We're Solving the Wrong Problem for LLMs and AI Overall

Part two of an accidental series. A few weeks ago, I wrote about LLMs having amnesia:

I wrote about LLMs and memory and the a16z continual learning essay, and I got the response I usually get when I write something like that and post on Linkedin: a bunch of people telling me I'm right for the wrong reasons, a few telling me I'm wrong for the right reasons, and one very long DM from someone I won't name who had clearly read every paper I cited and wanted to argue about test-time training for forty-five minutes. I'm not going to write about that email. What I am going to write about is a paper someone dropped in the comments that I almost didn't read because it was fifty pages long and published by a company called Onto AI, predating the a16z article. I assumed it was going to be the...

Copyright of this story solely belongs to hackernoon.com. To see the full text click HERE