When Robin Hanson was writing Age of Em in 2016, he was thinking that Ems will be possible with advancement of brain-scanning technology in a hundred years; what he didn't anticipate was the transformer models, and their capabilities, as well as introduction of AutoGPT paradigm (I proposed the name ICA Simulacra in this post, but it is obvious now that AutoGPT is the label everyone will be going with).

If capabilities will increase further as they did before, the "new paradigm" of worker+ChatGPT will not even live for several years; AutoGPT will replace that.

Why do you need a marketing department?

Just hire an org who can set up a fleet of AutoGPT "Ems" to do these tasks, hire one expert marketer to verify their outputs and make sure they are all in line, and save thousands of dollars.

Hell, when you see that this expert marketing guy is just signing off without any changes, setup another AutoGPT in his place. 

And then some crypto guys, riding the wave of AI hype, will make a DAO consisting solely of AutoGPT Ems. And this or the other one down the line will succeed.

And some other guys, or the same ones, make an org with a sole objective of introducing better AutoGPT Ems.

If government will act quick and smart, they will manage to introduce UBI before wide-scale riots and society collapse.

And then we die either way.

Age of Em will be a decade of Em at best.