pondering if I want to run Ublue Kinoite or Ublue Aurora 🤔.
The Explicit Sync news was very well timed for me. I am trying to migrate from a laptop setup which would murder my back (also wouldn't allow me to use a decent keyboard) to a more desktopy one, so I've attached my old 60Hz monitor. And sadly, whereas I didn't have any issues on my 240Hz monitor, this one was just working like an absolute crap. This forced me to switch from Hyprland to Mate for the time being. Weirdly enough, the next day in the morning Brodie suddenly releases the video about exactly this issue (or at least I hope so) being fixed. Odd, but very glad I will be able to move back. Sadly, X11 also works like garbage. If I turn off my laptop monitor, the whole system becomes barely usable (any window takes about 10 seconds to respond), so Xorg really showing its age here
I feel like Fedora will probably have a decent implementation. I find Linux out of the box feels ancient for not having AI features out of the box. People coming from macOS take AI for granted. I've been able to mnually add AI powered features to my own Linux box, and it is a lot cooler for that. I think the best decision is if they make the AI features locally run or make it an optional component.
@@grantschilb8019 Optional, I guess is fair enough. I'm personally not a big fan of it and I don't want it shoved down my throat on Linux like it is on Windows and Mac.
@@diotitus Yeah, I wish Microsoft didn't force copilot on Windows. I like copilot, but I prefer it as an opt-in web-based feature. Apple at least makes Siri optional. Although this year, the new macOS 15 will integrate with Google's Gemini, which bothers me. Google's AI, from a technical perspective, is very dysfunctional and I hope apple doesn't give it kernel level access. If Apple oversteps, then maybe I'll stop using macOS. Only time will tell.
I am so happy about explicit sync finally coming soon, saying I am TIRED of this issue is saying nothing! haha I'm glad process is being done to fix this, dedication to make it a lot usable
05:10 my only issue here is Snapshot not being able to select which camera you want to use. My Laptop has an IR Camera wich always ends up as video0 and Snapshot never works because my real camera is video1.
Glad to hear a notable distro is actually looking at ai, no matter what your opinion of the various implementations of ai we have now is the fact of the matter is that its here to stay, perhaps not to the degree microsoft/google/apple would like to sell you (llms that run off remote servers being built directly into the desktop is foolish on a lot of levels...), but there are plenty of practical uses for the tech that aren't going away and the more involvement of the open source community we can get in these things the better, I do not want to see a future of ai tools that may one day prove vital to business being dominated solely by large for profit companies.
I can not speak for everyone, but I have been considering moving back to Mint or Ubuntu from my MacBook because I want to get away from GPT style AI on everything. 😅
Wasn't league already gone on Linux? I hadn't been able to run it for a very long time. All of the reports on Lutris were saying the same thing for like a year. Maybe that's why they only had 800 users, because nobody else could run their garbage.
Could A.I. be able to: Actively monitor kernel for more efficient resource usage and/or streamlining background processes? Assist in writing use specific kernels? I'm just curious on what ways linux could harness the power of A.I. and possibly ai/ML
For one, ChatGPT is exceptionally good at writing scripts and solving problems on Linux today. I keep hearing complaints from Windows people trying Linux - that they often find themselves stuck in BASH trying to look up and remember commands for doing certain things. This could help them learn, or at least automate things by accepting human language and writing the scripts for them... Software names can also be a mystery to newcomers: it could help you choose software based on the needs you describe to it and figure out the difference between the different software; solve problems you encounter trying to do something... Sure you have a search engine for that, but this is more of a hands-on approach. :P Not vibing it with the default desktop backgrounds? You could ask an image generating AI to make something more appropriate for you. I would. An AI layer could be interesting IMO, though I don't think I'd have enough resources available locally to make good use of it. I WOULD like a standardized API for such things so that I would have a choice of whether to hook to ChatGPT, some other service, or even a cluster of laptops at home - that's what having AI on Linux would be all about, compared to the proprietary shit...
That AI thing would also be useful to help you with using the terminal and cli-only programs. Many people are still afraid of it, so an AI helping you through it would be nice (would also eliminate falling for the rm -rf trolls)
I'm usually not advertizing, but I just have to. I've got one of these monster Stellaris laptops. It's as good as you would expect it to be. It has a few downsides, mostly related to the firmware (slow boots) and Nvidia. But I knew what I get myself into there. What I didn't know is that you can actually control the backlight of all keys individually. There is no software for that (afaik), but it's easy to do (echo r g b > /sys/class/led/...). Can't wait to find the time to integrate that into my keybindings
I started using Siver blue after year of Fedora. Installing drivers was a bit shit but it so stable and faster updates via ostree the dnf. AI I am leaving to SUSE Micro Aeon if only I could work out how to install codecs extra on it. That said haven't use SUSE for years. DNF is just easier to use for me. A yrst on Gome is janky as hell still. While my aim is in making gnome setting cover what you need, then running one distro only software. Also fail to find a way to install Waydroid on Suse Micro yet.
Really like your Channel and the News. A year ago I was still using windows and lots of proprietary software, but you got me to give linux a try and go (almost) fully open source. Started that 6 months ago and found my way (currently with Kubuntu) and really like it. As a "proud" owner of an Nvidia Card (thx to being a long term windows user) I'm really looking forward to wayland working correctly with my Nvidia card
It's just an opinion from Mathew Miller, in order to get an AI implementation like a local model running in a workstation iso for whatever, it should be presented as a proposal with a definitive solution, let's say, a client that let you download a llama2-3B-Fedora model, as a very specific example. Don't think Matthew Miller dictates everything on the Fedora project, there are proposal, discussion and elections within the Fedora project community.
And Ubuntu 22.04 got a STIG from DISA in April (2024). So much more security for everyone & also for the people DISA works for. In a more open context, CIS benchmarks are doing good things for the large corps utilizing open OS & select apps.
AI can be used to characterize you with behavior without the details. It’s always free. It’s always nice. It’s always good in the beginning with everything that we’ve had so far. But I assure you this time will be different!
I am facing a lot of flickering issues on wayland in my setup running tumbleweed and rtx 3080 (desktop). More consistent issues in chrome and electron based apps.
Artist transitioning to tech, disillusioned and disappointed by the ridiculously over-hyped and unethical 'content' generating AI we have now. A solution to a nonexistent problem built on theft in a black box. I've never hit a subscribe button harder than after hearing "public doesn't mean free to use for every use case." Thank you.
As for the AI stuff, ya it needs to be worked on for AMD Radeon GPUs but some of the problem is their published documentation which may or may not have improved. Then there's the upcoming RISC-V hardware and also things like Raspberry Pi.
Explicit sync is a nice to have. Doubly so with it being the way Vulkan wants to handle things. Should help a lot. While I don't care about AI, I cannot deny that it's there and it's not going away. It'll be a good thing to have an open implementation about AI. If it's gonna happen, then we might as well get ahead of things and be ready. Rest in peace League on Linux. I've had my fun with the game, but I decided to get rid of it when the Vanguard announcement came and I do not miss it.
Hopefully Fedora AI doesn't turn out like Nouveau where it's there to exist and adhere to the morals of open source nature but doesn't end up excelling in a single area compared to proprietary counterparts.
Fully agree on No Kernel Access for something as frivolous as anticheat (that doesn't work 100 per cent anyway!). I'm okay missing out on badware. Thanks, Nick! 🥰 🐧
I see AI as a huge opportunity for copyleft, so long as it is taken advantage of. The copyright aspects of AI training are a huge roadblock for most models, but that doesn't apply if explicit permission is granted in the license to use the work and make derivatives. That means that if an LLM is trained on the vast corpus of CC-BY-SA and compatible work, its outputs would all need to be CC-BY-SA. Same goes for image generators. Coding LLMs too, substituting CC-BY-SA for GPL. While it would still cause fragmentation around the different incompatible Free licenses, that's still a huge improvement over the situation with closed models, where the data all needs to be either Free but non-copyleft or specifically (expensively) licensed the way that Adobe is doing with Firefly or Google is trying with Reddit. Imagine what it would mean if the most powerful tools for writing code meant that all the code had to be copyleft. But none of that can happen if it's just ignored. I'm sorry if this was an incoherent mess, I'm running on three hours of sleep.
I've pondered this issue as well, and I've come to see intellectual property as a house of cards propped up by profiteering lobbyists. The expectation that a machine learning model must strictly adhere to the licensing of its training data mirrors the notion that a human artist drawing inspiration from existing work must conform to the wishes of the original creator - an unnecessary and restrictive constraint on individual autonomy.
@@noderunner_ That's certainly a debatable point, but the fact is that *at the moment*, the law is not clear. For all we know, the outputs will all be public domain. This is just one possibility, and I think that the potential upside is worth the risk. Even if it doesn't work out, we still get an ethical AI, which we don't really have.
@@skywz There's a prevalent sentiment that Machine Learning is deemed 'unethical' when models are trained on copyrighted data. Advocates of 'ethical AI,' pushing for more stringent intellectual property laws in ML, risk steering us into a dystopia where generated content becomes monopolized by corporate entities under complex IP regulations. This stifling landscape will impede individual freedom, leaving only those with the resources to navigate legal complexities to benefit from this transformative technology.
@@noderunner_ I've actually made this argument myself in person. When I say "ethical" I mean that it is ethical by promoting copyleft works through its output and ethical by being local, private, and transparent.
13:14 they didn't say they had 800 daily linux players. They said they had 800 linux players "yesterday", which seems insanely disingenuous when you think how often companies use "statistics" to "lie". For all we know that was also only on a specific server, which they happily just decided to ignore. This also comes with the fact that linux players already knew about vanguard, so they probably stopped playing as much.
Don't be giving me a heart attack. Windows 11 forcing AI that I have no control over is one of the main reasons for switching to linux. They better not be forcing that shit on us in LInux as well. I love AI that I can localize and control, but anything else being on my system against my will is a no go. Very concerning from somebody who loves AI and Linux. Also...I bet if you were a portrait artist in the 1800s, you would have been furious and fighting it every step of the way huh?
@@-KillerHawk- For over 15 years yeah, and an artist of many other types for over 30. And programmer, father, philosopher, buisness owner, and trained soldier. I am also a musician and game developer. I guess we found what kids today are calling an "NPC". Also spent my teen years on the streets involved in organized crime. Can you please explain with your wisdom, who I am?
@@-KillerHawk-found the nonsequitor.. Who cares what anyone does for a living? Having a constant data feed to some company hosted "AI" (not actual AI..) is a bad idea period. They're not worried about running out of training data when people are literally paying them to harvest their data while using the AI.
Does Nick go into why they don’t like AI? I would like to hear their take as I have mostly optimism for the trajectory of generative AI in its current state
Kernel-level anti-cheat really isn't the solution. But I wish Linux would start to implement tools that provide applications with a similar level of trust in the system they're running on, without being as invasive as a rootkit. Things like a userspace API that prevents other programs from hooking into certain processes, or a cryptographic system to verify that the kernel comes from a trusted source like the distro's official repos.
Aww look at you coming up with actual ideas for real anticheat solutions, hahhaha.. Don't tell me you really thought their rootkits were for anticheat... These companies have never been interested in stopping cheaters, those are their best customers.
Client-side anticheats never work. Those are good ideas in theory, but an API call can just be removed from the code. A cryptographic system could be spoofed and would prevent people from legitimately working on kernel stuff. And of course, kernel anticheats can also be disabled/spoofed and are useless too. The best anticheat solution is to be cautious of what power and data you give to the client from the server. Wallhacks wouldn't be an issue if the server was smart enough to not send info of players you can't see and can't hear walking. Fly hacks/item hacks in games like Mario Kart for instance pretty much only occur due to the game having a terrible architecture that lets the client do way too much. Aimbots are the toughest to fight against but even kernel-level ACs do a bad job against those currently. The best solution for those is probably behavioral analysis like reCAPTCHAs work. Of course yeah you never will be able to get rid of all cheats, but you'd reduce significantly their impact. Like you could have a lesser version of a wallhack that renders sound sources, but it would be pretty limited in scope, and is already present in some games as an accessibility feature. Or a lesser aimbot that only works in screen-space and that would have to make a significant effort to look legit. On Minecraft for example, you can't prevent people from installing Xrays, but what you can do instead is to make them useless by feeding the client with garbage data on blocks they cannot see.
Head to squarespace.com/thelinuxexperiment to save 10% off your first purchase of a website or domain using code thelinuxexperiment
No.
@@j0hnny_R3db34rd No need to make him feel bad. The guy's gotta earn money somehow.
“Nvidia is fixed on Wayland” is like a headline out of Groundhog’s Day for the Linux world
- removed herobrine
- fixed nvidia
Next is going to be Hyprland:
- Added Nvidia support
Once this has been fully rolled out, it will be the Year of the Linux Desktop!
Yeah. "I've been using wayland on nvidia for a week now and it justs works" *use is just launching a browser with youtube* "See ! Just works."
Yeah, but this time, it's like the last major hurdle is now officially passed
The question with AI is very simple and it is the same as with any technology: if we let big corporations to monopolize it, then all use cases (even strongly legitimate and ethical ones) will push people to their closed ecosystem. Occupy ALL spaces, dispute ALL attempts to enclosure users. This is the way.
I agree. The current AI landscape has a few proprietary giants but there's a ton of competing open source alternatives and I think it's generally in a pretty good spot for now.
"even strongly legitimate and ethical ones" there is currently no such thing in this 'ai' hype bubble
you're just creating an 'open' alternative, but not excluding the exploitation, which very strongly misses the point of open source/ethical software
@@noderunner_ Open models are currently kicking the proprietary models' asses. The open source community has nothing to fear. Like everything else, open source is better than the proprietary stuff, and there are healthy communities around it.
@@ThylineTheGay If you think the models are unethical, you don't understand what they are and how they work.
I strongly believe AI should not be considered open unless its training dataset is available
I agree, and I'm not sure how you could argue otherwise.
100%
Agreed. Open weights is not truly open source. But it is a lot better than the more closed down versions being pushed by the big companies right now.
Honestly datasets are way more important than models. Compare usefulness of GPT-J to Pile.
From the ground experience on applied statistics, this can't be more true!
Nick, I just wanted to say that I very much appreciate the fact that your background music is barely audible and your videos are not filled with "memes" and flying useless effects and cards everywhere. I miss out on many good Linux TH-cam channels because of how distracting their videos are. Thank you for that.
I don’t know of any linux youtuber that does this. Also this kind of comment is turning into a meme of its own lol. But yeah Nick’s videos are actually great!
League of Legends requiring Vanguard is what made me switch to Linux. Gaming was my only reason to use Windows. Now that my most played game is unplayable, Linux is the superior option.
Wait but why did you keep playing LoL on Windows if it was possible to play it on Linux? You could have deleted Windows far earlier.
@@axeldewater9491 it was playable but a pretty bad experience imo, frequent game freezing for a couple seconds and consistent cursor bugs at the start of every match for me.
Helldivers 2 is what made me finally convert the gaming rig to Linux full time. Don't feel like getting rooted on Windows. 😅
Going with a malware grade anti-cheat to deal with cheaters is like someone taking a lobotomy to deal with a headache.
Yes! I was afraid there wouldn't be a new video. As a Proton customer - very excited about Standard Notes - that was one of my requests and I actually got an email about them addressing this. Good times and warm Spring weather. Hope you have a great weekend Nick!
With Proton aquiring Standard Notes, they almost have a have their office suite completed, with Standard Notes offering both word processing and table calculator features.
With Evernote doing what they did I am going to use Joplin ( used in the past, much better now ) Also dnld LogSec and have been using a little bit for notes. I also dnld Standard Notes but didn't use it yet, but looks very interesting. Once I saw this title, I browsed around the program and noticed it also has self hosting with Wireguard built in with very easy setup for people, I assume Netbird also. Only thing is it installed to Admin which I'm going to try to see if it'll just work with Lmt user acct with Win 11 Pro. We'll see. Just want to add markdown plugin and learn that later as I guess that can be very powerful once you figure that out. Also will work with my new Linux laptop.
I hope Proton gets something similar to Google photos.
(I know there's proton drive but that's more akin to Google drive I want something more tailored to photos and videos)
Once they have that they'll have an alternative to all the google services I use and I can finally break free.
12:05 I love how the captions for this section said "Bye LoL" 😂
Having ethical AI seems totally fine. It's a new technology like any other.
On my Fedora install. I've replaced the bad espeak tts voices for Pipers TTS voices that run using pytorch and onnx and they sound better than some of the Microsoft Natural Voices on Winodws. (theres and app that's called Pied that makes this super easy to do ).So all TTS related stuff that calls on speech-dispatcher now references theses voices!
There is a new Linux speech API and framework called "Spiel" in the works meant for neural voices like these
Just recently, we have open-sourced LLM models nipping at the heals of GPT-4 (Mixtral 8x22b and Command R+)
Fedora including Pytorch in there F40 repos is one of the best decisions because the modularity of packages in Linux is what makes this platform the best for running AI locally. But it will take time for all the proper packages to come.
@@tommy12331 Yeah, but I guess that's the price we gotta pay if we want to progress. Most of humanity's work was built upon theft, forced colonization, slavery, and destruction.
@@tommy12331saying AI is built in theft is like saying Google is built in theft. It's just a information retrieval technology. Get over it.
@@tommy12331 Honestly, the existence and widespread usage of Creative Commons and other forms of copyleft makes me think that, if there was a *truly* ethically built AI that could be used explicitly for non-commercial purposes, there would be individuals who would be willing to volunteer their work to aide in creating such an AI. The logistics for actually *building* that, though, is far beyond me.
When Canonical and Ubuntu will try to implement it, people are gonna rage at them lol
You can’t train an AI on “ethically sourced data.” There’s not enough data of that type in existence. You’re talking just public domain stuff, which for an LLM would make the AI sound more like Shakespeare than a useful modern assistant. And for images, the largest source of public domain images are those that were generated by AI. The number of human produced images in the public domain is minuscule considering the length of copyrights and the relatively short period of time cameras have been available to the general public.
They don't just need to make AI ethical and open source, they need to make it optional and more importantly removable. Better yet they can not do anything with AI.
I think AI in Linux could be a good and useful tool like many of the other things in Linux however, what I'm vehemently opposed to is including AI in the operating system by default furthermore I'm opposed to AI that needs Cloud infrastructure to run. If Fedora and other distro maintainers want to include AI it should be available as a voluntary download and it should be a locally running application this way everyone remains honest, and those of us who don't want AI integration in our operating system won't have it shoved down our throat like Copilot.
Exactly This. One of the main reasons I just put Debian 12.5 on older laptop. Will take a little bit to get it fully up and running but I'm learning. Another is talk of the Explorer Patcher and similar apps getting stopped with 24.. Only reason Win 11 is bearable. With AI also getting shoved down our throats, tea leaves have been very heavy for a year.
What's wrong with including a proper speech synthesis with the OS? It's a very important accessibility feature.
@@speedytruck I agree speech synthesis is a important accessibility feature but my issue is when the program relies on cloud infrastructure that is to say the program sends data to a server to be processed and then sent back
@@mcash2189 Yeah, I think they're working on local AI.
Too bad users with legacy Nvidia cards will be forever left out but hey, the progress is exciting for anyone else!!
Legacy Nvidia cards so like… last generation?
No I have a 1060 and it's not legacy @@Freshbott2
@@Freshbott2 the ones that are not supported by the latest proprietary driver or NVK
I think the pascal series will be supported, my 1070 did run well under wayland if only sticking to apps designed for wayland protocol otherwise its basically not usable like steam which runs under xwayland has flickering problems and can't even be used or playing games flickering problem or the cursor is magnified 4x times or illegible text and with x11 there's the vsync issue, too much screen tearing that makes it unbearable.
@@Freshbott2 think 700-series or earlier
Yeah, I agree. It would be great if more open source developers would self-host their forums and other stuff. It's annoying that I forced to use Discord just to communicate with the devs/community. The problem is that it requires a phone number on many servers to post.
it's the dev's choice on whether to enable phone number verification for their server, no?
@@wabbajocky8235 yep, it's under "Verification Level" in moderation where there are 5 options
none (no restrictions)
low (discord account needs email verification to chat)
medium (also needs to be in the server for at least 5 mins to chat)
high (now needs 10 mins to chat)
highest (needs a phone number)
in this case it's the "highest" option making it required
@@wabbajocky8235 Still, I don't think that discord should be used for forums.
I think matrix can be self hosted ?
There are good open source solutions for self-hosting Forums.
The most popular is probably discourse, but Matrix is also an option if you want something more Discord-like
"We can't find a way to affect the kernel modules and boot state on Linux."
So what you're saying is, you don't know how to competently make a client and server-side paired anticheat without outsourcing it to some other company that also doesn't know how to do the same thing...
Sounds like a skill issue imho.
Well, Valve is also not competent enough (or they may not be trying), seems to be a widespread skill issue if it actually is one.
It's not a skill issue, Linux just isn't ready made to be rootkitted like windows. It has actual security that has to be patched through.
The entire "cheat detector" is still security through obscurity and making one for Linux would expose too much of how they work, making it even more trivial to create workarounds. The cheaters make workarounds anyway, they are always a step ahead of slow moving companies, but as consolation the game companies will now have hardware level access to collect data on your systems... so still worth for them. Idk why anyone would install this trash.. It already exists for many games and are those games are free of cheaters? Nope.
@@craigslist6988 That was my point though. They apparently can't be bothered to make better anti-cheat that doesn't rely on trapping the system. And yea, there are overhead memory access cheats that will probably never be able to be patched to work with something like sandboxed or server-side anti-cheat, but their lack of willingness to try and absolute laziness to just say "well, we know how the game works and how it would be exploited for cheating, but nah, We're just going to put a third-party scanner on people's hardware that hardly works and call it a day"-- it's pathetic laziness that these big companies can't solve. And their responses of "oh its an endless battle" as these games with kernel level anti-cheat see just as much cheaters as TF2 and it's antiquated setup is horrendous.
So maybe it isn't a skill issue. They just don't even try.
Embracing server-sided anti-cheat is imperative to safeguarding users from the risks of complete OS compromise by foreign entities.
They actually said attest, not affect.
Thanks for your news! as always.
I'm glad they're fixing the sync issues on Nvidia with Wayland but they're a bit late because I already got rid of my Nvidia card because of the problem.... X11 just felt horrible to use and I was getting frame flicker on any game. So I went and bought a 7800 XT and I haven't had any issues since
I like that ethical AI initiative from Fedora.
Fedora really is the Linux flagship distro that pushes the Linux desktop forward.
stable and innovative, fedora has it all
*Morally Correct.
12:05 I agree. Hearing that they can't get Vanguard running on Linux sounds like a good thing to me, because the PC sercurity channel showed recantly why Vanguard could be spyware. Riot games being owned by tenscent is enough reason for me anyway.
14:00 Ah. Discords moderration, a wonderfull dream of nightmares again.
If Proton gets notes I'll probably buy a proton subscription. The only thing I wouldn't use from them would be their password manager (unless bitwarden goes to shit somehow).
I agree, the fact they prioritized the Proton Pass product when Bitwarden exists is a headscratcher.
@@noderunner_ It's nice they offer it. However, may not be the best thing to put too much in one basket. Unless find out they have 2 extra backup servers in numerous countries outside of 15 eyes safe from 500 year tsunamis and such. Just for the business customers or top tier consumer level maybe.
Ethical AI, sure, let's increase energy consumption because climate change isn't a thing. We should also run our military drones and guided missiles with FOSS, so they're ethical.
Yeah if Fedora start putting AI crap in the OS, I'm done with them. If it is just a program I can install at my discretion, then I am fine with it.
It's actually pretty amusing that in 2024, developers are still having trouble with their anti-cheat systems. They keep making them more aggressive, borderline sanctioned malware at this point, but it's not working. It's like they're doubling down on their failed strategies instead of trying something new.
By 2028 you will need to run their own custom OS while having a always on webcam pointed at your mouse/keyboard to prove your not cheating ... By 2032 this will not be enough so they will remove your ability to run the game locally, while forcing folks to stream it over the internet instead (custom OS/webcam will still be mandatory obviously).
Ability to map the super key in KDE without having to run commands through the Terminal....finally! Gnome file picker using Nautilus, yes! Its the little things....lol. Great video as always Nick. thanks for the updates!
The guy who runs the PirateSoftware channel used to catch cheaters for various Blizzard games when he worked there, including world of warcraft and diablo 3, and he did it using heuristics of what people were doing in the game itself. For example, some of the shop UI in diablo 3 was loaded off screen and so invisible to the player but visible to a bot, so he caught a load of botters by building a heuristic to detect people clicking on UI elements that they wouldn't be able to see. He loathes kernel level anti-cheat because while cheat vs anti-cheat is always an arms race, detecting cheating by in game behaviour and then doing bans in waves (every 3-6 months or so) not only makes it so that the people who make the software then have to guess about how they were caught, if they charge money for their software they also have a bunch of pissed off customers wanting their money back.
Nvidia issues finally being fixed soon is wonderful, but they tool so long that I changed to AMD a week ago (7700 XT yayy)
I had the stuttering in games like Yakuza 4 and 5 and kept me on X11. I could have swapped to Wayland for everything else but flipping back and forth is annoying, so now on AMD I don't have to wait for the next driver and Plasma 6.1. I can just use Wayland for everything now.
Nvidia Wayland: I activate my spell card "Fix issue".
RogueRen: Ha, you run into my trap "Too late; already switched to AMD."
Nvidia Wayland: Oh no!
I was excited for this, but no hdmi 120hz issue made it a no choice 😮💨
@@bengtakevarg wait what? I use HDMI 100hz?
"Sorry, we don't support operating systems that don't let us tamper with the kernel" -Riot Games
I disagree with the integrated AI idea. Many people are so put off by this integration (especially online) that they would rather switch to Linux. This should basically just be an installation option, never a kernel component that I have to compile out at the end (just to get rid of openai)....
Agree, I think it would be best for it to be something like an extension or external package that can be installed.
You know local models that are 100% offline exists, right ? Rwkv is already part of the linux Fondation.
There's one local use of LLMs on Linux freely available now, and it's great. Speech Note does speech to text, text to speech, and translation, using one of a number of models that can be downloaded. The speech to text, using the large Whisper model, is better than anything I have ever seen: it got "less well known eleventh century hagiographer" accurately on the first test we did. Haven't tried text to speech, and a quick test of translation suggested something at about the same level as Google Translate--useful, but with basic grammatical errors.
The speech to text, though not real time, made me realise that there is a use for LLMs in my little world, and Speech Note handles recorded voice files, which is the obvious use for the desktop.
a small 3b local llm, is pretty performant even in some lower end cpu only computers
would probably work well for a copilot alternative in linux
The biggest thing of use could be content generation for small game devs
I want to switch away from Windows because I don't want to have AI integrated in the OS. Fedora was the one I was looking at because it has updated software and a stable release. This just cut Fedora off from what I want to install. Arch, Debian or Suse it is then.
Yay! Proton Notes!
woohoo i can stop using Standard Notes now =)
I'm writing this on a computer with Fedora, as an experiment. It's not all I hoped for, as I was thinking of using it at work. And AI definitely made me stay wirh anything else. I'm going back to install some sort of Debian instead. AI? Really? Nope from me.
Don't worry, I'll still watch your videos :-)
As I could not wait for Tuxedo adding Plasma 6 packages in Tuxedo OS (3.0), have switched to OpenSUSE Tumbleweed with Plasma 6. At first changes/improvements seemed not that big, but now using it for a while, would not go back to Plasma 5.27...
Hey, I know this is not the place I'm supposed to ask this in, but have you got any idea on how to connect to wifi on a Musl LLVM Gentoo install without wpa_supplicant ; nmcli ; iwctl/wifi or any simular commands. USB tethering doesn't work. I've tried to boot from a live USB of another system and copy both the kernel and the binary command files themselfes, but it doesn't seem to work. I've made sure my wifi itself works properly. Help would be greatly appreciated. Thanks!
Even when AI is trained in an 'ethical' way, whether it is ethical depends on who's using it and for what.
That's true of most technology. The same Linux kernel the drives the open source community, coul also be used in nefariously by hackers, scammers and even by the servers that run AI algorithms
The game server should detect abnormal behavior, not the client...
Totally!
B..but without a rootkit how can they safeguard you from cheaters by looking at (and totally unrelated maybe selling) your browser history, entire file system, active user time, etc? This is the ONLY way to milk you for ca... I mean.. combat cheaters.
14:59 A company will definitely try to Suyu!
Nah. Ai/ML on the desktop is mostly a stupid fad with few exceptions. Just like 3D displays were in general.
Most people I know are entirely uninterested in using anything that is 'AI enabled'.
Massive W for wayland nvidia
W-ayland.
I don't want Ai in my operating system unless i install it myself or something, I don't really trust the idea of the idea having an AI baked in to the os.
Play DOTA, it runs natively on Linux
is also just an objectively better game lol
Fr
I don't care about what Fedora decides to put in or not put in, but having a decent Open-Source LLM would be huge
I'm personally pretty excited for the Standard Notes being integrated into Proton, as I tried Standard notes but disliked the paywall, yet I do pay for Proton as I use their VPN E-Mail and Cloud Storage.
It sounds like this will fix the flickering issues I've had when I tried to run Wayland before on my 2060. Excited to give it a shot in June/July when KWin gets updated!!
As someone who is somewhat interested in AI, but does NOT want to use any that is unethical or even somewhat questionable, hearing Fedora focus on helping to develop and create ethically trained AI and other safe and privacy focused AI tools that of course would be Open Source, AND they seem to be taking their time to do it right rather than just go all bandwagon on the AI trend, gives me some hope for humanity and makes me much more excited personally! (Side note: I even plan on tinkering with an open source AI source code to make an artist/user/privacy friendly generative AI tool that never connects to the web, only trains on data you explicitly give it, along with custom made public domain, royalty free data as part of it’s factory settings, and is easy to configure and above all, easy to 100% delete ALL data you give it, and reset to it’s factory settings!)
I needed my Linux news fix, thanks.
I will NEVER ditch Linux for ANY game
I was willing to lose the ability to play half my library when I made the move in 2019. Turned out I needn't have worried, as even back then Proton was pretty damn good.
Nobara? Why not Bazzite? I think that's better for handheld PC/Console, especially as you can still follow the SteamOS model of whole-image upgrades BUT with Bazzite's Blue Build infrastructure means you can also stick whatever else you need in the system itself (say, VPN apps).
+1, Bazzite feels like a better SteamOS, eve better than Valve's, plus it is built on an intended image-based system compared to twisting Arch to be one lol
There's Fedora Silvernobara if you wanna have image upgrades. It's just a repo you add to your Fedora Silverblue in order to get the latest improvement from Nobara into your Silverblue.
Seeing people mentioning bazzite more and more online makes me happy. Jorge Castro and his team are doing awesome work with Ublue.
I'm still pondering if I want to run Ublue Kinoite or Ublue Aurora 🤔.
The Explicit Sync news was very well timed for me. I am trying to migrate from a laptop setup which would murder my back (also wouldn't allow me to use a decent keyboard) to a more desktopy one, so I've attached my old 60Hz monitor. And sadly, whereas I didn't have any issues on my 240Hz monitor, this one was just working like an absolute crap. This forced me to switch from Hyprland to Mate for the time being. Weirdly enough, the next day in the morning Brodie suddenly releases the video about exactly this issue (or at least I hope so) being fixed. Odd, but very glad I will be able to move back.
Sadly, X11 also works like garbage. If I turn off my laptop monitor, the whole system becomes barely usable (any window takes about 10 seconds to respond), so Xorg really showing its age here
I don't want AI being shoved into my operating system.
I feel like Fedora will probably have a decent implementation. I find Linux out of the box feels ancient for not having AI features out of the box. People coming from macOS take AI for granted. I've been able to mnually add AI powered features to my own Linux box, and it is a lot cooler for that. I think the best decision is if they make the AI features locally run or make it an optional component.
@@grantschilb8019 Optional, I guess is fair enough. I'm personally not a big fan of it and I don't want it shoved down my throat on Linux like it is on Windows and Mac.
@@diotitus Yeah, I wish Microsoft didn't force copilot on Windows. I like copilot, but I prefer it as an opt-in web-based feature. Apple at least makes Siri optional. Although this year, the new macOS 15 will integrate with Google's Gemini, which bothers me. Google's AI, from a technical perspective, is very dysfunctional and I hope apple doesn't give it kernel level access. If Apple oversteps, then maybe I'll stop using macOS. Only time will tell.
@@grantschilb8019 Oh yea I heard something about Apple looking to partner with Google for a lot of AI features. Is that true?
I am so happy about explicit sync finally coming soon, saying I am TIRED of this issue is saying nothing! haha I'm glad process is being done to fix this, dedication to make it a lot usable
05:10 my only issue here is Snapshot not being able to select which camera you want to use. My Laptop has an IR Camera wich always ends up as video0 and Snapshot never works because my real camera is video1.
Glad to hear a notable distro is actually looking at ai, no matter what your opinion of the various implementations of ai we have now is the fact of the matter is that its here to stay, perhaps not to the degree microsoft/google/apple would like to sell you (llms that run off remote servers being built directly into the desktop is foolish on a lot of levels...), but there are plenty of practical uses for the tech that aren't going away and the more involvement of the open source community we can get in these things the better, I do not want to see a future of ai tools that may one day prove vital to business being dominated solely by large for profit companies.
I can not speak for everyone, but I have been considering moving back to Mint or Ubuntu from my MacBook because I want to get away from GPT style AI on everything. 😅
As someone who just started using Nvidia graphics, I’m so glad there’s a fix 😂 started regretting all my life decisions.
I'm not a big fan of AI but,
I support Fedora in the Pursuit of AI (as long as it's FOSS of course).
Wasn't league already gone on Linux? I hadn't been able to run it for a very long time. All of the reports on Lutris were saying the same thing for like a year. Maybe that's why they only had 800 users, because nobody else could run their garbage.
I am really excited for 24.04
Could A.I. be able to:
Actively monitor kernel for more efficient resource usage and/or streamlining background processes?
Assist in writing use specific kernels?
I'm just curious on what ways linux could harness the power of A.I. and possibly ai/ML
No
For one, ChatGPT is exceptionally good at writing scripts and solving problems on Linux today. I keep hearing complaints from Windows people trying Linux - that they often find themselves stuck in BASH trying to look up and remember commands for doing certain things. This could help them learn, or at least automate things by accepting human language and writing the scripts for them...
Software names can also be a mystery to newcomers: it could help you choose software based on the needs you describe to it and figure out the difference between the different software; solve problems you encounter trying to do something... Sure you have a search engine for that, but this is more of a hands-on approach. :P
Not vibing it with the default desktop backgrounds? You could ask an image generating AI to make something more appropriate for you. I would.
An AI layer could be interesting IMO, though I don't think I'd have enough resources available locally to make good use of it. I WOULD like a standardized API for such things so that I would have a choice of whether to hook to ChatGPT, some other service, or even a cluster of laptops at home - that's what having AI on Linux would be all about, compared to the proprietary shit...
@@GrumpyDerg Some very solid points
10:10 actually, it's through a config file only now.
I, for one, am glad to be behind on the AI hype train. But it certainly is good to remain open minded about it!
Fedora pushing for AI makes me nervous ngl
That AI thing would also be useful to help you with using the terminal and cli-only programs. Many people are still afraid of it, so an AI helping you through it would be nice (would also eliminate falling for the rm -rf trolls)
I'm usually not advertizing, but I just have to. I've got one of these monster Stellaris laptops. It's as good as you would expect it to be. It has a few downsides, mostly related to the firmware (slow boots) and Nvidia. But I knew what I get myself into there.
What I didn't know is that you can actually control the backlight of all keys individually. There is no software for that (afaik), but it's easy to do (echo r g b > /sys/class/led/...). Can't wait to find the time to integrate that into my keybindings
I switched from my RTX4090 to my 7900XTX to test Fedora. Good distro, very stable and coherent.
As a former League player for very years i know i speak for many that people being forced to leave LoL is good news
I started using Siver blue after year of Fedora. Installing drivers was a bit shit but it so stable and faster updates via ostree the dnf. AI I am leaving to SUSE Micro Aeon if only I could work out how to install codecs extra on it. That said haven't use SUSE for years. DNF is just easier to use for me. A yrst on Gome is janky as hell still. While my aim is in making gnome setting cover what you need, then running one distro only software. Also fail to find a way to install Waydroid on Suse Micro yet.
im using nvidia-open , very similar to proprietary , flickering is only in games and apps that use xwayland, native Wayland has no problem
12:05
Welp, i guess that a part of the LoL players ares gonna take a shower and touch grass for once since a little while
I use both Proton and Standard Notes. Hope this is good news...
Really like your Channel and the News. A year ago I was still using windows and lots of proprietary software, but you got me to give linux a try and go (almost) fully open source. Started that 6 months ago and found my way (currently with Kubuntu) and really like it. As a "proud" owner of an Nvidia Card (thx to being a long term windows user) I'm really looking forward to wayland working correctly with my Nvidia card
1:41 I don't want Fedora then.
It's just an opinion from Mathew Miller, in order to get an AI implementation like a local model running in a workstation iso for whatever, it should be presented as a proposal with a definitive solution, let's say, a client that let you download a llama2-3B-Fedora model, as a very specific example. Don't think Matthew Miller dictates everything on the Fedora project, there are proposal, discussion and elections within the Fedora project community.
So i changed to linux to not have AI embedded in OS and they want do it now anyway...
And Ubuntu 22.04 got a STIG from DISA in April (2024). So much more security for everyone & also for the people DISA works for. In a more open context, CIS benchmarks are doing good things for the large corps utilizing open OS & select apps.
AI can be used to characterize you with behavior without the details. It’s always free. It’s always nice. It’s always good in the beginning with everything that we’ve had so far. But I assure you this time will be different!
I am facing a lot of flickering issues on wayland in my setup running tumbleweed and rtx 3080 (desktop). More consistent issues in chrome and electron based apps.
Artist transitioning to tech, disillusioned and disappointed by the ridiculously over-hyped and unethical 'content' generating AI we have now. A solution to a nonexistent problem built on theft in a black box. I've never hit a subscribe button harder than after hearing "public doesn't mean free to use for every use case." Thank you.
Onlyoffice and Zoom really flicker under Wayland + Nvidia, which is really unusable, glad something is being done about it.
As for the AI stuff, ya it needs to be worked on for AMD Radeon GPUs but some of the problem is their published documentation which may or may not have improved. Then there's the upcoming RISC-V hardware and also things like Raspberry Pi.
I need explicit sync also on alsa for usb audio devices!!
Explicit sync is a nice to have. Doubly so with it being the way Vulkan wants to handle things. Should help a lot.
While I don't care about AI, I cannot deny that it's there and it's not going away. It'll be a good thing to have an open implementation about AI. If it's gonna happen, then we might as well get ahead of things and be ready.
Rest in peace League on Linux. I've had my fun with the game, but I decided to get rid of it when the Vanguard announcement came and I do not miss it.
Au for what exactly? To clear the ram?
I want local AI with the ability to create multiple "instances" that are each based on data I provide it, and 0 censorship.
Hopefully Fedora AI doesn't turn out like Nouveau where it's there to exist and adhere to the morals of open source nature but doesn't end up excelling in a single area compared to proprietary counterparts.
When do you think we can expect an AMD driver with "Adrenalin Software"? There are some features that I miss under Linux...
Fully agree on No Kernel Access for something as frivolous as anticheat (that doesn't work 100 per cent anyway!). I'm okay missing out on badware. Thanks, Nick! 🥰 🐧
I see AI as a huge opportunity for copyleft, so long as it is taken advantage of. The copyright aspects of AI training are a huge roadblock for most models, but that doesn't apply if explicit permission is granted in the license to use the work and make derivatives. That means that if an LLM is trained on the vast corpus of CC-BY-SA and compatible work, its outputs would all need to be CC-BY-SA. Same goes for image generators. Coding LLMs too, substituting CC-BY-SA for GPL. While it would still cause fragmentation around the different incompatible Free licenses, that's still a huge improvement over the situation with closed models, where the data all needs to be either Free but non-copyleft or specifically (expensively) licensed the way that Adobe is doing with Firefly or Google is trying with Reddit. Imagine what it would mean if the most powerful tools for writing code meant that all the code had to be copyleft. But none of that can happen if it's just ignored.
I'm sorry if this was an incoherent mess, I'm running on three hours of sleep.
I've pondered this issue as well, and I've come to see intellectual property as a house of cards propped up by profiteering lobbyists. The expectation that a machine learning model must strictly adhere to the licensing of its training data mirrors the notion that a human artist drawing inspiration from existing work must conform to the wishes of the original creator - an unnecessary and restrictive constraint on individual autonomy.
@@noderunner_ That's certainly a debatable point, but the fact is that *at the moment*, the law is not clear. For all we know, the outputs will all be public domain. This is just one possibility, and I think that the potential upside is worth the risk. Even if it doesn't work out, we still get an ethical AI, which we don't really have.
@@skywz There's a prevalent sentiment that Machine Learning is deemed 'unethical' when models are trained on copyrighted data. Advocates of 'ethical AI,' pushing for more stringent intellectual property laws in ML, risk steering us into a dystopia where generated content becomes monopolized by corporate entities under complex IP regulations. This stifling landscape will impede individual freedom, leaving only those with the resources to navigate legal complexities to benefit from this transformative technology.
@@noderunner_ I've actually made this argument myself in person. When I say "ethical" I mean that it is ethical by promoting copyleft works through its output and ethical by being local, private, and transparent.
@@skywz I certainly align with those values, not just in AI but in all software.
They're saying that our Linux systems are hard to install a rootkit onto for their anti-cheat... I say that's a feature not a bug.
13:14 they didn't say they had 800 daily linux players. They said they had 800 linux players "yesterday", which seems insanely disingenuous when you think how often companies use "statistics" to "lie". For all we know that was also only on a specific server, which they happily just decided to ignore. This also comes with the fact that linux players already knew about vanguard, so they probably stopped playing as much.
Don't be giving me a heart attack. Windows 11 forcing AI that I have no control over is one of the main reasons for switching to linux. They better not be forcing that shit on us in LInux as well. I love AI that I can localize and control, but anything else being on my system against my will is a no go. Very concerning from somebody who loves AI and Linux. Also...I bet if you were a portrait artist in the 1800s, you would have been furious and fighting it every step of the way huh?
Gentoo?
Found the digital artist
@@-KillerHawk- For over 15 years yeah, and an artist of many other types for over 30. And programmer, father, philosopher, buisness owner, and trained soldier. I am also a musician and game developer. I guess we found what kids today are calling an "NPC". Also spent my teen years on the streets involved in organized crime. Can you please explain with your wisdom, who I am?
@@-KillerHawk-found the nonsequitor.. Who cares what anyone does for a living? Having a constant data feed to some company hosted "AI" (not actual AI..) is a bad idea period. They're not worried about running out of training data when people are literally paying them to harvest their data while using the AI.
It's Linux, it'll be optional.
Good that Fedora does not shy away from taking the heat. We definitely do not want to lag behind.
We should be the ones to bring the new things instead
I just wish that
Well first no AI built into my OS, but if it is, that it is opt-in and not opt-out, that and easily uninstallable
The Steamwebhelper crash is still not fixed that prevents using Steam on some Distros.🙄
I must admit I never encountered that game on any distro I ever tried gaming on
@@TheLinuxEXP What are you talking about ?
Does Nick go into why they don’t like AI? I would like to hear their take as I have mostly optimism for the trajectory of generative AI in its current state
Ending music is very French. I love it!
Kernel-level anti-cheat really isn't the solution. But I wish Linux would start to implement tools that provide applications with a similar level of trust in the system they're running on, without being as invasive as a rootkit. Things like a userspace API that prevents other programs from hooking into certain processes, or a cryptographic system to verify that the kernel comes from a trusted source like the distro's official repos.
Aww look at you coming up with actual ideas for real anticheat solutions, hahhaha.. Don't tell me you really thought their rootkits were for anticheat... These companies have never been interested in stopping cheaters, those are their best customers.
Client-side anticheats never work.
Those are good ideas in theory, but an API call can just be removed from the code. A cryptographic system could be spoofed and would prevent people from legitimately working on kernel stuff.
And of course, kernel anticheats can also be disabled/spoofed and are useless too.
The best anticheat solution is to be cautious of what power and data you give to the client from the server.
Wallhacks wouldn't be an issue if the server was smart enough to not send info of players you can't see and can't hear walking.
Fly hacks/item hacks in games like Mario Kart for instance pretty much only occur due to the game having a terrible architecture that lets the client do way too much.
Aimbots are the toughest to fight against but even kernel-level ACs do a bad job against those currently. The best solution for those is probably behavioral analysis like reCAPTCHAs work.
Of course yeah you never will be able to get rid of all cheats, but you'd reduce significantly their impact. Like you could have a lesser version of a wallhack that renders sound sources, but it would be pretty limited in scope, and is already present in some games as an accessibility feature. Or a lesser aimbot that only works in screen-space and that would have to make a significant effort to look legit. On Minecraft for example, you can't prevent people from installing Xrays, but what you can do instead is to make them useless by feeding the client with garbage data on blocks they cannot see.