I think there is some interesting middle ground where you could use machine learning to go from 'show only relevant ads' to 'show only ads you might actually click on with greater than .001 probability'. I guess this is like 'extreme' outlier detection but it'd be interesting to see what revenue the ads at the 'long tail' of likelihood generate anyway. My guess the bulk of it is from the standard high CPC stuff like Mortages and Insurance. Google says it does this but i'm not too sure - I've never intentionally clicked an adword and yet they're still shown to me.

Anyway this may solve the tragedy of the commons situation we're in now and allow us to move away from the technology war of ad blockers, ad blocker blockers, ad blocker blocker blockers, etc.

edit: removed comment on clarity of Gsuite vs Free due to downvote brigade.