Actor-Network Theory is such a weird creature: romantic about becoming, but hates metaphysics. No ghosts, no generative gaps, no imagination! 🍄
Anna Tsing's "worlding" fixes this: actors experimentally construct worlds through gaps & mistranslations. ANT's flat ontology misses the very […]
I have to admit I am currently deep in the rabbit hole of frugal and #selfhosted AI.
With a used Geforce rtx 3060 (12 gb vram), gwen3.5 (35B A3B) is an amazing toy.
For more context on frugal AI:
https://restofworld.org/2026/frugal-ai-big-tech/
Three days at #stsing in Bochum, thanks to the organisers for putting this fantastic conference together! STS' future in Germany is in good hands! And I hope all the smart people without permanent positions that I met will soon be as lucky as I was when I was in their situation!
Does this make sense?
"Quillx is an open standard for disclosing AI involvement in software projects - expressed through the language of authorship. Not a judgment. Just transparency."
https://github.com/QAInsights/Quillx
The ruins of Ephes, the rest of a temple with 7 Ionic columns in the front, and what may have been a street lined with monuments in the back. (Getty images, published under the unsplash+ license)
In an age of #AI #sycophancy, a worthwhile task for the #humanities might be to be the slave who whispers into the ear of the pandered-to individual: remember you are only one human.
Apropos the Roman tradition: during a triumphal procession, a slave stood […]
[Original post on tikatech.org]
Today a student reminded me that hacking/appropriation is the better alternative to pragmatic acceptance and principled rejection, which are my two reactions to new tech.
In this spirit: I finally found a way to have command line access to my work-emails and appointments which are involuntarily […]
Little reminder for those who still take marketing at face value:
If you are paying someone else to use a tool you don’t own, it is not a "technology", it is a "service".
You are not "learning" or "mastering" or "building", you are "spending".
You are not an "entrepreneur", you are a […]
After having "created" a native #SailfishOS app with claude code, I am happy that refusing to buy into Android or iOS is getting so much easier. This could weaken the grip that app and play store have on the mobile os market. However, it is clear that this is individualising something which […]
On the other hand: maybe the hope to be able to change for the better is not only ideology: the author of the article scores high on the "openness" trait, which should support change of the kind reported in the underlying research.
Of course in this day and age we won't accept that some parts of ourselves cannot be optimised: How to change your personality in six weeks:
www.bbc.com/future/article/20260102-...
Not that many people here will need it, but as an indicator it's interesting:
In case you're looking for an alternative New Year resolution, Denmark has a new people's movement: Danmark skifter, Denmark switches, which aims to help people regain focus, change their social media habits and get […]
"Technical reports often compare LLMs’ outputs with 'human' performance on various tests. Here, we ask, “Which humans?” [...] We show that LLMs’ responses to psychological measures are an outlier [...], and that their performance on cognitive psychological tasks most resembles that of people […]
This year's dark season distraction: doing research about the largest cities in the world, and practicing some good oldfashioned, climate friendly daydreaming about distant places.
An untidily stacked pile of computer print outs.
#Academic housekeeping: This is the pile of drafts, exams and applications that an average prof in Norway reads and comments on in one year. Every text is an expression of hopes and aspirations, which I am quite aware of while reading. Not much AI-mediocrity in […]
[Original post on tikatech.org]
Intelligence amplification or assistance (IA) instead of Artificial Intelligence (AI) makes a lot of sense to me.
@viznut reminds us that algorithms are neither good or evil and that while big-tech creates big and centralized AI we should look for small and decentralized @permacomputing […]
An open old Huawei server and a Geforce gpu and a lot of cables and stuff.
My Frankenstein data centre has a new addition bought from a gamer who I guess was upgrading. Here no games but 4K #jellyfin transcoding and quicker #selfhosted LLM answers. Turns out a consumer GPU does not fit at all in a server and I needed a riser cable and […]
[Original post on tikatech.org]
A weekend with some more experiments with local #LLM: these monsters are so much more useful when fed with a lot of context - mostly about me - which for obvious reasons is verboten in the case of chat-gpt & co. Reasonably good results with Alibaba's #qwen (通义千问 3:235b-a22b, via #ollama). It is […]
Simple and good about "What AI is Really For":
https://www.chrbutler.com/what-ai-is-really-for
"When a private company can construct what is essentially a new energy city with no people and no elected representation, and do this dozens of times a year [...], you have a sudden imbalance of […]
Nerve wrecking joy: updating my mastodon instance without having a clear idea of I am doing. I think I made a backup of the database, but maybe I did not? Usually there is next to nothing at stake when botching an update of my #selfhosted services. But losing my dear followers would indeed be […]
Mmmmh, #writerdeck, small, cheap computers that can do only one thing: distraction free text prosessing. Very tempting, and many diy options. On the other hand, I have bought so many devices through the years, and usually they gather dust in a corner after a week of intensive use […]
Just realised that a lot of what I talked about at the oppositional responses to ai workshop (#cscw2025) today was inspired by Sally Wyatt's brilliant idea to do research on the coming AI winter.
sallywyatt.nl/imaginary-research-about...
A monitor showing the niri desktop with the settings for dms open. A wallpaper from Scavengers reign shows a pot of leaves.
Unreasonably excited about my new Linux desktop environment: The wonderfully functional and performant #Niri (1) made beautiful and complete with #DMS (2). Redefining the sweet spot between bling and function, the holy grail of desktop environments […]
[Original post on tikatech.org]
Back to my #android ( #grapheneos ) phone after another week with the #sailfisos C2. Defeat. Most things work well enough on the C2. But my employer #ntnu is locked into the Microsoft ecosystem and has disabled mobile access to email and calendar from outside Outlook. Not being able to ckeck my […]
I am teaching an introduction to #sts on the MA level this fall and did a little anonymous mapping of #ai use at the beginning. Most interesting findings: for almost all of these 20 students AI = #chatgpt. And almost all expected that AI will lead to profound changes in the next decade, with the […]
Trondheim's Good infrastructures lab is looking for a #PhD student who wants to write about the fediverse as infrastructural hope. Norway treats its PhD students well and we will help you to make your application competitive. For the call, see here […]
Finally switched from #plex to #jellyfin for my streaming needs. Main reason: using a tool instead of a (wannabe) platform is such a mental and esthetic pleasure as it eliminates clutter and reduces complexity. Jellyfin does fewer things and it does them well. Platforms all dream of becoming […]
"Young people have always felt misunderstood by their parents, but new research shows that #GenAlpha might also be misunderstood by #AI."
Rosie Thomas of @404mediaco summarizes the findings of a really interesting paper that focuses on how #LLMs (mis)interpret teenage slang and meme culture […]
@skyglowberlin
Claude (which obviously has Internet access):
"University: Christopher Kyba received his PhD in Experimental Particle Physics from the University of Pennsylvania in 2006
Thesis Title: "Measurement of the Atmospheric Neutrino Induced Muon Flux at the Sudbury Neutrino […]
@skyglowberlin I just asked Claude and it got place and title of my really obscure thesis exactly right. What is wrong with LLMs is unfortunately not only that they are wrong.
That no ownership of the means of production always was linked to exploitation of work should be an easy way to convince knowledge workers to not use any cloud products (including non-local LLMs). Unfortunately, it is not.
#selfhosting #localfirst #work