The author of the tweet is a physicist. Her work is at the edge of the boundary of human knowledge. LLM’s are useless in this domain, at last when applied direc

The author of the tweet is a physicist. Her work is at the edge of the boundary of human knowledge. LLM’s are useless in this domain, at last when applied directly. When I use LLM’s to explore applications of cutting edge nonlinear optics, I too am appalled about the quality of the output. When I use an LLM to implement a React program, something that has been done hundreds of times before by others, I find it performs well.