It's not that the laws of physics are different at that scale so much as that different laws become far more important at smaller scales, e.g. electrostatics and air viscosity play a bigger role with a micro-membrane than with a great big cone. There is no new physics, only physics that engineers are not so used to dealing with.
One of the issues that I recall seeing on the videos from AP was that the response from the MEMS was linear, albeit sloping. This contrasts with almost every speaker I've ever heard of (and I am admittedly no expert), where the response characteristic is a significantly variable curvy line. This is a direct and unavoidable result of the actual mechanical properties of the physical structure of the speaker itself. This is all very familiar physics, so engineers are quite used to designing exquisitely beneficial structures and cases to get as good a response as possible from their speaker system. And this approach is definitely required to get that. This is why good speakers have traditionally cost a small fortune.
But the whole point of AP's system is that none of this is required. The MEMS array response is linear, and is able to be made
dead flat if necessary, so the original (albeit) digital sound is reproduced with negligible acoustic artefacts from the speaker itself. This is one of AP's principal claims to being a
very good idea!
Of course this is what has been promised for the technology once they get it working correctly. I will leave it to the experts to comment on how well they have achieved their hoped for performance. But I'll bet Mike Klasko will have something to say fairly quickly once he has heard it!
The AP website has an interesting video on this from 2016:
http://www.audiopixels.com.au/index.cfm/news/videos/the-distortion-of-sound/
I have not seen it for a long time, but no comment on the video's contents was made on the AP site interestingly, but there is a message there for sure, or they would not have put it up.
(20min delay)