Mastodon Feed: Posts

Boosted by stephen@social.calmabiding.me ("Stephen Castro-Starkey"):
chadmccullough@bsd.cafe ("Chad McCullough") wrote:

Vim, with no AI.

Vim Classic

https://vim-classic.org/

#vim #noAI

Boosted by pixxl:
stux@mstdn.social ("stux⚡️") wrote:

Tralalala

Boosted by pixxl:
markmetz@sfba.social ("Mr.Mark") wrote:

“In the meantime, some of the interviews with the billionaires are just mind numbing. Jensen Huan on Ezra Klein last week seemed to suggest that the task of automating our field had essentially been completed. Ross Douthat interviewing Dario Amodei claimed that it's well-established that AI can read a "scan" better than a radiologist. Don't forget Geoff Hinton saying in 2016 that AI would replace all radiologists in 5 years...

These people have zero idea what we do. And If they have zero idea what WE do, then that means they've got zero idea what anyone else does. Rather than being the first ones to be replaced, I contend that we are canaries in the coal mine for this technology and why it will not wreak the havoc that our billionaire overlords are hoping to scare everybody into.

They are just too far removed and self-satisfied to realize how incorrect their predictions are. Not to mention how useless their technology is in the real world when lives are at stake.”

more ->

Boosted by pixxl:
markmetz@sfba.social ("Mr.Mark") wrote:

“I'm truly just blown away by how little impact AI has had in our field, even when it comes to pure diagnostics.

My hospital employs a limited number of "models" for basic tasks like pulmonary nodule detection, and even then they suck to such a point that I stopped using them early on in my residency.

Beyond that, I have never heard of anything remotely approaching the point of analyzing CT, MR, US, PET, or catheter angiography accurately and producing a complete report.

The technology is not just in "early stages"...it's plainly nonexistent. And the "cutting edge" research that I review ain't exactly making much progress (it's usually for much more banal bullshit, like fixing reporting errors, and triaging critical findings...which is all well and good, but only modestly impactful in the setting of a true clinical workflow).”

more->

Boosted by pixxl:
markmetz@sfba.social ("Mr.Mark") wrote:

So, about AI taking all those jobs… this from a radiologist on Reddit.

“Real talk here in radiology

Just need to get this out there, since some of the interviews and opinions I have heard about the state of radiology are just unmoored from reality...

I am a senior resident in diagnostic/interventional radiology with a background in applied ML research from the pre-LLM era. I also review papers for a leading medical AI journal.”

more->

Boosted by pixxl:
atax1a@infosec.exchange ("mx alex tax1a - 2020 (6)") wrote:

shout out to the guy who's saying "hallucinations havent been a problem for a while now" while asserting that his personal experience is meaningful while ours isn't. fully cooked techbro brains

Boosted by pixxl:
mcc wrote:

Bluesky is having an odd little debate about use of the phrase "stochastic parrot" and how it appears to be the one phrase that just sends AI advocates into a rage, and I'm really torn, because while I'm in favor of making Ai advocates angry, I sincerely believe that parrots are capable of both understanding concepts and using language to intentionally communicate about their understandings of concepts. And I believe this is not true of LLMs

Boosted by pixxl:
atax1a@infosec.exchange ("mx alex tax1a - 2020 (6)") wrote:

imagine retaining respect for someone after they become a sloperator

Boosted by pixxl:
davidgerard@circumstances.run ("David Gerard") wrote:

Boosted by pixxl:
arclight@oldbytes.space wrote:

@davidgerard I often think about how - if I hadn't done the reading - I'd think that the rationalist AI sex cult was just some crazy made-up story that unhinged spittle-flecked AI haters use to scare passers-by like a bug-eyed street preacher. It's honestly hard to believe. EXCEPT WE HAVE A STACK OF RECEIPTS. A VERY TALL NOTARIZED STACK OF RECEIPTS.

This remains the stupidest timeline.

Boosted by pixxl:
jedbrown@neuromatch.social ("Jed Brown") wrote:

This afternoon, the Boulder Faculty Assembly passed a resolution (after strengthening via amendment) calling for no extension of the OpenAI contract. Vote was 39-8-1 after suspending the bylaws, as necessary to meet the chancellor's artificial decision deadline of tomorrow.

The CU Student Government meeting was this evening and public comment was fired up about their corresponding resolution, talking about trust, societal impact, and genuine commitment to learning. So energizing.

CU Student Government passed their corresponding resolution (with lots of context and calls for shared governance) unanimously on first reading, as necessary to meet the chancellor's deadline. I am so proud of their curiosity, purpose, commitment to shared governance, and parliamentary skills. ❤️

University administrators play on our division. They have claimed that there is no coherent message from faculty or students about "AI". That refusal is fringe. That OpenAI contracts are ordinary procurements not subject to shared governance.

May the record show that faculty and students speak as one. ✊️

Today was a good day for democracy, no matter what the chancellor deigns to do with this information tomorrow.

Boosted by pixxl:
nicholasgrossman.bsky.social@bsky.brid.gy ("Nicholas Grossman") wrote:

AI is now the biggest investment in all of US history as a percentage of GDP.

Datacenter construction is up, all other construction is down.

"AI Is Squeezing Out the Rest of the Stock Market," reported WSJ 2 days ago.

We basically bet the whole economy on the most optimistic visions panning out.

RE: https://bsky.app/profile/did:plc:x4qyokjtdzgl7gmqhsw4ajqj/post/3mx2oktgvbk2c

Boosted by cstanhope@social.coop ("Your weary 'net denizen"):
ajroach42@retro.social ("Andrew (making toys)") wrote:

Making choices is hard. Sometimes it's exhausting.
Sometimes it's easier to not make a choice.

I do it all the time.

But not choosing is also a choice.

By allowing an algorithm to decide what I'm going to watch, or even just to provide the options from which I made my choice, I am ceding my agency to the machine. I am choosing to trust the people who programmed that algorithm, and the choices that they made.

And, increasingly, if this isn't pointing you towards political extremism, it's pulling you back to the mean, taking you from whatever potentially unique and interesting and maybe independent media you might be consuming back to whatever had the most marketing dollars pumped into it.

Boosted by soatok@furry.engineer ("Soatok Dreamseeker"):
animesh ("baksignal") wrote:

"Programming is building a mental model. It is not mechanical code production. The shared mental model in developers' minds is the real product. A program is "living" only while the team understands this theory. When programmers leave, it becomes "dead". Without the original theory, new maintainers treat software as a black box, creating patchwork fixes that cause side effects and architectural decay."

Absolute truth.

Originally from Peter Naur's "Programming as Theory Building" (1985)

Boosted by soatok@furry.engineer ("Soatok Dreamseeker"):
wordshaper@weatherishappening.network ("Dan Sugalski") wrote:

@soatok wait, what? Someone actually thought you were talking about floor joists or roof trusses or something?

soatok@furry.engineer ("Soatok Dreamseeker") wrote:

https://soatok.blog/2026/10/02/touching-grass-isnt-enough-the-furry-fandoms-issues-are-structural/#structural

I am extremely disappointed that I needed to actually clarify this. For fuck sake.

timnitGebru@dair-community.social ("Timnit Gebru (she/her).") wrote:

"In 2021, co-founder Anna Salamon, by then CFAR’s president, acknowledged that at least three employees had effectively been compelled to undergo “debugging,” the community’s term for intervention in a person’s own psychology."

Nothing to see here.

NfNitLoop ("Cody Casterline 🏳️‍🌈") wrote:

When you're on a web site and it asks you to click that checkbox to prove that you're a human, do you concentrate on moving your mouse like a human would to click it? Do you consciously perform your humanity for the bot-detecting bot? ... just me? 😅

Boosted by kornel ("Kornel"):
scottjenson@social.coop ("Scott Jenson") wrote:

My "Future of Desktop UX (part 2)" talk at #KDE Akademy is now up and it's racking up lots of views. I'm excited people want to discuss this more.

Now I just need someone with Wayland experience to help me prototype this.
https://www.youtube.com/watch?v=V7AfAcQwLW0

dysfun@treehouse.systems ("gaytabase") wrote:

it's spiteful when a dvd or bluray puts the titles out of sequence so they don't play in order without using the menu

ireneista@irenes.space ("Irenes (many)") wrote:

we made the call originally to do things we knew were inefficient, because we had a lot of conceptual complexity in the code we were writing, a lot of novel problems in need of solving, and we chose to spend our limited ability to juggle that complexity on the essential parts of the problem rather than the "easy" stuff

but since we're solidly post-release now, we've been revisiting those things as we run into them, and making them more robust

Boosted by brennan@social.lol ("Brennan Kenneth Brown 👻 BOO!"):
moonsea@omg.sayitditto.net ("melo") wrote:

my blog can now render pages as plain markdown, not for weird gross AI reasons but because @brennan made a plain text version of his site and i was inspired to do something similar lol

if i put html in the post though that isn't converted to markdown bc i don't know how to do that on rails so just deal

just add `.md` to the end of any post url and it should work! like here https://bubblegum.girlonthemoon.xyz/articles/reverting-to-gnu-coreutils-and-sudo-on-ubuntu-26-04.md

timnitGebru@dair-community.social ("Timnit Gebru (she/her).") wrote:

"Yarvin was not an outsider to the rationalist community; he had been posting in the community’s own forums before he started his own blog, and by 2011 LessWrong’s annual survey was asking members whether they had been influenced by him, because enough of them had that the question seemed worth tabulating."

Curtis Yarvin is in there too.

ireneista@irenes.space ("Irenes (many)") wrote:

we had thought we might also want to keep the array sorted at all times, which would have been a semantics change we'd have to test carefully, but it turned out we could get the performance improvement without doing that (it's probably still mostly sorted)

Boosted by glyph ("Glyph"):
dryad.technology@bsky.brid.gy ("mcc") wrote:

Have you noticed, lately, or noticed people around you commenting, that there just seems to be no money anymore. There's no money for companies to make the tv shows you like, the video games you like, the local diner you like shut down, no one you know can find a job.

Think about that in context of

RE: https://bsky.app/profile/did:plc:756gaaeewovgn4axja34ajne/post/3mx2p4of7o22z

ireneista@irenes.space ("Irenes (many)") wrote:

we figure it's worth mentioning this because last time we talked about optimizations, someone expressed surprise that just changing a basic algorithm choice like that has such an impact.

yes - if anything this sort of thing has far, far more impact than all the fancy techniques modern compilers use. the fancy stuff is for when you've already gotten this impact and still need more.

ireneista@irenes.space ("Irenes (many)") wrote:

this is, once again, a change to the metadata entry array used by the hex transform. we now maintain a counter of how many items are in the array, instead of scanning through it to count.

ireneista@irenes.space ("Irenes (many)") wrote:

we made some tiny but substantial improvements to Evocation's build times last night, they're under a minute in most cases now

they'd been creeping up to fifteen minutes and another change we're in the process of making had doubled that to thirty, so it was time to do this work

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
uglyreykjavik.bsky.social@bsky.brid.gy ("Ugly Reykjavik") wrote:

The trees really are lovely.

#Iceland #Reykjavik #photography #streetphotography #nature #abandoned #decay #trees

Boosted by glyph ("Glyph"):
mhoye@cosocial.ca ("mhoye but seasonally spooky") wrote:

Posh. Baby. Sporty. Pumpkin. Ginger. Old. But they were, all of them, deceived, for another Spice was made.