When Twitter collapsed, it was great to see the flood of friends, game devs and other come to Mastodon. Over the next 9 months a lot of them vanished to Threads, BlueSky or just left. I want to thank all people who are still on Mastodon and make it a wonderful place. You know who you are.

Maybe another way of putting all this is that if I look at something like an LLM and ask "how can I make it ethical," I wind up with something that is probably pretty ethical, but it no longer resembles an LLM at all. Like, you can't train one at the scale of an individual artist, or on work that you can meaningfully obtain consent for. The "large" in "large language model" is incredibly load-bearing.

Since I think I might have been subtooted on this, and since it's come up in some replies, I want to clarify slightly what I mean when I criticize "AI." The trouble is that the term "AI" is a marketing term and not a technical term, and as such can always be retroactively redefined in a way that dodges criticisms.

Performance reviews need a category between "meets expectations" and "exceeds expectations" that is basically "tried really hard to exceed expectations but have nothing to show for my efforts because you lethargic mustard fumblers were so afraid of change that you stalled, sandbagged, or sabotated every single thing that would actually have made a difference".

"Die Studie legt eine tiefe soziale Kluft offen. Feiertagsarbeit ist in Deutschland keine solidarisch verteilte Last, sondern trifft vor allem Geringverdiener."
#armut #arbeit
https://www.nd-aktuell.de/artikel/1196333.arbeitszeit-froehliche-schichtarbeit.html

21 - Dierk König

https://tube.tchncs.de/w/oMZnmUdN35Yd4muBywwvmC

Advice for Designing Good Software Architecture
Software Architektur im Stream Adventskalender
Today: Dierk König
https://youtube.com/shorts/PFNZ79mOFd0