I feel that the trend towards training models to autonomously go off and try to do everything themselves is anti-human. We should, IMO, be training LLMs to support humans in their learning, creativity, and iterative experimentation.

→ View original post on X — @jeremyphoward