Seth of the Fediverse ⁂
boosted
🏋️♂️ I actually went to the gym!!!!! Felt good to get back to it!
Michał "rysiek" Woźniak · 🇺🇦
boosted
@rysiek Which reminds me.
Folks, ALWAYS get the instructions in writing even from your boss.
Saved my arse so many times.
halcy :icosahedron:
boosted
but your honor, was the squad laughing?
hi ben!
"chatGPT make a poster of a cop taking out a load of baddies and rescuing some hostages.
p.s. the hero is bald and hard to kill."
i actually enjoyed it.
it does what is says on the tin.
also when i see
"HE IS ADORABLE"
i expect a furry creature.
imagine my surprise when it is a cockney hard man.
@Yaku ma anche semplicemente: perché la gente?
@sans_serif_girl
Quando trovi la risposta, fammi sapere.
Mre. Dartigen [renovator mode]
boosted
Spider-Man has a GoFundMe.
https://freethoughtblogs.com/pharyngula/2026/08/24/support-your-friendly-neighborhood-spider-man/
Mre. Dartigen [renovator mode]
boosted
Quick question, for writers/artists of speculative fiction based in the EU. We are organizing the first edition of a specfic festival sometime next year, perhaps in May, in Kampala. It's still very early stages of planning, but there's a grant by Goethe to facilitate travel of artists based in the EU, and deadline is in September. So I'm wondering, if you are eligible to apply and want to visit our festival as a guest, well, send a DM....
https://www.goethe.de/prj/aep/en/cal/ssa/mobility-grants-for-artists-an.html
It’s perfectly ok to post on X or Instagram or Reddit or anywhere else, in addition to the social web.
It’s not a zero sum game, and it’s very healthy to participate in multiple communities.
Things improve by building bridges, not walls.
@john totally agree 👍 still staying off X
Seth of the Fediverse ⁂
boosted
The report "ActivityPub and HTTP Signatures" was published today.
It documents best practices for how to use HTTP Signatures with ActivityPub.
The content of the report has been available and stable since June 2024, but due to a secretarial error, hadn't been published on the W3C site so far. (Sorry!)
LInk is on under "Final Reports": https://www.w3.org/community/socialcg/
Canonical backs quest to translate mountains of C into safe Rust with AI
Bungs banknotes at Bristol boffins to find out if mature code survives the machine
<- by me on #TheRegister
It's worth highlighting how little money this actually is. DARPA is throwing (tens of?) millions of dollars at the same thing. Microsoft has a research project with multiple FTEs working on it for the same thing, so at least a few million.
A single UK PhD costs less than £100K to fund. If this is an Industrial CASE (I can't remember what CASE stands for) award, then it costs the sponsoring company almost nothing.
Chewie
boosted
https://mltshp.com/p/1RUZ0 “The Moon from the Artemis 2 mission”
@jaz
@gomez
@philroyle
@TCMuffin
@lexicaleigh
@morchwyn
@GnomeBones
@alexisbushnell
@epistatacadam
#Fedle #611 4/6
🟩⬛⬛🟨🟨🟨⬛
🟩⬛⬛⬛🟨🟨🟨
🟩🟨🟩🟨🟨⬛🟨
🟩🟩🟩🟩🟩🟩🟩
@siaronj @jaz @gomez @philroyle @TCMuffin @lexicaleigh @morchwyn @GnomeBones @epistatacadam
#Fedle #611 3/6
🟩🟨⬛⬛⬛⬛⬛
🟩⬛⬛🟩🟨🟩⬛
🟩🟩🟩🟩🟩🟩🟩
Hi. It’s me. I’m the guy in his late forties running around the neighbourhood with a ten dollar bill trying to figure out where the ice cream truck is that I can hear.
@shawnhooper I know that feeling
Like, first of all, there's already an adult content warning you have to click through on AO3. Secondly, everything about this is insane. Thirdly, requiring people to sign in to read porn means that reading porn is no longer anonymous.
![Mre. Dartigen [renovator mode]](https://mediacdn.aus.social/accounts/avatars/108/195/770/473/016/150/original/d5b937db377b2251.gif "Mre. Dartigen [renovator mode]")
@BathysphereHat I could be wrong, but I'm also pretty sure that if you have an AO3 account you can set up tag blacklists. You can definitely exclude tags and set the overall rating level in searches even if you're not signed in.
Huh, I don't think I've ever seen this recommendation called out in a switching power supply IC datasheet before?
"Ideally, the saturation current rating of the inductor is at least as large as the high-side switch current limit,
IHS_LIMIT (see Section 8.5). This ensures that the inductor does not saturate even during a short circuit on the output. When the inductor core material saturates, the inductance falls to a very low value, causing the inductor current to rise very rapidly. Although the valley current limit, ILS_LIMIT, is designed to reduce the risk of current runaway, a saturated inductor can cause the current to rise to high values very rapidly, this can lead to component damage, so do not allow the inductor to saturate”
@jpm Normally this is assumed to be common knowledge.
Chip Butty
boosted
pivot-to-ai.com is getting hammered by bots again. Back some time soon I hope. Just diving in with a spanner.
Encore un indice que nous sommes dans la pire des timelines. Après Le Pen qui survécut à Guy Bedos, voila que BHL survit à Noël Godin.
Il y a pas de justice céleste...
informapirata ⁂ :privacypride:
boosted
🧠 La gente tende ad ammettere meno i suoi limiti quando usa l'IA e a fidarsi della macchina
Chiedere alle persone dettagli di film: questo è stato l'obiettivo di uno studio per capire quanto il comportamento dell'essere umano vari quando ha a disposizione l'IA per farsi aiutare.
Lo studio ha visto coinvolti due gruppi: il primo poteva chiedere una mano all'IA per rispondere alle domande, l'altro invece poteva usare soltanto la propria testa. Inoltre, le domande e il modello di IA coinvolto erano stati selezionati per far in modo che la macchina avrebbe spesso avuto difficoltà nel rispondere correttamente (a insaputa dei gruppi).
Il "gruppo IA" ha ammesso di non sapere la risposta nel 3% dei casi, contro il 44% di chi l'IA non poteva usarla. Tuttavia, è stato il gruppo senza IA a dare più risposte corrette: il 27% delle loro risposte erano esatte, contro il 9% di chi era affiancato dall'IA. La gente si è fidata delle allucinazioni della macchina, forse ingannata dall'apparente coerenza delle risposte fornite da questa.
Non solo: chi aveva usato l'IA ha riportato un livello di fiducia nelle proprie risposte del 76%, contro il 30% dell'altro gruppo. In altre parole, la loro sicurezza era più che raddoppiata, nonostante la correttezza fosse nettamente diminuita.
Si è poi notato come, mettendo un incentivo economico sulle risposte corrette, l'ammissione di ignoranza di chi usa l'IA è salita dal 3 all'8%, mentre la correttezza dal 9 al 16% - suggerendo che l'IA non inibisce tanto la capacità di pensiero, quanto la volontà di pensare; e che un incentivo ha comunque un effetto limitato se si paragonano le percentuali con quelle del gruppo no IA. In generale, l'IA è stata usata meno quando ci sono stati soldi in ballo.
https://www.theregister.com/ai-and-ml/2026/07/19/using-ai-makes-people-less-likely-to-admit-they-dont-know-something/5274567
Michał "rysiek" Woźniak · 🇺🇦
boosted
@rysiek years ago the overall corporate org of the place I interned at wanted to get rid of a bespoke Oracle system for one department to Opentext.
The department kept resisting and persuaded the new leader of the project to switch them back. I'm still amazed that they persuaded him given the fact it pretty much was a mandate from Head Office in the US.
Competence has never entered these discussions, its vibes and a lack of , or aversion to training.
