Bokeh, AI, and selection

I've started playing with Adobe's new Neural bokeh filters and Skylum Neo's bokeh AI filter.  Of the two, Neo beats Adobe for people (really only good for people), Adobe beats Neo for general subjects. Neo beats Adobe hands down as it has a selection brush to fine-tune - which is pretty much essential in many images (C'mon Adobe! Really, no adjustment brush?).
Now hear me out - I try hard to get it right in-camera always. I'll choose a 1.2 or 1.8 lens to get the subject isolation and quality of bokeh I want, but sometimes I have to use a slower lens like an F4 or variable aperture that doesn't sport 2.8 or wider. Sometimes even the 2.8's don't give me enough. And I'm good at PS and LR with blur and masking, but the new crop of AI filters are virtually automatic with easy adjustments for faster and dedicated post-processing.

My question is - as these AI bokeh filters get better and better (and they are already good enough for production use for many images), how will that affect lens selection - both the purchase of fast lenses and the use of them even when they are in the bag? Already we've gained incredible low-light machines with better high-iso, great NR, and up to 5 or more stops of IBIS in the latest cameras and lenses.  As someone that came from the film era in the 70s and the earliest DSLRs, it's rather astonishing.
Now one still can't beat getting it perfect in-camera to avoid extra post-processing time (can't wait until we get it in LR and can do non-destructive batch), and the extra stops of light is always more desirable - especially in very low light venues, but will all of these things combine change the decision process of purchasing and using really fast expensive glass?
Mike