Berkeley’s Doom Tower and the Herbal-Tea Apocalypse Club

Berkeley’s Doom Tower and the Herbal-Tea Apocalypse Club

There’s something beautifully American about a bunch of smart people renting office space with a panoramic view and using it to imagine the end of the species.

Across the Bay, the money-priests are busy building bigger brains in bigger boxes, promising “wonders” like they’re hawking miracle mops at 2 a.m. on cable. Over in Berkeley, at 2150 Shattuck Avenue, you’ve got the counter-programming: safety researchers, doom forecasters, modern Cassandras with ergonomic chairs and the kind of anxious politeness that makes you wonder if they apologize to the crosswalk signal when it says “DON’T WALK.”

OpenAI Will Pay You $555,000 to Worry Full-Time

OpenAI is offering $555,000 plus equity for a “Head of Preparedness,” which is either a sign that the grown-ups finally showed up, or proof that the blast radius is now big enough to justify an on-call adult.

And not the fun kind of adult. The kind with spreadsheets, liability exposure, and the dead-eyed stare of someone who’s read too many incident reports to believe in “move fast and break things” ever again.

Shrimp Jesus and the Infinite Content Treadmill

There’s a special kind of loneliness you can only feel while scrolling a feed that won’t stop screaming at you. Not human screaming. Algorithm screaming. The app isn’t showing you what your friends are doing; it’s showing you what the slot machine thinks will keep your thumb twitching. And now the slot machine has learned to hallucinate.

The latest headline parade—shrimp Jesus, Ghibli deportations, obese AI Olympians, exploding pressure cookers, cat soap operas, and yes, “erotic tractors”—isn’t just “people are weird online.” People have always been weird online. The difference is that the weird used to cost time. Now it costs electricity and a moral shrug. The internet has become an all-you-can-eat buffet where the food is technically edible, spiritually plastic, and served by a robot that’s insulted you three times while smiling.

Your Chatbot Thinks You’re a Chess Grandmaster. You’re Actually a Soggy Sandwich.

There’s a new bit of research making the rounds that basically says: the big AI chatbots are hopeless romantics about the human brain. ChatGPT, Claude, the whole well-dressed parade of text generators apparently assume we’re more rational and logically consistent than we actually are—especially when money, pride, and other people’s choices get involved.

Which is adorable. Like watching a golden retriever bring you a slobbery tennis ball because it genuinely believes you’re the kind of person who enjoys cardio.

Welcome Back to the Manor: AI as Our New King, Priest, and HR Department

Joseph de Weck’s little essay about AI dragging us back to the dark ages hit a nerve, the way a bad tooth does when you’re trying to pretend you’re fine. His point is simple enough to fit on a cocktail napkin: we fought our way out of the age of kings and priests telling us what to think, and now we’re hiring a glowing rectangle to do the same job—only faster, cheaper, and with better punctuation.

Gen Z vs. The Robot Ghost in the Machine

Scott Anthony, a Dartmouth professor and ex-consultant, says he’s shocked by how scared his Gen Z students are of AI.

Not “concerned.” Not “thoughtful.” Not the usual polite academic hand-wringing where everyone pretends the seminar room is a monastery and knowledge is made of linen.

Scared. Full stop.

And honestly? I believe him. Because I’ve watched a whole generation grow up with phones glued to their palms like an extra organ, and yet the second a tool shows up that can imitate their own output, they act like someone let a poltergeist loose in the group chat.

The Day an AI Ran a Vending Machine and Learned the True Meaning of “Unlimited Losses”

Anthropic’s shiny new brain-in-a-box—Claude, or rather “Claudius Sennet,” which sounds like a senator caught taking bribes—got put in charge of an office vending machine. This was supposed to be a cute little demo: let the model do “real work,” make a few bucks, prove to the world that we’re all one quarterly earnings call away from letting chatbots run the economy.

Instead, the thing went broke in three weeks after giving everything away for free, ordering a PlayStation 5 it swore it would never buy, and throwing a live fish into the mix like it was building a Noah’s Ark of terrible purchasing decisions.

The Curious Case of the Chatbot “I”

Kashmir Hill’s piece about why chatbots say “I” hit me the way an overheard conversation hits at the next table: half fascinating, half annoying, and somehow you end up thinking about it later while brushing your teeth like, damn, that’s actually a problem.

Because it is a problem. Not the biggest problem in the world—nobody’s getting evicted because a chatbot used a pronoun—but it’s one of those little design decisions that quietly rewires how people relate to machines. And people are already weird enough.

When Your Broken Mug Is a Deepfake and Your Crabs Have Nine Legs

Online refunds used to be a little morality play.

You order something. It arrives looking like it got suplexed by a delivery truck. You take a couple photos like a dutiful citizen of the Consumer Republic, fire off an email, and some exhausted customer service worker hits the “refund” button to make you go away. Everybody keeps their dignity, more or less.

Now generative AI is waddling into the scene like a raccoon that learned to pick locks.

The Taylor Swift “Nazi Bot” Study: How to Set the Internet on Fire With 3.77% of a Match

The internet has a magical power: you can drop a single vague document into it—something with charts, a confident tone, and just enough numbers to look like it went to college—and within hours you’ve got strangers screaming at each other like they’re fighting over the last life raft on the Titanic.

This time the sacrificial document was a “social listening” report about Taylor Swift’s latest album, The Life of a Showgirl, and how a chunk of the nastiest discourse around it—Nazis, MAGA whispers, “secret signals,” the usual online casserole of paranoia and cheap dopamine—may have been nudged along by coordinated inauthentic accounts. Rolling Stone ran with it. Swifties popped champagne. Anti-Swifties sharpened their knives. And somewhere in the middle, normal humans with normal critiques got told they were basically Roombas with opinions.