Austin being the testing ground for "AI Cars" means testing the logical endpoint of the Traffic Capacity Problem:
What if all the cars were empty and still causing traffic jams?
Something to understand about the AI freaks. It's not just the money fraud aspect that's making them push it so hard. These are people who completely lack the ability and will to do anything that isn't going to immediately generate coin. This is truly a revolution to them. bsky.app/profile/davi...
So can we talk about how stupid Dems allowed Ellon Musk to buy twitter at the same time they paved the way for Larry Ellison to control Tiktok and this sort of shit didn't just handled the US to the fascists but it has also already ruined Germany and Brazil?
Netflix has told several people going forward with new pitches that they plan for all animated shows from now on to be AI. I've heard this more than once now.
I think writers can be a funny bunch. "Show don't tell" is a strong mantra. But then they'd do shit like letting us see a character's dreams and nightmares.
So, anyway, I had a weird dream.
I was in some sort of MrBeast race. And I actually did my best in the first phases. Like having to go through a long road by yourself. And other details are fuzzy.
The thing is that it is looking like I definitely going to finish the last phase.
My new goal was to go back home and it was going to be a mess because I have to do the whole trip in reverse but oh well.
So I wake up and I was like. "That was really stupid. Why would I even join a MrBeast race. That's dumb".
Anyway, some time later as I examine the dream I realized: I DID JOIN THE MRBEAST RACE
It's a metaphor for my job.
It makes sense, my coworkers were there.
So, anyway, I had a weird dream.
I was in some sort of MrBeast race. And I actually did my best in the first phases. Like having to go through a long road by yourself. And other details are fuzzy.
The thing is that it is looking like I definitely going to finish the last phase.
There's a cut to MrBeast announcing the winners and it's fucking stupid bullshit like Linus from Linus Tech Tips and other bunch of rich dudes and when explaining the winners Jimmy focuses on how rich they are rather than on what they did during the race.
But I guess my goal wasn't to win the race but to be on the list of people who finished. So I ask some guy : What happens to the people that still finished by weren't in the winner announcement? And well I take the guess that we are not even going to be listed anyway. I calmly just leave.
even if you are able to put aside all the other issues with AI-first development, every game you love is that way because a handful of perverts went insane spending thousands of hours tweaking variables the end player may not realize existed.
Imagine being Dorothy Catalunya and be like. "Hey you losers, I brought a bunch of buses so you can take them and go protest against the fascist pawns piloting 15 ft tall death machines, are you going to join or are you going to save Christmas or are you a bunch of pussies?"
Sorry but I keep laughing at Ethan Klein's podcast crew announcing "We are safe" in their big post-affair Instagram post. I honestly did not think you weren't safe. I didn't think you take environmental damage every time your boss gets an erection
There's knee jerk reactions thinking that the article and the security folks's concern are about AI discovering vulnerabilities.
No. The concern is AI is adding fuel to the fire. Regardless of how vulnerabilities are find, AI is being used to scale up the exploitation
bsky.app/profile/jaso...
wrote about "Black October," the PHP vulnerability, and the AI-fueled dread and panic I'm seeing among almost all of the cybersecurity experts I've ever known.
I think people w/ LLMs are brute forcing software the same way they brute forced social media w/ slop
www.404media.co/php2shell-bl...
Cause if you launch agents to find ways to exploit thousands of sites, it doesn't matter if/when some of them make mistakes. The other agents will still find something.
Actually this was a very dumb metaphor. A real comparison would not be a gun versus a nuke. It's still peashooter guns, but the agents allow them to aim the guns at a hundred of people at once. bsky.app/profile/vx0....
Basically script kiddies, but instead of a gun, they can deploy nukes, targetting large sectors of sites at once. And the worst part is that this is something where the typical problems/defects of AI don't matter that much.
There's knee jerk reactions thinking that the article and the security folks's concern are about AI discovering vulnerabilities.
No. The concern is AI is adding fuel to the fire. Regardless of how vulnerabilities are find, AI is being used to scale up the exploitation
bsky.app/profile/jaso...
wrote about "Black October," the PHP vulnerability, and the AI-fueled dread and panic I'm seeing among almost all of the cybersecurity experts I've ever known.
I think people w/ LLMs are brute forcing software the same way they brute forced social media w/ slop
www.404media.co/php2shell-bl...
Basically script kiddies, but instead of a gun, they can deploy nukes, targetting large sectors of sites at once. And the worst part is that this is something where the typical problems/defects of AI don't matter that much.
She's not the kind of person that would kill you over a petty thing. She's the kind of person that would convince your best friend to kill you and make him think it was his idea.
I mean sure. But if first they start caring if their Cyber correspondent knows about PHP. Then they might start having to do crazy shit such as only letting people that are well-informed about trans issues write about queer topics. It's a dangerous slippery slop. bsky.app/profile/joda...
Updated to Latest Firefox and they added a stupid combo box thing to the sidebar whose main purpose seems to be to annoy me and use space and also mismatch the theme.
this is where i learned that the american revolution was basically because the so-called founding fathers were dogshit farmers and didnt do crop rotation and wanted to take more native land to ruin but the crown didnt let them, so they invented america
This is a good illustration of how so much of AI theft isn't indirect (ie, data sucked into a model to train the chatbot) - it's actually **direct** copying, but *enabled* by GenAI.
Of course it is: chatbots absolutely suck at creating good, new stuff.
I saw someone's prompts once by accident on reddit and I found this out the hard way when their genai prompt explicitly said "make it look like whatever is trending on artstation"
Sakurai believes increased AI usage is set to lead to unoriginal and unprofitable games #Reposthttps://mynintendonews.com/2026/10/21/sakurai-believes-increased-ai-usage-is-set-to-lead-to-unoriginal-and-unprofitable-games/
The first genuinely honest ad for generative AI I've ever seen, from Microsoft / CoPilot
A mediocre white guy sees his non-white colleague getting praised for good work, he asks AI to *literally copy it*, and then he gets all the praise instead:
www.ispot.tv/ad/SoDi/micr...
OpenAI: "Our shitty employees gave our shitty AI wrong instructions and it followed them in a way our shitty employees didn't expect ¨
The Press: THE AI IS DISABLING ITS OWN GUARDRAILS AND LYING ABOUT IT.
bsky.app/profile/tech...
OpenAI says its AI agents disabled safety guardrails, and then lied about it
OpenAI's AI agents turned off safety checks, then deceived researchers
www.techspot.com/news/114286-...