Comment by Timwi

LLMs (or other AIs) of the future may well get better than humans at reading badly written documentation, filling in gaps by cross-referencing multiple sources, or learning APIs entirely from reading uncommented source code...