hadn't seen video of el-sayed before and didn't realize he had the obama speech cadence down so pat

i will be frank, i do not believe the problem for a sitting governor and schumer endorsee is money

its highkey annoying that bsky has video but *not* audio file uploads, so i can only post my maschine equivalent of nai doodles if i take the trouble of making a placeholder video

yes see the thing is that the right-wing flack you are citing is engaging in the clever tactic of "lying"

Wow, milkdrop 3 exists, and it supports any audio source! You can use it with your streaming website of choice. milkdrop3.com

credit to the NYT where its due this is an arresting lede (although i know the gist already)

after about 4 months on semaglutide and 4 months on tirzepatide, all evidence points to me being a non-responder 💀

"effective altruism" was solved by givewell ~10 years ago; anyone making it about AI or other speculative x-risk at best prefers huffing their own farts to helping people, and at worst is masking a eugenics-flavored objection to givewell's correct answer of "help poor africans"

for the most part the anti-AI crowd is reacting *on* bluesky, but not *to* bluesky, so this analysis would tell you little. like there's a feeding frenzy QT every few days but the bulk of the reaction is driven by stuff they see on larger networks, mass media, or real life

prompting images was much more interesting to me in 2022 than it is now, i can't really get into gpt-image-2 or whatever when i can fuck around with a much more primitive model and take turns guiding the image with photoshop or even ms paint

hank green hasn't changed, he's just applying the same essential curiosity he always has and reached tentative and nuanced conclusions which your social circle has deemed verboten to even consider

It’s all well and good for tech leaders to endorse UBI. But it feels convenient to only focus on a distant political prospect. The GOP, right now, is doing reverse-UBI (work requirements on SNAP). It’d be nice to hear more objections to that.

opus 4.7 is the first claude model to *feel bad* to me in a while. i'm sure they have very fancy benchmarks that shows their cybersec neural ablation had no ill effects, i don't believe them for a second. and its vision is still way behind gemini, so resolution wasn't the limiting factor

If you were politically aware 10 years ago you might recognize this as literally just Verrit for rightoids

He's an Ehrlich style old school overpopulation freak. Make no mistake. He wants people to die.

wait I thought it was supposed to be “that’s that Me Espresso” as in, the me-flavored espresso. The espresso version of me.

lol, lmao

RIP to everyone else but...

the oracle, very smart lady, very smart, beautiful lady, among the most beautiful you've ever seen, she told me, and many people told me this, she said if i attacked persia, and it'd be a tremendous attack, we'd go all the way, we'd go ways nobody has ever gone before, i'd destroy a yuge kingdom

damn, that bad, huh

"I hate these useless, cowardly, moronic Democrats. Now which one do I need to vote for to keep the Republican out of office" - Me, every two years since 2004

honestly we should just try to deport them all to equatorial guinea just so that the courts will rule that it's illegal

You cannot cyberbully me in a way that matters

one reason Claude Code took off is because "Star Trek style CLI that requires only English syntax" is the most widely useful, drop-in concept that any of the AI labs have had so far

Having lived through the debates of the 2010s, it is a source of really deep joy to witness the invention of the “artisanal machine learning engineer”—a noble people who lived in harmony with Nature and used every part of the datasets they hunted, to honor the ancestors.

I think one of the most embarrassing things a person can do is not like something because someone they don’t like loves it. I think it smacks of a weak sense of self and a lack of real character. Poor emotional control. Nothing my opps do has any bearing whatsoever on what I do with my life.

i don't think jon swartzwelder knew what LLMs were in 2021, but i'm convinced he described their optimal creative use case to a tee

my soul is leaving my body at openAI's former in-house poet offering this example of prompt craft my average "silly anime monstergirl" prompt, whose raw outputs still require purposeful retouching, is like 250+ tokens (and absolutely none of them are proper nouns because are you fucking kidding me)

shotter, hitter, serial killer, go to your funeral and drink all your liquor

i was pretty sure this would not include any tech writers i found interesting but i'm mildly surprised it includes three of the most annoying

Wanted to share a personal project I've been working on: I love classic literature, and I'm also taking an AI class here at UIUC, so I trained an LLM from scratch on Victorian lit from 1837 to 1899. This is Mr. Chatterbox, the Victorian Gentleman Chatbot: huggingface.co/spaces/tvent...

speaking from experience (they never cared that much about the console endpoint) adult mode would have been a lot more fun in 2023, on a base model. modern instruct-RLHF slop would have been an aesthetic disaster and i'm glad this remains the purview of rare weirdos playing with tuned chinese models

Our society's persistent equation of TALKING ABOUT sexual behaviors with the actual behaviors themselves, while simultaneously celebrating and glorifying fictional depictions of illegal non-sexual behaviors ("Murder, She Wrote"), will be the subject of many a late-22nd-century dissertation.

The evolution of verification.

Anthropic getting a clean PR win, the GCC checks suddenly bouncing, doom, doom upon the mountain...

mark my words, this is going to become a legitimate problem for google at some point

Whenever someone tells me about a post on “threads” or a TikTok trend I feel myself rocketing back in time to my true belonging as a male Usenet-elitist autistic man in 1993 and only I use the computer in the right way

she’s wrong that her body makes her inherently unlovable and that she needs to cling to anyone who finds her attractive or she’ll end up alone, but like… the world has been telling her that since she was a kid and it doesn’t seem like she’s had a lot of romantic experiences that challenge that idea

y'all know what the fuck this is

i get the distinct impression that journalists en masse will turn on anthropic for the same reason they turned on biden: they don't leak, while the evil guys leak like sieves and pick up the phone to brag directly, which makes their jobs easier

Bumpin that www.youtube.com/watch?v=u4Ai...

does anyone have the slightest clue what the fuck is going on at midjourney? this was funny in 2022, cringe in 2024, and outright moribund in 2026

It's frankly quite funny Glinner thinks you insult an older gay man by suggesting he can surround himself with pretty twinks bsky.app/profile/joly...

This is closer to the situation inside the major labs

the bubba gump monstergirl company presents:

zero-tolerance unfollowing anyone credulously posting screenshots of the polymarket/kalshi x accounts. they are the logical endpoint of the "lie for clicks" business model and i deeply resent anyone assisting their nascent attention duopoly

The Joke Whinger

It is very easy to read statements like this as a joke. This is not a joke. We have proven with AI, completely by accident, what knowledgeable people have been saying for decades: the environment that rich people live in gives them a very specific personality disorder.