Submitted 1 week ago by cm0002@toast.ooo to technology@lemmy.zip
https://www.quantamagazine.org/distinct-ai-models-seem-to-converge-on-how-they-encode-reality-20260107/
To me this points to there being a lazy trick for AI Models that precludes a more nuanced expression of reality that they all find themselves at from optimizing more revealing encodings of reality away.
from the article :
“In a 2024 paper, four AI researchers at the Massachusetts Institute of Technology argued that these hints of convergence are no fluke.”
Now, publish your own findings. No ? i didn’t think so either.
Why the hell would I waste my time studying such a lame, environmentally destructive, fundamentally anti-humanist technology?
supersquirrel@sopuli.xyz 1 week ago
To me this points to there being a lazy trick for AI Models that precludes a more nuanced expression of reality that they all find themselves at from optimizing more revealing encodings of reality away.
A_A@lemmy.world 1 week ago
from the article :
Now, publish your own findings. No ? i didn’t think so either.
supersquirrel@sopuli.xyz 1 week ago
Why the hell would I waste my time studying such a lame, environmentally destructive, fundamentally anti-humanist technology?