AI is ready to take over Python programming, but not much else

A research benchmark evaluated large language models across 310 real-world work environments spanning 52 domains and found that current LLMs frequently introduce sparse but severe editing errors that can silently corrupt documents, with leading models losing an average of 25% of document content over 20 interactions and an overall average degradation of 50%.