Furries have already achieved the dream of every shit silicon valley technology and in a more ethical and less cringe way The metaverse: VRChat NFTs: adoptable fursonas Typing words into a text box and having art appear: commissions

Yeah, this pretty well describes my own experience with LLMs. They are genuinely groundbreaking as a learning tool, provided that you actually use them as a learning tool...which vendors do not encourage (my god, the glazing) and which hardly fucking anyone does. www.anthropic.com/research/AI-...

was epstein uniquely responsible for evil or is he just the one obscenely wealthy person whose emails we're getting to read

Worth remembering that something did happen: the journalist who revealed this died from a car bomb.

In the context of artificial intelligence, the term "world model" describes the ability of an AI system to model the world it operates within. For example, your world model as a human is how you know that if you drop an object that you're holding onto, it will then fall to the ground.

My pull-string Sheriff Woody doll said “there’s a snake in my boot” when there was not in fact a snake in his boot

remember, this is only the sprawling elite pedophile cabal that we know about

That's one of my favorite and most enthusiastically replayed games of all time you're talking shit about there

My attention was drawn recently to this paper. Some ML folks not only identified the fundamental design issue with LLMs which I've been going on about for a while, but it seems they also identified a way to get past it. Hm. The chatbots might be getting a lot smarter soon. arxiv.org/pdf/2412.067...

an often overlooked casualty of this whole scam is how gen-AI corpos have flattened "artificial intelligence" to mean LLM garbage neural networks, behavior trees, freaking path finding!, were all once under the umbrella of "AI", before it became synonymous with IP pillaging & chatbot sycophancy

There are two kinds of people: Those who tend to blame things going wrong on things outside of their control, whose coping strategy in a chaotic world tends toward recusing themselves, and those who tend to blame things going wrong on themselves, who cope by believing they're the ones in control.

Here's a few examples of what my 5 megabyte (1.4M parameters) language model wrote, after it was trained for about 20 minutes with my subpar GPU and 50kb of my own bsky posts (and absolutely nothing else)

Preface: More than anything else, I did this for the simple joy of doing something creative. I had fun! I haven't made anything *quite* like this before. But I also did it in the hope that Sos and others might learn something as well. So, please allow me to take you on a brief journey:

Wanting to believe that all those folks who use an LLM chatbot and their takeaway ISN'T "wow, this chatbot is really dumb and gets almost everything wrong" are secretly being A/B tested with a way better LLM. Since the alternative is believing that 50%+ of people can't tell the fucking difference...

Sufficiently advanced stupidity is indistinguishable from malice

This is a public service announcement: Any discussion of "AI safety" not including CICERO among its principal case studies should be disregarded as unserious horseshit. CICERO was 2022. Imagine what similar tech must be in use for astroturfing now, in 2026. www.technologyreview.com/2024/05/10/1...

Folks forget real easily just how gravely fucking concerned we should be about the network-connected always-on phone cameras and mics around us all the time www.techspot.com/news/111082-...

I stumbled upon this guy on YouTube who uploads EDM remixes of Rock/R&B songs set to stock footage and has been doing so for years, and they freaking slap. This one just landed today. Dear Matthias, please let me pay you for high quality versions of your remixes. www.youtube.com/watch?v=c5E7...

ICE agent asked why he's taking pictures of a legal observer's car, replies: "Cuz we have a nice little database and now you're considered a domestic terrorist. So have fun with that."

From reddit: www.reddit.com/r/greenland/...

I must find some time to try this

This is a public service announcement: Any discussion of "AI safety" not including CICERO among its principal case studies should be disregarded as unserious horseshit. CICERO was 2022. Imagine what similar tech must be in use for astroturfing now, in 2026. www.technologyreview.com/2024/05/10/1...

Here is a very good talk on how the world (and particularly the internet) might be a better place if everyone would do away with the digital anti-circumvention laws that the United States pressured other nations to incorporate as a part of trade agreements www.youtube.com/watch?v=39js...

Today I finished my re-read of Frank Herbert's six Dune novels. I've never read or seen or heard anything else since my first read some 15 years ago that took Dune's mantle as the high water mark against which all other art and media is judged, and I'm increasingly doubting that anything ever will

Am I really gonna have to spend a second day in a row talking about how our media institutions are failing to meet the moment? Because Jesus fucking Christ, Reuters

Never ketchup. The best is vinegar and old bay seasoning (I grew up on boardwalk fries)

you know that thing where nature keeps evolving things into crabs? the youtube algorithm does that but by making every video title and thumbnail look like it was made by mr beast

Angular lerping done wrong in production

For a while now, genuinely, there's been no one else in journalism who I hold in as high regard as @gamersnexus.bsky.social Thanks, Steve, and thanks to all y'all with GN for being the literal one and only substantial source of news that has yet to feel to me like a Gell-Mann tarpit

I'm not familiar with Look Back. But what I do know is that I have never seen a bad movie by Hirokazu Kore-eda, and I have seen several I really loved. Hell yeah

I object to the term "AI" (I would fall into the "mathy maths" camp mentioned in the article) but this is a good read regardless

Mark my words: I'm going to publish a game on Steam and when I do I'm going to insist on writing about NPC AI and procedurally-generated content in any "AI" disclosure, purely out of spite

www.huffpost.com/entry/us-cit... "They’re denying that any of her birth certificates, which are from Laurel, Maryland, her records of immunization, medical records — they’re denying the authenticity of them. It is something I’ve never encountered."

In home alone, the same actor who plays one of the bad guys plays the cop at the beginning of the movie.

mozilla's problem is leadership — not any particular leadership, but leadership as a concept. the C-suite assume they ought to be leading, when actually they had nothing to do with firefox's success in the first place, so they should shut the fuck up and let the nerds handle it

sometimes I think games just aren't for me anymore. then I play old games and nope they're a lot of fun

It can be very challenging to communicate ideas & intentions to concept artists in a way they can work with. The problem w/ using computer-generated images just to help communicate w/ artists isn't that it'll compromise the art. The problem is how evil the corps selling those image generators are.

"Some people never observe anything. Life just happens to them. They get by on little more than a kind of dumb persistence, and they resist with anger and resentment anything that might lift them out of that false serenity."

Personally I am getting real tired of people acting like the problem is some nebulous ill-defined "AI" and not a few dozen or so rotten individuals with massive lobbying budgets. You can't fight the abstract idea of computers doing statistical modeling, but you sure can throw some folks in jail

Drought and Flood #fanart #art #Illustration #pokemon