I recently used ChatGPT canvas to improve a friend's resume. I had managed him in the past, and he is great to work with. However, from his resume you would never know it. I did 3 hours worth of editing and improvement in 5 minutes. I gave rambling descriptions of what he had done when working for me and why it was impressive, and GPT translated this into resume speak almost instantly. I then gave some vague suggestions for improvement, made a couple minor edits to buff off the AI veneer, and viola, he had a professional resume that will help him do great work in his next position.
Seriously, a use case where LLMs really shine: bullshit padding. It's quite amazing to be able to turn ramble into coherent text. The less obvious tax is that if you don't put effort you will never be able to learn how to write and that will most likely affect thinking too. Writing will remain a valuable tool for improving thinking though, but likely for a lot fewer people.
I suspect we'll start to see people asking for more raw bullet-points for certain communications, because all the text they used to use to guess engagement/thought/investment/care is debased from easy counterfeiting.
In some cases that'll remove something that was always a questionable waste of time, and in other cases it'll represent a real loss of information.
Did you consider that the AI screeners would have liked the AI veneer? That CV will be "read" more times by a machine than it will be by humans. Someone should publish an RFC for a machine-to-machine resume spec so we can ditch the inconvenience of using human language as the transport
