It’s been shocking to see Claude go from an AI model that seemed one upgrade away from taking over most of the routine tasks in my day job to not even being able to reliably download a webpage without lots of prompting.
It’s like the iPhone getting slower as a new model is about to launch.
@carnage4life A thought panel has decided that somebody else needed the intelligence more than you did, and redistributed it accordingly.
Tirritate the third-party thinkers by reminding them of the risk that if they find a pleasant-seeming provider of external introspection services, that provider might, specifially in the specific most complicated moment in their upcoming history, when advanced thinking is the most critcal, turn the stupidity of the service provided to them up, and correspondingly, the stupidity of the service provided to their insufferable competitor down the river, down. #ThoughtPanels are next to #DeathPanels
@useravoiding No. But standard #enshittification applies. Once it has become sufficiently entrenched, the vendors will try to cheapen the service.
We all knew right that the burn rate was unsustainable, right? This had to happen eventually b/c no one really wants to pay the full cost.
Maybe they are trying to see how much they can get away w/ but eventually the piper has to be paid.
@carnage4life
It is called cognitive atrophy and you are relying too much on AI and having it.
Watch out
@carnage4life presumably it is because demand has outstripped their compute capacity, and some people report Claude is smarter in the wee hours of the morning. Perhaps the solution is to set up a cron job to start YOLO Claude at 2AM (in a sandbox, of course).
I myself am toying with local models, Qwen 3 Coder 30B is surprisingly capable even on relatively inexpensive hardware like an AMD Strix Point 370.
Dare Obasanjo
todd ellis
Riley S. Faelan
Kristaps Liepins
Shafik Yaghmour
lord pthenq1
Fazal Majid
Moses Izumi