Tyro TCAS is not a three-dimensional solution. It deals with three-dimensional problems in a two-dimensional way - ie it demands(it does not command) a vertical manouevre to avoid the perceived problem. It cannot demand a horizontal manouevre. Its thresholds are relatively generous, which is one of the reasons why a visual sighting of the 'threat' aircraft allows the pilot to override the demands of the system. But it's a good piece of kit, and probably the single most significant aid to flight safety introduced in the last fifty years,
However, it's a hell of a step to extrapolate the logic built into TCAS and extend it to the entirety of all the decisions that must be made in the execution of a long-haul scheduled passenger flight. There are far too many grey areas for any computer program to reliably and repeatably solve for it to be considered a safe way of conducting flights - for the moment. Believe me, I do this job for a living - with the aid of the best computers Airbus can give me. To put that in perspective, the computers in my brand-new Airbus are somewhat less capable than the Intel 486 I had on my desktop 10 years ago - and the software is mickey-mouse compared to FS2004! But it is reliable (more or less - and a lot more so than anything Microsoft) - and the passengers won't die if the software or hardware fails, because I and my colleagues are there to flexibly and appropriately apply our expertise.
It will come, but not in my professional lifetime.