" We been brainstorming various ideas in the team for how we can make use of AI to provide improved or new features to users of GNOME and Fedora Workstation."
I’m not crazy about the idea of anyone’s AI embedded in my DE or OS. Thoughts? Would the OM devs remove the AI from GNOME?
Another reason I don’t Gnome.
I would hope they would remove it. If it was possible.
An AI in an operating system should be the users choice and it should be local to the device. No call backs to some faceless central control data center doing whatever they want to do with your data.
This whole “AI” craze has been nothing more than a glorified data vaccum.
I think they’ll be getting a lot of pushback if they embed it in the DE or OS. Easily available, fine. Embedded, absolutely not.
As long as the code is open-source, I can see this as a good idea. I often use local large language models with an open source program based on llamacpp (there are many different ones, but I don’t use Ollama). There’s a huge variety of models available on Hugging Face, even tiny ones that can run on CPUs, so there’s no need for an LLM in the cloud, which would naturally be used for data collection of various kinds.
There was a project which sought to emulate Microsoft’s Windows Recall as an open-source project. I found the idea interesting, but there’s been no commits to Openrecall for at least 5 months. I assume it’s abandoned. There is Screenpipe, which I just found out about. As far as I’m concerned, as long as the project is fully open-source and uses local models, it could prove to be a useful tool for those who need it.
I agree that I do not want embedded AI in my OS.
I am also concerned about the future of hardware with built in NPUs that cannot be removed and how that will generally effect the Linux and various BSD world.
I am SO looking forward to seeing the AI Bubble implode.
Maybe the GNOME Project needs some artificial intelligence because there’s no actual intelligence to it’s design and engineering… 
(maybe)
The point for me of a Linux distro is that it doesn’t move unless I say it moves.
AI beats the point. Sure, having AI tools is nice, but having AI key components of the system, even if userland, is just a big no.
It won’t be mandatory. These will simply be available tools that if someone wants to, they can turn them on. Nothing more than, like with the AI plugin for Krita, if you want to, you can install it, if you don’t, then no.
I also can’t wait for the AI garbage to crash. So many places wanting to force it on users. and I am so so sick of having to catch the user-agents of the AI scrapper bots hitting my sites causing me to have to stop what I am doing then go in create a new UA bad-bot block. Then on top of that trying to get a new site to get any traction in search results cause of the AI slop spam. then google pushing the AI slop youtube channels. AI can just die.
So Deepseak is the newest AI LLM everyone is gargling the nuts of because apparently it’s so totally awesome. So I added it to my Ollama server and asked it for a brief biography of Anakin Skywalker, it was just the first thing that popped into my head. The results are, um, well, interesting. I swear on my grandmother’s grave I didn’t change a word.
deepseek-r1:latest
Thought for 27 seconds
Anakin Skywalker Biography
Early Life: Anakin Skywalker was born as the firstborn son of Count Dooku and Princess Leia Organa on Felucia. Born into the Imperial Court, he lived a sheltered life until he turned 19 when he entered the Clone Wars to protect the Resistance.
Rebellion: Encouraged by Obi-Wan Kenobi, Anakin joined the Rebellion against the Galactic Empire. His initial role was as a bodyguard to Princess Leia before he answered the call to fight for freedom.
Rise to Power: During the Clone Wars, Anakin faced Darth Maul in “The Empire Strikes Back,” where his skills earned him respect among rebels but also drew attention from the Empire. After being captured and taken to Felucia by Count Dooku, Anakin secretly rebelled against him, leading a group that overthrew Felucia.
Downfall and Return: Anakin’s rebellion led to his betrayal and execution by Dooku’s men. His death marked a turning point as he was revived through a cloning process, placed on Mosset 4 under the care of Grand Moff Tarkin.
Final Chapters: As a Sith Lord, Anakin served Count Dooku but eventually realized the darkness and sought redemption. In “The Force Unleashed” and “The Last Jedi,” he faced Darth Vader’sClone army in a pivotal battle where he killed Palpatine before his own defeat by Vader.
Legacy: Anakin’s journey symbolizes the struggle between light and dark, with his ultimate sacrifice highlighting the cost of resistance. Posthumously canonized as a Sith Lord, he remains a tragic yet inspiring figure in Star Wars lore.
I’m confusion.
My prediction for 2027:
GNOME requires at least 4 high-end GPUs with 2 TB RAM each to even start, because otherwise the AI that automatically determines what desktop background to use (let’s not confuse the user with configuration options) doesn’t make the decision quickly enough.
GNOME users say that this is perfectly normal because the Gods at Crapple do the same thing, and accuse KDE of being ass-backwards and decades behind for caring about working on cheap hardware.
GNOME receives millions of $$$ in donations from GPU hardware makers and the likes of Red Hate and Canonical that will sell an expensive support package for every computer newly bought to meet the new requirements.
My prediction for 2028:
The AI - assuming it is taken from a proper open source project and not written by the GNOMEs themselves [in which case it would certainly become ASS instead: Artificial Supreme Stupidity] built into GNOME has gathered enough data and learned enough about computers to make the right decision - instead of switching the desktop background, it runs rpm -qa |grep gtk |xargs sudo dnf -y erase ; sudo dnf install task-plasma7.
GNOME users of course have no idea what the AI did for them, and go on the internet to talk about how great that latest GNOME update is.
Back at my old job they had an AI bot running on Slack. I tested it with something I knew there wasn’t an answer to (the track listing of a fictional band’s 4th studio album). In canon the series ends when they graduate high school, and even my unfinished fanfiction of their later career doesn’t have a track listing.
The bot sat there hung on the question, churning away for ten minutes or so until it finally regurgitated a completely irrelevant answer. Unfortunately haven’t gotten ChatGPT to barf on the same question.
Would be handy if we could use the solution Number 6 used in “The General” (episode of “The Prisoner”).
Way in the early days of GNOME 3, when I could see the direction the project was going, I made this mockup of what I thought a GNOME 4 desktop would look like:
Might need to re-do it as GNOME 5…
What the f…???
If they have common sense, they would provide maintainers with an option to remove the AI at compile time or an “ai” toggle that uninstalls the necessary components when toggled off, but I can also SMELL red hat’s money sliding on GNOME’s bank account for integration with their “open source” models with a solution for a problem nobody had. Yeah they are truly open as the training dataset is open source and public, but normal models suck, local models suck even more, local & open models are just a pure joke. Local & open is good, don’t get me wrong, but AI scales with training data quantity and quality. I think AI has potential on the desktop, but not red hat’s AI with red hat’s biases, not facebook llama ai, not the CCP deepseek AI. In a few years we’ll see real AI, for now its still a tech demo, and then by 2030 we’ll probably have a decent open and local AI with hardware that can actually run it. Not today.