I don't know anything scientific about the Surround: AI. Supposedly it analyzes "the scene" and dynamically "adjusts channels" to more properly image the sound. I don't know what is meant by "the scene" and "adjusts channels". Is a scene a frame? Does channel adjustment mean volume and tone? And what is the practical effect on a listener?
You may not hear the effect due to the topology of your speakers.
I am generally a "Let my Onk pick the decoder automatically" guy. There are times when I over-ride it though. Nora Jones - Live in New Orleans now sounds horrible in Dolby Digital. I switch it to stereo and it's more enjoyable. Even more enjoyable is PLII or Neo (those are simulated surround settings).
I'll just add in here that I am running 4.5. I have total holography. I see no need to go to Atmos. Stereo is absolutely magnificent and holographic also.
Edit: I just found this totally laughable article by Yamaha. What a piece of shit! I would have been totally embarrassed posting this. What this describes is what every multi-channel receiver has done since the dawn of time. Absolutely deplorable. I really despise technical writers who not only aren't technical but also cannot write. Joy Frye, take a hike. You're such a waste of time.
Then again, maybe Surround:AI is nothing but marketecture.
https://hub.yamaha.com/what-is-surroundai/