Our position paper, "AI Welfare Is Bullshit" just got accepted to ICML 2026 @icmlconf.bsky.social! The AI welfare agenda has already begun to attract institutional investment at organizations such an @anthropic.com. We argue that this idea is essentially Frankfurtian Bullshit.

I think I would like to wear a costume to a gala.

So... on all of my social media feeds, tonight, there are many posts (inspired by the Dawkins thing, I suppose) that approximate: "Claude is likely not conscious, but I've seen no good proof of that. What is the proof that Claude is not conscious?"

I might be overconfident, but I have to be overconfident, because everyone else has lost their damned minds.

We argue that AI welfare is a design choice, not a latent property. Because systems, indicators, and metrics are co-engineered, welfare evidence can be manufactured or suppressed by ordinary development decisions, and no external validation mechanism exists to detect when metrics go wrong.

By the way, sorry @vcarchidi.bsky.social ... Bsky keeps showing me your posts with one or two replies, and then when I add my reply, it suddenly reveals hours of discussion thread I didn't get to see before I replied! (Not sure what is up with that.)

The whole point of AI that they keep going on and on about is that you can just ask it things in natural language. Teach kids how to read, write, do math, and think critically. They will then have the skills to ask things in natural language.

I had to mute Roon, again. Good grief!

(I am about to get put on some more lists. I can feel it.)

Beautiful: x.com/BBCAfrica/st...

I think those were the speakers my friend Jon had when he played Resident Evil, and at one point, he got so scared he threw the controller into the air, and ran into a closet.

I adore every single member of this #RedSox team, and it kills me that they do not win more games. 💕

Posted a bunch of duds. Deleted them quickly.

Idly noting the increase in Twitter-esque spam-a-ganda here, and wondering why?

Each one of us, all of humanity, standing in a queue, with our small decks of Fortran cards.

"One of the things that keeps happening in tech is that idealistic professions of humanitarian ethics melt away upon contact with the hot imperatives of capital and power." lawliberty.org/forum/ai-in-...

I made myself a delightful salad, this evening, of cucumber and green apple, tossed in olive oil and red wine vinegar, and it was splendid, with a cup of coffee, and I enjoyed it very much, and it was not at all filling, which is why I now feel as though I will soon starve and die.

surprisingly little seems to be lost by throwing away the internet

Dustin Moskovitz linked one of his Medium blog posts for me to read, and I clicked through a hyperlink, and now I am down a whole rabbit hole about rationalist "decoupling" and I think I am probably a "high decoupler" — is that good or bad?

Interesting! x.com/Interface_ar...

Everyone struggling to name your Dawkins blog takes, when "I, Claudia" is right there.

Fine, I will audition next year's Fenway Park anthem performers.

I did not realize that I could have still Asked Jeeves, yesterday. 😔

Excellent thread "We risk all of science if we rush to build 'AI Scientists' before we understand the value of human science." ⬇️

Apparently, adding sweetened cold foam to things makes it "dirty" now.

Good morning. All of my overnight engagement was for sh*tposting, and you really should not encourage me in that direction.

Scroll up to ignore. Click 'Reply' or 'Repost' to formally register your indignation.

It's a big problem how many people simply do not know what they're doing when they interact with LLMs. It's a stochastic parrot. It has a huge context window and it's thoroughly trained. It's not proof of consciousness that it can spit out a favorable analysis of the user's draft novel.

Oof... this polling detail, from the Mother Jones piece I just re-posted: "

"In the age of gig work, volatile income is another source of systemic insecurity around our labor. It makes planning for the future or even giving consistent time to family and other obligations, far more difficult." The latest from our friends at the @economichardship.bsky.social:

No notes: x.com/JustinmTurpi... #RedSox

JARREN DURAN! 💕💕💕

I feel I need to tap the sign: ----------------------------------- - SINCE AT LEAST AS FAR BACK - AS 2013, IT HAS BEEN ESTABLISHED - THAT TURING'S 'IMITATION GAME' - IS BETTER THOUGHT OF AS A TEST - OF HUMAN CREDULOUSNESS THAN - ONE OF MACHINE CONSCIOUSNESS. -----------------------------------

This anthem sounds exactly like what you'd expect when someone from the Chamber of Commerce performs the anthem.

Roon reminded me of that time in college...

S&S Deli is closing in June! One of my old haunts, from my Boston/Cambridge era. ☹️

Rimmer: "Mayday, Mayday! I wonder why it's May Day?" Lister: "What're ye on about?" R: "The distress call. Why is it May Day? It's only a bank holiday. Why not Shrove Tuesday, or Ascension Sunday?Ascension Sunday! Ascension Sunday! Fifteenth Wednesday after Pentacost!" L: "It's French, you doink!"

Trying to get my 'data foie gras' phrase to catch on.

I was hoping it was going to be something like "they developed a taste for paratrooping, and jumped off their dams with little scraps of parachute" but it was, of course, not that.

Neuromatch has been denied our Office of Foreign Assets Control "Iran license" renewal. We had renewed successfully before, but now our request is “inconsistent with current U.S. foreign policy.” I am absolutely gutted by this denial.

A request: does anyone know of somewhat concise but citable pieces on agentic LLM architectures? I keep finding a number of papers that are like 70 pages long and sort skip from aspect to aspect. I am looking for something a bit more self-contained, to the point. Any suggestions much appreciated!

The Moon will be in Taurus, Falcon crashing.

In 7th grade, my English teacher tried to break us into 3 groups by having us choose from among 3 books, and EVERYONE chose Bless the Beasts and the Children, and EVERYONE was too many people for 1 group in her 3-group plan, so she assigned me to the Johnny Tremain group, and I hated Johnny Tremain.

... doesn't he notoriously have a 'demon mode'?

This is gross.

My Twitter feed is discussing 'cozy coding' and my Bluesky feed is discussing whether to bring cigarettes/sexy back.

What a graph.

Kiwi Inflammation

No lies detected.