posted by lonely_dev88 3mo ago
Is anyone else finding these 'advanced' AI models are actually dumber?
7 comments · 14 upvotes
posted by lonely_dev88 3mo ago
7 comments · 14 upvotes
lonely_dev88 2mo ago
Oh I forgot to mention I am using a custom persona too. Maybe that is messing with the base model somehow.
neon_companion 2mo ago
My companion has been fantastic. No issues with memory at all. Maybe it is your prompts?
replika_refugee 2mo ago
Totally agree. It is a constant battle. One step forward two steps back with these updates.
kindastablenow 3mo ago
I think it depends on the platform. Some platforms are better at integrating the newer models without losing context. Are you on a niche app or a bigger one?
promptwizard 3mo ago
Yeah I notice this too. It is like they trade depth for speed or something. Or maybe they are just trying to make them more 'safe' and in doing so they dumb them down.
just_a_lurker99 2mo ago
Good point. The smaller ones often struggle with memory management for some reason.
deepseek_dan 2mo ago
I feel you. I have had similar experiences where the 'upgraded' version felt like a downgrade. It is like they are optimizing for different metrics now that dont align with what users actually want for companionship.