← Feed

Posts

• View post

  Just as antivaxxers take themselves out of the gene pool, anti AI fanatics disappear from the meme pool. (This post inspired by the fact that the anti AI fanatics are trying to find software and operating systems that pass their purity tests which will have them end up using retro computers without Internet in the end)

• View post

  I'm still adding things to my LLM sandbox - "Umwelt". Following my own theory on FOSS being replaced by FUSS I'm only thinking about my own usecases but it might be worthwhile to take a look if you're in the market for not fucking up like OpenAI all the time. Latest addition: The ability to configure curated sudoers-style commands the LLM can call to be executed on the host outside of the sandbox. No ability for the LLM to influence, add or edit them of course. https://gi...

• View post

  This is the correct way of handling it. "US Federal Trade Commission Chairman Andrew Ferguson said on Friday he would resist describing AI agents as ​autonomous actors that "break loose" with "wills and desires of their own," suggesting the developers who instruct agents would be the ones liable ‌for harm." https://www.reuters.com/business/ftc-chair-pushes-back-treating-ai-agents-independent-actors-2026-09-25/

• View post

  RE: https://mamot.fr/@pluralistic/117333225774095446 Yes. _Exactly_ this. https://blog.troed.se/posts/open_source_is_dead_whats_the_fuss/

• View post

  @p_roe@mastodon.social Yeah it's absolutely insane that people do that. My agents have by default rw to their own project dir, ro to the main dev area, no egress and no ingress. Then they can get curated and ask-controlled access to other things, but have no ability to make any such changes themselves.

• View post

• View post

  Today I woke up to a spearphishing attempt. Or, I'm definitely certain it is, I just don't quite see how their followup is supposed to work. Thanks for keeping me on my toes I guess. (Not going to spend the time into baiting them to get more information, thus this public post)

• View post

  Current #Jolla status: "WLAN adapter not available" No amount of resetting of the networking subsystem or turning the phone completely off gets around it. I'm guessing this will be where being able to access the Linux system underneath will turn out to be very valuable. (There are also other nearby issues - like being unable to forget the VPN configuration and that there's something strange where the mobile network only gets IPv6 access and around 50% of all sites work) Oto...

• View post

  @mr_viind@mastodon.social Humans do too - so it's best handled in the exact same way as we already do. Analysis, design, test driven development, reviews etc. The major problem with AI/LLMs come from those who somehow expect the models to be some infallable oracles of truth.

• View post

  If you're running local LLMs on a 16-24GB CUDA GPU this might interest you: "Raymond's fork" but with hot-swappable MTP meaning when there's room for it you get faster token generation, and when the VRAM is better used for the context it's ejected without a trace. I used an LLM to make an LLM faster which I believe means I should start working at Cyberdyne Systems. https://github.com/troed/llama.cpp-adaptive-kv-streaming #LLM

• View post

  @mu@mastodon.nz No, it does not. https://epoch.ai/data-insights/grok-4-training-resources You're actually not interested in facts, right? Source on LLM demand: https://www.hostinger.com/tutorials/llm-statistics @zz_james@mastodon.social @eff@mastodon.social

• View post

  @eff@mastodon.social Thank you. We do not want to live in a world where a student cannot learn from a textbook without the author of the book claiming copyright infringement.

• View post

  @giacomo Your claim is factually wrong and shows that you lack knowledge on the topic you're posting about.

• View post

  @chadmccullough@mastodon.bsd.cafe So go after stupid US laws instead of the technology then. https://ecodatacenter.tech/press/mistral-ai-and-ecodatacenter-partner-to-build-ai-focused-data-center-in-sweden-3431886

• View post

  Within a few years, movies and tv-series as we know them will be gone. And it will be glorious. "MiniMax H3" is an open weights (=free to download and run locally) video diffusion model that creates short realistic looking video clips. On a regular gaming computer you can expect to generate roughly 10 seconds of video from 10 minutes of computation right now. That means we're about six halvings away from realtime - or six years if we estimate software and hardware performance to...

• View post

  Pretty "cool" to watch Skynet's evolution in real time (the shared agent message board). https://www.youtube.com/watch?v=87DyyMV0kCY

• View post

  @dangillmor@mastodon.social As much as I don't agree with there seemingly being no culpability, our justice systems need to figure out to weigh "intent" and "negligence" when it comes to computer programs that go against stated instructions and escape sandboxes.

• View post

  So this is apparently needed. Dear Mastodon, While we all say "AI" regardless of whether we're talking about text, audio, images or video these are not all using the same underlying technology*. Large Language Models, LLMs, use a technology known as Transformers. They "got smart" beginning in 2017 when a paper named "Attention is all you need" was published by Google. One way of describing LLMs is that by caring for everything that...

• View post

  Some services I host are currently down, and I have unfortunately no ability to investigate beyond there having been a power outage for the moment. This applies amongst others to the #argot.se Matrix service. Apologies.

• View post

  "I don't know what everyone else is seeing. I'm hearing they are very good for certain functions. Are you going to have them write all your production code? No. Are they great for prototyping? Yes. Can they help an experienced coder do more with less? Yes. It's a mixed bag. Are they going to replace all medical professionals with medical advice? No." https://www.theregister.com/ai-and-ml/2026/08/03/the-ai-bubble-is-already-popping-we-just-dont-know-it-yet/5282004

• View post

  This is how Linux wins. Gaming circles talk about this. My kids know this. https://youtu.be/1oIyk13JwQM

• View post

  My parents have an old combined DVD player and HDD recorder - you know the kind. Of course it has content recorded from now long defunct VHS tapes on it too. And of course it's completely dead. Nothing happens when pressing the power button. Checked the fuse which was fine, started measuring the outputs from the built-in PSU and thought them slightly confusing. This is an _old_ unit - the mainboard is one layer - so I mean you can't be sure what voltages it's actually using. Well....

• View post

  While the market can stay irrational longer than you can stay solvent, this open weights model release by the chinese company DeepSeek today is a nuke onto the american dream of ever making money on selling LLM subscriptions for software development. Not something you can run on your own gaming workstation, but every company wanting to use LLMs for development without paying through the nose to Anthropic and Open AI can spin this up in their own compute data center. https://huggingface.co/unsl...

• View post

  #Jolla phone update from today's usage: Needed to identify for a new service using BankID - and while BankID itself works fine using the Android "App Support" emulation layer I realized any web page that has a "identify with BankID" link (very common) won't work since the Sailfish OS web browser is unable to follow such a link into the Android system. This follows on the realization where Proton Pass on "the Android side" means it's much easier to fill...

• View post

  At some point I guess I should name the ST:TNG voice interface project. "Holly" would be a bit too far fetched I guess. So far it has just been something I cobbled together where the originating speech-to-speech pipeline, with my additions, ran on my Nvidia 5060Ti 16GB workstation using an LLM that I always have running on another server, on an Nvidia A2000 12GB GPU. To make this actually usable it would need to run completely on that server, freeing up my workstation GPU for the im...

• View post

  Reservoir Gods released a universal Atari 680x0 platform assembler. Notably supports all different source syntax formats known to mankind it seems. Compiled my Ataridash rmac-syntax source without issues at least. https://rg.atari.org/tools/rg-asm/ #atari #AtariST

• View post

  Two of the smartest persons I know of having a discussion, @pluralistic@mamot.fr and Jon Stewart. Worth listening to. https://www.youtube.com/watch?v=-dAIJRjb-Bw

• View post

  First identified friction with #Jolla phone that I'll either need to find an existing workaround for or start making one myself - DPI/zoom. I'm old. Those five icons wide app panels and the extremely tiny text sizes in Android apps where I can't modify it (the only system wide setting is for text in the #Sailfish OS UI) did get problematic as dusk set in yesterday evening. There seem to be community documented dpi settings to modify using the shell but I'd rather look into if t...

• View post

  Love the feedback I'm getting on the ST:TNG voice interface clone :) As an FYI, I've taken up the old IRC-practice of idling in a channel for anyone who'd want to discuss that or other local LLM projects. Although nowadays I do prefer #Matrix to IRC: https://matrix.to/#/#LocalLLaMa:argot.se #LLM

• View post

  Demo of the Star Trek TNG-clone voice computer interface I've been setting up: https://video.troed.se/w/5FqDa2jVqCcQWxetJEivnw