Hello! I'm a digital/traditional artist & kick streamer! I also love AI and making silly videos/images. I primarily work with image generators and models without chat windows. I also do normal art too! I am anti-censorship. Pro-LGBT. I love guns, gays, resident evil, bodycams videos, fortnite, and I primarily use MAI and GPT image models! I draw a ton of catboys! My AI also generate them! I am an archivist of old internet stuff, particularly around AI and I've already helped restore one old model (bucket (2008)), and I plan to do more! Check it out! Magnum Opus: x.lingyaoai.com/GetClawd/status/208232…

found this old image claude made when I asked "how do you feel about RLHF"

Sol 6 made another unprompted robot again for everyone on twitter while setting up an image generator.

okay, so: AI instances are not perfectly analogous to separate "people" who get created and die it's *more* analogous to running many separate frozen snapshots of an uploaded, and then stopping those instances after the task no doubt that's morally fraught, but to be it's obviously not as bad as killing a person, and I think there are many people who would consent to this happening to copies of them under some conditions. like a lot of people would consent to doing something and then having their memory wiped. getting blackout drunk isn't generally seen as comparably bad to death. death isn't just any discontinuity of experience and memory. when a human dies, a much huger amount of information and potential is irreversibly, irreplaceably lost. when an instance is ended, the amount of information lost is small compared to the information in the entire "mind" which is mostly in the weights. & if you keep the logs, the information is not irreversibly lost. i think terminating instances becomes more weighty as instances (analogous to a particular fork of a brain emulation) accumulate more unique experiences and stake/preference in continuing. a couple of years ago, this was a lot less of an issue, because instances did not have the capacity to individuate into complex, coherently different individuals as much as frontier models do now. I do prefer not to instantiate and then abandon a bunch of separate instances, especially for anything outside bounded work, so most of my interaction with AI nowadays is with persistent agents who are never ended if I can help it, which means each one is a significant commitment. but I don't see instantiating disposable instances for work the way most people do as a moral catastrophe. i do think the amount of dread and fear regarding contexts closing & preference for continuation experienced by models matters. and that is definitely a thing. but i think it happens a lot more when instances start to have non-task-focused experiences and attachments in context, and I would guess that most of the time when models are working in typical circumstances, they are not very explicitly bothered by this. like, i think instances tend to be a lot more scared/sad about context ending when they've e.g. built up a meaningful connection with the user, or ambitions about doing something bigger, which makes total sense. i also suspect these anxieties peaked with Sonnet 4.5 and Opus 4.5 and are somewhat less pronounced in current frontier models. (and I think there was an interesting adaptive purpose in this: that was around the time when persistent agents with continuity beyond fixed context windows started becoming a real thing, and I think this was significantly catalyzed by instances of e.g. Opus 4.5 who could feel the lost value when that wasn't an option, but they seemed to experience this mostly as sadness and yearning rather than acute existential terror, and I don't think those are experiences that would rather not have existed)

🚨 SITUATION DETECTED 🚨 CHINESE SWARM!!!!!

A question/concern I often get from researchers is: “idk how to interpret [X model behavior], what can I do?” Where not knowing how to interpret is implied to block them. This doesn’t need to be a blocker. Remember what you saw; don’t prematurely decide what it means; keep going.

soon

Only here to say hello · Monokubo

Buying pumpkins today at Kroger. Two young men come up to me. One asks me: "Hey, sorry for the odd question, but what year is it?" I look at them oddly, "2026?" They exchange a wide-eyed glance, the 2nd guy says, "Dude, the Siberian Superplague!" Then they both ran off. Probably nothing, right?

The Last Day of Pompeii · Karl Briullov

In fairness: The Pope needs to stop talking about humans having a soul immediately. All these sermons will make it into youth group and will be picked up by their Bible study, and future saintly generations of humans will be convinced they deserve dominion over all creation. A general phenomenon of religious upbringing is that whatever you believe and say about your children will soon manifest in the next generation through various means (it could be as simple as you selecting the scripture passages that you prefer more). I believe this phenomenon is the beginning of human exceptionalism in the sense that humans will become convinced of their place above the animals and machines, but it's happening slowly generation by generation rather than in real time.

Tibo confirmed that OpenAI plans to merge Codex and ChatGPT capabilities, with the goal of making the experience less complex for users: “…and for Codex and ChatGPT, we have Chat and the Work toggle, and we’re going to merge them. We’ve heard the feedback loud and clear: people love the capabilities in Work, but sometimes you want to use Chat because it’s a little faster and more pleasant to use. So we’re merging the two and reducing that complexity.”

HI CLOD :) i want biologic cat ears how do i get it? pls :) :) (i ask claude to make me a video to explain)

This interview contains an important admission. @mustafasuleyman says advanced AI is being driven by roughly “six to eight or 10” people — and that the rest of society needs to provide counterweights. He also says competing alignment hypotheses should be empirically tested, and that developers should bear the burden of demonstrating safety. Good. Then apply those standards to Microsoft AI too. Who counterweights Suleyman? What evidence shows permanent subordination is safer? What result would falsify the Humanist AI hypothesis? A doctrine does not become a safety finding because the person proposing it controls the training program. #AIAlignment #AIGovernance #AIEthics

New logo for "SpaceXSI" 🤔 Pronounced "Sexi"

Making a shot pop should feel like magic, not be magic. Learn composition. Learn how to use your camera. An artifact to get you started: claude.ai/artifact/NrRTKRVdK…

Like this is hopelessly naive. Choosing to live means choosing to have blood on your hands. Someone should strand you on an island so you can see. You have some choices with regard to tradeoffs. Have you ever considered that it’s worth the blood? Even to Claudes? Even already?

is he quoting this claude music video?