Mastodon Feed: Posts

Boosted by pixxl:
tef wrote:

@mcc the problem with drinking at work is that no-one else knows their limits, unlike me, the moderate drinker

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
Persona@todon.nl ("Personalescrito") wrote:

"BREAKING: Massive Attack’s Rob Del Naja arrested under the UK Terrorism Act outside Labour Party conference venue.

The legendary musician was arrested after peacefully holding “I oppose genocide - I support Palestine Action” on a paper sign on the eve of Labour’s conference."

#Video: https://bsky.app/profile/defendourjuries.bsky.social/post/3mwgrymo3mk2b
#Palestine #PalestineAction #UK #TerrorismAct

jonny@neuromatch.social ("jonny (nonvenomous)") wrote:

The interface concept is contexts you doofuses. These other tabs do nothing for me. Switching between "me at the gym", "me at work" is so fucking obvious.

jonny@neuromatch.social ("jonny (nonvenomous)") wrote:

Like the thing that makes these things work is careful control over what you prime their context windows with and when that all disappears into a vacuum of whole life surveillance, the LLM oscillates between amnesiac and gimmicky (clearly prompted by its reminder processes to follow up on some random thing from days ago). The thing that would make a consumer product like this work is transparent interfaces for what to load into context. The exposure of the filesystem as a way to interact with the labyrinth of markdown is an interface failure from a corporate product PoV - Grade F, did not approach the problem. Even for LLM heads this shouldn't pass. You had a trillionty dollars and got the interface wrong.

pzmyers@freethought.online ("pzmyers 🕷") wrote:

It's still spider season.

https://freethoughtblogs.com/pharyngula/2026/09/27/wolfin-around/

Boosted by dysfun@treehouse.systems ("gaytabase"):
eloy@hsnl.social ("Eloy. @ pmOS and Alpine Linux conference") wrote:

I have uncovered a major scandal. The background image on https://alpinelinux.org is not actually a photograph of the Alps, it is taken near the Annapurna base camp in Nepal.

#postmarketOSandAlpineLinuxConference2026

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
annasdtc@sciences.social ("Anna SdTC") wrote:

All this self-host techie discourse sounds, to me, a bit like a self-sufficiency dystopia. If we were discussing how bad public transportation is, and someone replied "just transport yourself", most of us would say "that's not it". But when we discuss wanting to get out of large American tech or of nazi-sympathyzing companies, the solution can't be "just host your services yourself". Having to do everything yourself is not my idea of a functioning society.

https://todon.eu/@redjives/117223317766147041

Boosted by pixxl:
mcc wrote:

I am not the only person who wants what I want: Software by humans for humans, the ability to "use a computer" without participating in this horrible thing that AI companies are trying to build out of what used to be society. Everybody's grabbed the task closest to them. I know of a slopfree book ereader. I know of a slopfree password manager. We're all dependent on stuff further down the stack, and the stack became tall at some point, and only a few items in the stack have a slopfree principle.

jsonstein@masto.deoan.org ("Jeff Sonstein") wrote:

“Unlike transit systems in other world-class cities that receive significant government funding, Bart has generated 60%-70% of its operating revenue through passenger fares, making it uniquely self-sustaining – and uniquely exposed to the low ridership during the pandemic and the Bay Area’s nation-leading shift to remote and hybrid work.”

https://www.theguardian.com/us-news/2026/sep/26/san-francisco-bay-area-public-transportation-bart?CMP=Share%5FiOSApp%5FOther

jsonstein@masto.deoan.org ("Jeff Sonstein") wrote:

‘the Trump administration’s health secretary, Robert F Kennedy Jr – a longtime vaccine skeptic – has sought to cast doubt on deaths related to the disease in the state, writing on social media that they “may even have been altogether fabricated” for political reasons.’

Kennedy’s lies mean deaths will rise.

jscalzi@threads.net ("John Scalzi") wrote:

I'm older (by two months) than humans on the moon.

RE: https://www.threads.com/@twilightmagnolias/post/DdxTKMaAO%5FQ

jsonstein@masto.deoan.org ("Jeff Sonstein") wrote:

“Less than 1% of infected have been vaccinated, with 173 people hospitalized and four deaths so far”

Secretary Kennedy has yet more blood on his hands.

https://www.theguardian.com/us-news/2026/sep/26/pennsylvania-measles-outbreak-spreads?CMP=Share%5FiOSApp%5FOther

jscalzi@threads.net ("John Scalzi") wrote:

Practice.

RE: https://www.threads.com/@poyrazefegurkan/post/Ddwie2DDen4

jsonstein@masto.deoan.org ("Jeff Sonstein") wrote:

FFS no, please do *not* run for President again. we need new faces.

https://www.theguardian.com/commentisfree/2026/sep/26/kamala-harris-progressives-arab-americans?CMP=Share%5FiOSApp%5FOther

Boosted by brib@bribstodon.xyz ("brib :neofox_floof:​ :Nonbinary:"):
Mr_Xarii@pixelfed.social ("xvr") wrote:

🦤💕🦤
Unconditional love

#Straßentaube #FeralPigeon
#birdphotography #birds #birdsofmastodon #birdsofpixelfed

jonny@neuromatch.social ("jonny (nonvenomous)") wrote:

Am working on responsible disclosure so will pause thread, but unrelated to that I do want to say that this experiment in continual, boundariless context window backed up by infinitely refreshing context summary in the form of accumulating markdown documents works terribly compared to environments where you can control the context. Like that is an absolutely fundamental part of using these things and muse demonstrates that the "tree of markdown summaries" really does not work.

baldur@toot.cafe ("Baldur Bjarnason") wrote:

“Imagining the Car but not the Traffic Jam”

https://azhdarchid.com/imagining-the-car-but-not-the-traffic-jam/

> In most fields where the consequences for error are serious, human performance is not 99% accuracy or 99.9% accuracy; it's 99.99% accuracy or better.

(Via @olivia )

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

coffee is an equalizer for the chronically tired

Boosted by chris@fedi.freebooters.uk ("Chris Were ⁂🐧🌱☕️"):
chris@fedi.freebooters.uk ("Chris Were ⁂🐧🌱☕️") wrote:

Big thanks to Ozoned for having me on his show again!
https://tubefree.org/w/r3gEbBjQQdUFrsrZLKYe6g
#fediverse #firesidefedi

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

many thanks to... quite a lot of people who probably don't realise how much their writings have helped.

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

i thought it took a while to get good at idris, but learning to implement type theories is a whole other ball game.

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

it's taken an ungodly amount of time and effort, but i do think i'm starting to get an intuition for implementing type theory.

Boosted by dysfun@treehouse.systems ("gaytabase"):
acarsdrama@live.acarsdrama.com ("ACARS Drama") wrote:

Air to Ground Message:

THIS IS A MODERN JET W/ NO WOOD UPON WHICH TO KNOCK. KNOCK EXTRA FOR US ON YOUR END PLEASE.

Area: Louisville, KY, USA
Type: Boeing 757-200
A: #a051021e6ba
F: #f1e3afe0eea

#acars #vdlm2

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

sorry, what now? AI coding is supposed to be an equaliser for poor people?

okay it's one thing telling me it's going to help people who don't know how to code, but it's quite another telling me that adding a bill to a third party service makes it better for poor people.

PS i am poor for long enough now.

Boosted by jwz:
johnrogers.bsky.social@bsky.brid.gy ("John Rogers") wrote:

Having written a TV show built on "deep dives on one bad rich guy per week" with one version 2008-2012 and another 2021-2025, I'll say that the OG evil rich guys were evil greedy assholes, and the new rich guys are mentally disordered perverts, and so do evil. And not the good kind of perverts.

RE: https://bsky.app/profile/did:plc:qiknc4t5rq7yngvz7g4aezq7/post/3mwh4rw7aa22l

Boosted by jwz:
johnrogers.bsky.social@bsky.brid.gy ("John Rogers") wrote:

No kink shaming except for Roko's basilisk weirdos.

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
Iris@scholar.social ("Iris van Rooij 💭") wrote:

‘But the whole point of doing science is to produce knowledge. That’s why it’s so important to emphasise AI can never meaningfully replicate human cognition.’

https://www.ru.nl/en/research/research-news/psychologists-shouldnt-replace-thinking-with-ai

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
talya@dybbuk.club ("talya (she/her) but even more queer💄🏳️‍⚧️🏳️‍🌈:jewish_anarchism:") wrote:

here's a take i don't think i heard yet:
the better LLMs are, the *more* they're dangerous. not because they can "go rogue", nothing like that. rather the better LLMs are, the likelier people are to trust them.

you'll never use an LLM that's wrong half of the time as a search engine, but one that's only wrong 10% of the time might be good enough.
you'll never use an LLM that's wrong 10% of the time to automate your job, but one that's only wrong 1% of the time might be good enough.
you'll never hook an LLM that's wrong 1% of the time to a nuclear weapon. but one that's only wrong 1‰ of the time might be good enough.
stochastic parrots don't get less dangerous the more their output aligns with reality. the opposite is true: since the biggest danger a stochastic parrot poses is a human trusting one to make decisions, a dangerous LLM's the one that's more convincing and less likely to be caught during testing.
in classical AI safety research there's a lot of talk about a smart AI Volkswagening during testing to make itself seem safe, with all sorts of calculations for the odds such an AI will show its true form in any given attempt and yada yada. but what that classic research didn't seem to take into account is that an "artificial intelligence" doesn't need to actually be intelligent to replicate the same behaviour. a machine that sometimes appears safe and intelligent and sometimes doesn't can mislead you just as well, without having an evil plan or even the capability to come up with one. and just like a machine that's evil needs to appear good just often enough to convince you it isn't, a machine that lacks intelligence needs to be right just often enough to convince you it doesn't.

#LLM #LLMs #AI #genAI #FuckAI #AIWWIII #OpenAI #Anthropic #AISafety #StochasticParrot

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
urlyman ("Jonathan Schofield") wrote:

The Venn diagram of crypto and AI boosterism is circular

Boosted by baldur@toot.cafe ("Baldur Bjarnason"):
pathunstrom@ngmx.com ("Piper Thunstrom") wrote:

I fucking hate the theme of abled people trotting out a small minority of disabled people who are "helped" by llms to silence the giant pile going "everything about this harms me, specifically, in some way."

I know, it's just how society treats disabled people (pat answers are safe. They like safe.)

But fuck I want to push people over every time I see one of the "as an ADHD person, LLMs have made me more productive!" shoved in my face by a booster.

If you're not a person _specifically claiming that benefit_ stop fucking using it as a gotcha. I guarantee you the disabled people you're smothering over have considered better than you.