peter purgathofer

what would i know?

peter purgathoferpeterpur@hci.social
2025-06-24

contrary to my intent and against better knowledge, i have created yet another collection of ressources advocating a critical view on generative AI systems.

i will probably maintain this list for the forseeable future, which for me usually means 6 weeks or so.

if you have something that fits it, please let me know (eg. via a comment there) and i will add it.

taskcards.de/#/board/aa077e4f-

peter purgathoferpeterpur@hci.social
2025-06-22

»it is never possible to please dead people«

laurie anderson 🧡

peter purgathoferpeterpur@hci.social
2025-06-21

now reality has troubles with fingers? is this some sort of AI-spillover?

peter purgathofer boosted:
2025-06-21

God, this bubble burst is going to be so brutal

It’s a google screenshot showing the results for the search “all you can eat buffet near me.” The result is an AI overview advising that the user type “all you can eat buffet near me” into a search engine
peter purgathofer boosted:
2025-06-17

I went to a talk lately that was mostly about something else, but the speaker came out with:

“If you only remember one thing from this talk, remember this. Everyone in this room who likes helping people, raise your hand.”

Every hand, or nearly every hand, went up.

“If you like asking other people for help, keep your hand up.”

Almost every hand went back down.

“As you can see, people like helping you. When you ask for help, you’re making them feel good, even if you don’t like asking.”

I’ve genuinely forgotten the rest of the presentation but I won’t forget that.

peter purgathofer boosted:
2025-06-16
Craig Shackleton
‪@craigshackleton.com

It is really really sinking in for me the degree to which LLMs are the ultimate expression of right wing anti-intellectualism. Its proponents are literally mocking the idea that anyone would ever want to learn anything, know anything, develop any actual skill, or have a thought of their own.
peter purgathoferpeterpur@hci.social
2025-06-16

»The use of LLM had a measurable impact on participants, and while the benefits were initially apparent, as we demonstrated over the course of 4 months, the LLM group's participants performed worse than their counterparts in the Brain-only group at all levels: neural, linguistic, scoring.«

arxiv.org/pdf/2506.08872

peter purgathofer boosted:
2025-06-11

Et pour fermer la gueule des pro-AI qui se croient malins:

La console Atari de 1977 bat systématiquement ChatGPT aux échecs niveau débutant.

Cerise sur le gâteau, ChatGPT est mauvais perdant et dit que c'est parce que les icônes de l'Atari sont difficiles à reconnaître. Pov' bichon.🥺

futurism.com/atari-beats-chatg

#ai #llm #code #programming #tech #ordinateur #chatgpt #techbros #news #science #recherche #IA #game #atari

peter purgathofer boosted:
peter purgathoferpeterpur@wien.rocks
2025-06-11
peter purgathofer boosted:
Dr PenDrPen
2025-06-09

Another happy story from the world of extractive techno-capitalism.

A screenshot of a tweet with a about how someone has fed all of someone else's writing into chatGPT thinking it's fine. The writer whose work has been fed into AI said it hurts their chest.
peter purgathofer boosted:
bison ✅bison
2025-06-06

Conni fackelt das AI Rechenzentrum ab

**TL;DR:** Parody cover of a Conni children's book shows her burning down an AI data center with a match.

**Extracted text:**
*Conni fackelt das AI Rechenzentrum ab*

**Description:**
This image mimics the style of a German "Conni" children's book cover. It shows Conni, a cartoon girl with blonde hair tied in a red bow, smiling while holding a lit match. She stands in a server room, and one of the server racks labeled "AI" is engulfed in flames and smoke. The illustration is colorful and cheerful in tone, despite the destructive act. A small cartoon cat appears above the title. The title translates to "Conni burns down the AI data center," making this a satirical or meme version of a typical Conni story.
peter purgathofer boosted:
2025-06-06

What's been lost...
Entomologists from Krefeld, Germany, collected flying insects for two weeks in August 1994 (left) and—at the same site, with an identical trap— in August 2016 (right). Similar data from 63 German protected areas overall gave a shocking result: a 76 percent drop in insect biomass between 1989 and 2016.
It has worsened significantly in the past 9 years.
Photographed at Entomological Society of Krefeld

Two cylindrical 2 liter lab glasses, both containing a light brown fluid with a mass of insects at the bottom. The left glass gas more than 1500 mL of insects, the right glass around 250 mL.
peter purgathoferpeterpur@hci.social
2025-06-03

💙❤️

peter purgathofer boosted:
Debora Weber-WulffWiseWoman@fediscience.org
2025-05-26

@JMarkOckerbloom @peterpur

I wrote to the "author", he felt he was not able to use AI and didn't know how this sentence got in there. Right. I got a journalist to publish an article about it in German in the FAZ, in print and online (behind a paywall, unfortunately): faz.net/aktuell/wissen/forschu

peter purgathofer boosted:
peter purgathoferpeterpur@wien.rocks
2025-05-25

das fühlt sich falsch an, ganz falsch. fast obszön.

gestern wurde beschlossen, diese servierform »nacktschnitte« zu nennen.

eine packung mannerschnitten, aber die packung ist so aufgerissen, dass alle schnitten als geschlossener block sichtbar sind.
peter purgathofer boosted:
oleschri@chaos.socialoleschri@chaos.social
2025-05-25

" 'Diese Modelle wurden gebaut, um Wörter vorherzusagen, nicht um die Wahrheit vorherzusagen oder wiederzugeben', erklärt Informatik-Professor Jannik Strötgen von der Hochschule Karlsruhe im Gespräch mit ZDFheute."

zdf.de/nachrichten/panorama/ku

#ai #llm #chatgpt #gemini

peter purgathofer boosted:
2025-05-24

Zwei Engländer haben eine Anleitung für ein #Mikroskop aus dem #3DDrucker publiziert. Zunächst fällt es durch seine quietsch­bunten Plastikteile auf, dann aber mit seiner hohen Präzision und Automatisierbarkeit. Andrea Pitzschke hat sich's genauer angeschaut: laborjournal.de/editorials/319

Zwei Aufnahmen eines fertigen bunten Mikroskops aus dem 3D-Drucker. Rechts daneben Überschrift "Flexibler Mikroskop-Würfel" des im Post verlinkten Artikels, gefolgt von dessen Vorspann.
peter purgathoferpeterpur@hci.social
2025-05-19

the speed with which literally everything is turning into shit is overwhelming. while i accept that there are some legitimate productive uses for genAI systems, the rate with which people are finding horrible uses that ultimately damage every aspect of the internet is fascinating, in a terrible way.

if i had a chance, i would cut the genAI computing infrastructure from the electric grid.

peter purgathoferpeterpur@hci.social
2025-05-13

i have no idea why i find this funny; it is the kind of cannot-stop-grinning-funny that seems to draw from some strange, old, internal wound.

Client Info

Server: https://mastodon.social
Version: 2025.04
Repository: https://github.com/cyevgeniy/lmst