The LLM tech bros all talk about “frontier models”
and it strikes me that this use of settler colonialist language is surprisingly appropriate.
They’re enclosing the commons of our knowledge and our culture. They’re putting up barbed wire, so they can make us pay rent for everything.
@slothrop The language they use tells you exactly who they are.
I wish I had understood and been able to get more people to understand this when Homesteading the Noosphere was published.
@dalias @slothrop What's currently marketed as "AI" is just a tiny fraction of what computer scientists and software engineers are doing with ML, and ML is just a tiny fraction of the entire research field of AI. There is also symbolic AI, which is human-readable right from the start, like systems designed for natural language processing using carefully built knowledge graphs instead of automatic data mining. There's the field of evolutionary algorithms, self-modifying code that gets selected for how good it is at solving problems. Also, autonomic robotics. All that kind of stuff.
Generative AI, like diffusion based models and GANs (e.g., image generators, 3D model generators, video generators, audio generators) or LLMs (chatbots etc.) can also be quite useful for many things. It's just that the sheer scale at which they're pushed into anything is completely bonkers, and that's just because a whole bunch of investors has sunk immense sums of money into it thinking they were going to get immensely rich, and now they're desperate to sell it.
We just need to take the big money out of AI research and take the AI models out of the hands of the rich. And we will have to live with the fact that it is just as impossible to trust anything you haven't witnessed yourself anymore just as it was before the invention of photography and audio recording, but I totally expected something like that to happen sooner or later ever since I got my first multimedia upgrade for my 286 back in 1989. It used to be hard and take a lot of work and expertise to modify or fabricate natural looking digital images, audio, and video, and now it's all automated, and that technology isn't going to vanish as long as there are computers. Everything a large computing centre can do within seconds, a desktop PC can do within minutes to hours, just like back in the day my 286 could render a photorealistic 3D scene in a day or two. Nobody can squeeze toothpaste back into the tube.
@LordCaramac @dalias You would just like to interject for a moment, wouldn't you?
Bye.
@slothrop Not just extract rent: They're building slot machines where you pay by the pull:
Shot: Everything about “The Man Who Killed Google Search” is fantastic except for the title. Under blamelessness, we recognize that given the systemic incentives and affordances, if it wasn’t Prabhakar Raghavan, it would have been someone else: https://www.wheresyoured.at/the-men-who-killed-google/ Given Google and Capitalism the enshittification of search to incrementally goose revenues was inevitable. 👇🏽
@mark JFC is it the heat, mate?
@slothrop Sure, but I don't get the metaphor. Settlers took commons and parceled it up for exclusive use; "You can't use these woods, they're my woods now." How does one do that with LLMs when the primary data is still there, reachable the way they reached it? The fact Anthropic read a bunch of books didn't make the books disappear, nor did image generators being trained on DeviantArt's contents make DeviantArt go away.
I don't get what the fence represents here.
(Sidebar: wow, but how timely and appropriate is that image Wikipedia has right now to visualize "metaphor?" Democratic party can't even catch a break historically!)
Agreed — which is why we need to actively resist the enclosure. Here's a solid example of how to do that: https://tldr.nettime.org/@asrg/114742667183459482 via @asrg
Attached: 1 image **"Trapping AI" – Slight Update!** 🌀 Activity in the **"Trapping AI"** project is accelerating: in just under a month, over **26 million requests** have hit our tarpit URLs 🕳️. Vast volumes of meaningless content were devoured by AI crawlers — ruthless digital leeches that relentlessly scour and pillage the web, leaving no data untouched. In the coming days, we’ll roll out a new layer of complexity — amplifying both the *intensity* and *offensiveness* of our approach. This escalation builds on `fakejpeg`, a tool developed by @pengfold@social.ty-penguin.org.uk. 🖼️ `fakejpeg` generates fake JPEGs on the fly. You "train" it with a collection of existing JPEGs, and once trained, it can produce an arbitrary number of things that *look* like real JPEGs — perfect for feeding aggressive web crawlers junk 🗑️. Explore `fakejpeg`: [https://github.com/gw1urf/fakejpeg](https://github.com/gw1urf/fakejpeg) Learn more about *"Trapping AI"*: [https://algorithmic-sabotage.github.io/asrg/trapping-ai/#expanding-the-offensiveness](https://algorithmic-sabotage.github.io/asrg/trapping-ai/#expanding-the-offensiveness) See the tarpit in action: [https://content.asrg.site/](https://content.asrg.site/)
@kris_of_pnictogen @slothrop Oh wow. Yeah, taught me a new word today.
Agentic as in "agent." This is describing AI that will be able to do its own planning and execution, whereas the current generation just responds to input. So, comes up with its own questions, creates its own sub-goals and acts on them.
Oh good. CEOs are trying to automate VPs. Nobody tell them that a CEO is just a fancy kind of VP, yeah?
Absolutely. "Frontier" means 'the place where no rule or force can stop us'. Thats all that word has ever meant.
Rocketman
Quincy
Cassandrich
Lord Caramac the Clueless, KSC
tiny_m
Mark T. Tomczak
Reg Braithwaite 🍓
Kevin Russell
feld
fedops 💙💛
Nini
Manic Pixie Dream Gremlin
Rossana Trotta
Kris Dreemurr (they/them) of the Pnictogen Wing
tom jennings