People who were reading computer magazines at the time should remember all of the craze and buzz about the idea of the convergence of Unix (which was always too expensive for us mere mortals) and personal computers. Also, everyone talked about Unix System V Release 4 as though it was going to solve the Unix market fragmentation problem. Both things didn't happen in the way that the journalists imagined - no inexpensive Unix or Unix-like OS came to the personal or microcomputer market, and the Unix market itself remained as fragmented as ever.
Lots of people - myself included - thought that the advent of real CPUs and MMUs would herald a new era of proper OSes which didn't crash, didn't need to be rebooted, were secure, and internetworked easily. The Dark Ages of computing lasted MUCH longer. In 1985, I thought it'd just be a handful of years, in 1990 I thought it was imminent, but when nothing materialized I realized that it wasn't going to happen for a much longer time. I'd say that it took until 2005 before we began our Age of Enlightenment and people started realizing that computers didn't need to crash, didn't need proprietary networking protocols, and didn't need to be as insecure as a screen door.
Unix is what has saved the day, just as everyone had predicted, but it took 15 years longer than we all thought to get to OS X. OS X is NeXTSTEP merged with Mac OS. A/UX could have been the beginning, but AT&T Unix was not the way to go. NeXT is much more BSD-ish and NeXT was not afraid to make completely new software APIs along with its new interface. If Steve had done NeXTSTEP as a project at Apple, we might've had OS X in the mid 1990s.
The perfect time to do the transition would've been during the transition from m68k to PowerPC. m68k code needed to be emulated, so that would've been an excellent time to encapsulate the "classic" code in a "classic" mode while running a new OS with protected memory, preemptive multitasking, a new, updated filesystem, and a permissions model.
Regardless, from the perspective of anyone in the late 1980s, A/UX was certainly reasonable. Or at least the idea of A/UX was reasonable - Unix was the future, as Steve proved with NeXTSTEP.
Oh, well!