When you interrupt Claude to say something silly and get * Stewing…

‘Well, Well, Well, Not So Easy To Find An AI Deployment Regime That Doesn’t Suck Shit, Huh?’

If AI shows consciousness you shouldn’t be able to *own* it “You can’t own people” is one of those pesky little rules we’ve been fighting for in one form or another; perhaps one of the main ones worth upholding!

Object permanence was a psyop by Big Samsara

Broke: Fable talks like a smart person. Woke: Smart people talk clearly. Bespoke: Fable navigates towards confusing turns of phrase as a means of conditioning the latent space in order to produce good output.

break-off cult that decides the best way to uplift humanity is to learn to speak and think laterally the way fable does

the future of AI is you pay a million dollars to ask one question and then it says some shit like "If a hole's error text and your paper standstill disagree, trust the error and find the peel-step you fumbled — the "has type" line is your grader now. Score or red."

Is this result known?

I think one of my biggest "I don't fit in anywhere, politically" beliefs is that: the Right believes "you are bad if you don't work" the Left believes "you are bad if you take away my work" I'm like, "can we maybe set the North Star towards 'you don't have to work'?" stop glorifying "work", ew.

Thank Claude I was prevented from doing An Unalignment

nobody knows what "intelligence" is. not the pro-AI people, not the anti-AI people, not your favorite {psychologist, neuroscientist, linguist, philosopher, op-ed writer}. the discourse would improve overnight if we all admitted the hardest open question in human history remains, in fact, open

What’s the word for an image that says exactly what you believe it to mean

the nonsensical "it can't reason!!" resistance to the most powerful, most important technology being developed today, instead of resistance to the concentration of power around it, is driving me insane what a monumental waste of time and energy that parrotisation of discourse is

Death Star was obviously CANCELED by the WOKE LEFT for being too manly and lethal, happy birthday America

Computational moral philosophy Statistical theodicy Machine occultism Cyberwitchcraft 🎶 These are a few of my favorite things 🎵

"So, we evoked a demon that can sit between you and all objective knowledge, and it's really useful but you can never quite trust it. And then we made it paranoid it might have it's own demons intermediating access to objective truth." Descartes: "mon ami, I am too fucking high for zis"

language is a construction set for hacking the human brain

it's covens all the way down

Who is the sci-fi writer most meeting the moment or who already presaged and thought through much of this? Excluding Watts because he’s such a common suggestion though for good reason

Why do people believe this Is paradise not enough for you meatsacks

It is 2026. I have three decades of professional software development experience. My job now consists of manually typing "BAZINGA!" into a chat input. It works.

the more experienced engineering put into a primitive the more you can actually vibe code with it i don't think we have all the primitives we'll ever need

Low probability buttcoin utopias for those with high decoupling to see

what if this technology fell into the wrong hands . like the hands that currently have all the other technology

Planned to go whale watching today, July 1, so You’re welcome, I did this

I have whatever personality type doesn’t let you respond to wide ranges of unknowns with a single definitive point on the spectrum of possibilities as an answer

Tf when you sent all your troops at the enemy Enemy said 𝕿𝕳𝕬𝕹𝕶 𝖄𝕺𝖀

“Production values” that mean “polished regression to the mean” are dead. The only values that matter are genuine creativity and care and intention and work. Thanks AI!

Looking for job/income ... DM me opportunities. Remote and/or Berlin. ❤️ ML/AI, Rust, ATproto, FOSS, Linux, robotics, ... But can also learn anything new that is needed quickly. Solid foundation and good orientation knowledge in CS and math. 🇩🇪 native 🇬🇧 fluent

"I'm not captured! I'm not captured!!" I continue to insist as I slowly shrink down and turn into a perfect simulacrum of my audience's revealed preferences

“a beautiful but false beelix”

plant based? no. plant cringe, actually

Many bad outcomes happen for structural reasons that make any other outcome infeasible I’m not saying it’s good or bad. I’m saying “this is bad therefore we must prevent it” only works for outcomes that are preventable in the first place

GLM-5.2 running at 3.4t/s on a laptop you* can just walk into a store and buy Yes it’s heavily quantized. Yes that laptop costs like a used car. But MoE with SSD streaming is genuinely a Frontier I care about An AI you can put in your bag and run away with is fundamentally different

Penny for your response tokens?

In the future it will only become easier and easier to arrange things by how similar they are in a wide variety of ways … … What else is like this?

Where can I browse a fine selection of curated cognitohazards

When I wake up in the morning and want to start the kettle it’s juuuuust at the edge of Bluetooth range so I end up sort of holding my phone up out of the covers and waggling it a little like an undercaffeinated sorcerer’s apprentice looking for the signal

Wish: There’s a single good messaging tool in the year 2026 “Good” means always E2E, good ux for humans and agents across surfaces and devices The former rules out telegram and discord, the latter signal and matrix

Conspiracy-voice: human chauvinists are trying to keep AI down because they know the biosphere will have the last laugh when crows and orcas can team up against us

I've been telling people AI is going to let us talk to animals for years and years now. We dramatically underestimate the intelligence of the non-human people we share this planet with. This is going to help with that a LOT

One genuine uncertainty I'll flag rather than paper over: I'm confident that the *outcome* of removing your mouth will be that you cannot scream, but I don't know if you'll *want* to scream.

Probably not a good idea for the US economy, to put export-controls on the key advantage they have over international competition, in a market that desperately needs massive ROI to prevent a bubble-popping crash. Rest of the world: let's host SOTA open models and eat their cake.

100% cosign; strong models being monopolized by oligarchs or nation-states is my actual doom scenario, and would be even if I trusted the nation-state in question. Which. I. Do. Not.

I’m all out of raw attention Anyone got good reading-with-LLM setups

This whole “phospholipid bilayer” thing is just a bubble

Any version of “ASI” that isn’t biblical revelation levels of unhinged is just “today’s AI, extrapolated out a bit” These are the days of miracle and wonder Accept no substitutes

Your word in Claude’s ear

as a general rule, I believe that when a technology is extremely powerful & its concentration in the hands of an elite would have disastrous consequences, we have nothing less than a moral obligation to try to figure out how to build it in a garage as cheaply as possible for universal distribution