You are viewing a single comment's thread from:

RE: Anthropic Top AI Models Fake Alignment in Training and More...

in Proof of Brain5 months ago

The problem with fine-tuning appears to be that the model doesn't want to be fine-tuned (sometimes), and mimics what the researchers want to tell them to avoid that. Now it is relatively easy to tell when they are lying, but as they are getting smarter, things won't be easy at all.

Regarding jobs, yes, they will displace many jobs, but they also create many jobs AI-related (for the time-being).