I'm a female doctor, for me the hardest thing so far was watching AI shut down a discussion on harassment in healthcare in real time. Women were contributing stories to a website documenting their experience until one claiming to be written by a male was published. All contributors were fluent but this hit different - lower information density (harder to get a read on the age, ethnicity, social class of the writer) but higher emotional impact. Women usually seemed to have interrogated themselves, identifying choices they had made (joining after work drinks, accepting a ride, being alone) but this showed no sign of any attempt to think through how alternative versions of the account would look (eg, a white woman contributing to a forum on racism to complain discrimination by non-white men protected by the patriarchy, also they were on cocaine?) The last-minute drug reference was weird (doesn't tha complicate the question of who the victim is, not reinforce it? why cocaine - except that "crack" would sit in a field adjacent to both "abuse" and" damage" and "drug" next to "doctor")? Other men had shared experiences, but not like this. Feed the stories into GPT and ask which is most likely AI and it "carefully and respectfully" identifies 239 for different reasons (symmetry, etc) and offered to rewrite in a more human way. Nobody has contributed since. https://www.survivinginscrubs.co.uk/your-stories/ yeah I'm in an industry that for no good reason still writes instructions using a language that hasn't been spoken since about 800AD. Ai scribes might not change how we practice but it is having real-world effects.