The way Automodding seems to work now is that if X% of the pop group is working in job Y, then automodding pushes that pop group pop toward having X% of the Y trait. So if you have e.g. 33% researchers and 66% bureaucrats, you'll get 33% of the effects of Traditional and 66% of the effects of Intelligent.
The game tells you what that % is ("efficiency" under each trait) and what % of pops have the right trait for their job overall at the bottom (under e.g. Adaptive Frames: Current Efficiency).
The net result is that Automodding is nearly perfect on very specialized worlds, but only 1/3 to 1/4 effect on planets with mixed pops. Which makes it especially bad on your capital to start.
Aka, it's now a rough substitute for pop modding micro without the work, rather than a "all your pops get perfect traits for free" trait. There's actual value now in taking Efficient Processors instead of Adaptive Frames.
Overall, I like it. It's a much needed nerf. But I do have a problems with its current implementation: Jobs that have no trait should be excluded from the trait weighting. If I have 250 researchers, 250 bureaucrats, and 500 metallurgists, I should get a 50/50 split on Intelligent/Traditional, rather than a 25/25/50 split between Intelligent/Traditional/nothing at all.
Side note: the jobs that have an effect on ethics and an attraction for pops working a certain ethic (like researchers and priests with Spiritualist/Materialist) should then end up having better performance, since the pop groups working them would prefer working that job over others.
i.e. If your spiritualist pop group is working all the priest jobs because they have slightly higher weight, then no other ethic's pop groups will have Traditional sucking up automod preference, and your spiritualist pop group should have a correspondingly higher Traditional %. So a 50/50 priest/researcher split may end up with 75%/75% efficiency, rather than 50%/50%.
I have no idea if those job weightings/ethic effects even work in 4.0 anymore though.