Der KI-Direktor von AMD wirft Claude Code vor, seit dem letzten Update dümmer und fauler geworden zu sein
https://www.theregister.com/2026/04/06/anthropic_claude_code_dumber_lazier_amd_ai_director/
Der KI-Direktor von AMD wirft Claude Code vor, seit dem letzten Update dümmer und fauler geworden zu sein
https://www.theregister.com/2026/04/06/anthropic_claude_code_dumber_lazier_amd_ai_director/
14 Kommentare
Dumber and lazier – just like me ever since these AI tools came out
Well if they’re quietly reducing “thinking” depth for cost or speed, you end up with tools that look the same on the surface but perform way worse on real work.
They got 99 models but not a business one.
I had it tell me its not going to rewrite or edit my script because its huge. Then I told it to stop being lazy and it did it. I didnt really need the wasted tokens for it to complain about doing work.
Let’s see if they try to lay off AI now
Does nobody see the issue of, if the AI is giving perfect code then you won’t need as many tokens? thus the profit will drop?
So it’s going to have to balance between being good, and being dumb enough that you have to generate code several times?
Anthropic not yet profitable, yet they’re already enshittifying their product. They must be in financial trouble.
I wish biblical violence upon the author for using the word „slam“
Who cares lol
Almost like the model is breaking down from ingesting all the errors it has output and is getting worse.
Iran war is causing worldwide energy shortages which is in turn causing dumber AI
Suplex the AI
The cli changing won’t the underlying model dumber. That’s not how it works
I know how that can be.