On the 747, for setting up the IRS, I literally just have to set the knobs to NAV and copy the GPS position into the IRS field on the CDU. This is nonsense. Why do you need three different navigation methods when you can get one GPS?
Both IRS and GPS (as well as VOR, and DME, when needed and available) contribute to the aircraft's position determination being used by the navigation system's computer(s) at any given moment.
IRS can provide a highly accurate aircraft position determination without re-updating (for hours) and is totally internal and self-contained (not reliant on outside navigation inputs if they are not available for some reason, such as GPS, DME, etc.) for nav purposes.
Nav system design makes the GPS a component of position determination. Maybe one day it will be the only component for navigation position determination. But today, for the type of aircraft in your question, it's only part of the ultimate position determination in use based on the aircraft's nav system design scheme.
While the other answers are correct about why GPS (or similar GNSS, such as GLONASS, etc.) aren't used exclusively for aircraft navigation, I would say that, these days, the premise of this question is no longer correct and GPS is the primary means of navigation for aircraft. It just is not the exclusive one and it most likely never will be because the aviation industry very understandably likes having backups to their backups to their backups, ideally ones that do not have common modes of failure, as would be the case with using other GNSSs.
That said, due to the extreme cost of getting new hardware certified and the slow nature of regulatory changes, the aviation industry has a lot of momentum and major changes take a long time, both to be approved for initial use and then to actually become ubiquitous. Airplanes, once built, tend to have very long service lives, especially when talking about an entire type of airplane rather than just one particular airplane. As an example, I'm in my late 30s and every airplane I've flown was built before I was born and was based on a type designed in the 1950s and/or 1960s. While airliners don't usually stay in service that long, we're still talking about production runs that last for decades and then service lives of decades beyond that.
Of the 4 aircraft that you mentioned, the second-newest of those types is the 747-8. Aside from being based on an airframe originally designed in the 1960s and designed to be flown by pilots who had been trained on previous versions of the 747, even the 747-8 itself began design in the early 2000s. The A310 was designed before GPS was even allowed to be used for civilian purposes and the A320 family (of which the A319 is a part) was also already well into its design phase before GPS was opened up to civilian aviation, though it had become available before it first flew. At the time the A310 and A320 were designed, airliners were still sometimes navigated at night in part by looking at stars with a sextant!
As for inertial reference systems specifically and why they are used in addition to GPS, they are actually great and can give much faster and more precise measurements than GPS in the short term. Orders of magnitude more precise. Additionally, they provide orientation information that GPS does not (i.e. GPS will not tell you if you are rolled to 35 degrees right bank angle and pitched 15 degrees nose-up, whereas the IRS will.) Over the course of a whole flight, though, their drift can become significant without something like GPS or radionavigation to give them occasional fixes. This is actually how your smart phone (or other portable electronic device) works, too. It uses GPS (and cell towers and other such things) for occasional fixes to factor out drift, but it uses an IRS with accelerometers and such that measure acceleration in different axes in order to work out short-term position and orientation changes. The same is true for virtual/augmented reality systems and such, except that they use optical systems for their fixes rather than GPS or radio signals (which wouldn't be nearly precise enough for their purposes.)
For many purposes the GPS is good is enough. Systems certified for RNP after all provide a 1x10-7 per flight hour containment guarantee and decent availability numbers. For other purposes, like total loss of position in IFR conditions, that may not meet the safety numbers wanted by engineers and airlines.
There are numerous reasons why IRU's are still installed, so often GPS provides the en-route navigation and perhaps the approach, while IRU provides a backup or cross-check for Position Alert messages. Of course you'll still need ILS, VOR, etc. for some approaches for mostly historical reasons that are too complex to get into here.
The biggest issue with relying solely on GPS is jamming. According to reporting by IEEE Spectrum lost or intermittent GPS was reported to the FAA 233 times in eight months in 2017-2018. That's alarming if you thought GPS was rock-solid reliable. Compared against millions of flights per year, it's nothing to lose sleep over but much worse than a good IRU since IRU have virtually no external interference. GPS issues may not be common at your home airport, but there are geographic areas where GPS issues are pretty common like Tel Aviv Airport, Ukraine (even before the war), the Turkey-Iraq border, and even White Sands Missile Range in New Mexico. (source: ASRS reports and discussions with pilots).
I literally just have to set the knobs to NAV and copy the GPS position into the IRS field on the CDU.
The user interface here is clunky, but the point is that you should only copy the GPS position if you trust it. It's like manually entering baro correction on both sides of the cockpit, or setting the heading on a directional gyro- if automated too much then a single fault could have catastrophic consequences. Could you design a unit that automatically syncs the GPS position repeatedly to the IRS as long as they don't start diverging? Yes, and that's more or less how a GPAHRS/GPIRU works.
Is GPS widely used, and increasing in usage? Yes. Coupled GPS-AHRS are replacing one or all of the IRUs in many new airplanes, so many commercial aircraft out there like CRJ 700's won't dispatch if all their GPS units experience issues. Overlay approaches are replacing VOR/DME, and there are now more LPV approaches than CAT I ILS approaches (source). The ADS-B mandate probably never would have happened without widespread availability of certified GPS units. However GPS's not the only source of position in any cockpit, just like you don't rely solely on one pitot tube, and like how magnetic compasses struggle in areas of magnetic unreliability like Manitou Springs.
This doesn't seem to be explicitly stated in any other answer: a significant part of procedures for flying airplanes comes from a desire for reliability, redundancy and fail-safes.
I can't tell from your question whether you are a simulator-only pilot or whether you fly physical airplanes. I can relate from my own experience, many hours of an instructor shutting off or covering up a random piece of equipment in my cockpit.
I enjoy simulators as well, and sometimes when things aren't going my way I just reset the simulation. This is not an option at 1500 feet going 80 knots. It was scary enough with advance notice of precisely what would fail and an instructor sitting next to me.
Luckily, I've never been in the position of having a real problem of this sort. I can just imagine what it would be like though, with 300 people behind me and approaching an airport, to find out my GPS doesn't work. Some other system may not work, but I only need one of them to work to land the plane.
IRS are used for attitude reference, not just location. GPSS has no way to know if you are flying upside down and sidewards, it can only show where your antenna was located when the last set of signals was received, not its angle or acceleration.
GPSS signals are disabled for testing frequently, not the whole system but targeted bubbles with a radius of about 400miles. (horizontal radius at ground is much smaller than the horizontal radius at commercial cruise altitudes). These zones are in NOTAMS that affect my area of the USA about once per month. In many mountain valley airports, there are not just enough satellites in view regardless of testing or outages.
Civilian GPSS alone does not meet the accuracy and reliability standards for precision approaches, and the position is always slightly behind (shows where you were, not where you are) due the nature of receiving signals and processing them. The IRS is much more accurate over short times, distances, and rapid changes in attitude and speed. Accuracy in this case is not judged on normal, best-case, or average performance it is judged based on expected worst-case performance(ie bottom 1% or 0.001%). Vertical reliability and accuracy seems to be the most limiting factor, though I can't explain the full engineering reason, approaches have the least buffer in the vertical dimension and GPSS is about half as accurate in the vertical.
If augmented with a local (at the destination airport) ground-based signal correction system(LAAS/GBAS) it can qualify as "precision" however the IRS or other combination of instruments and NAV aids are still needed to deal with a sudden failure of any part of that augmented GPSS, multiple external signals or internal processing. Only a few airports have installed local augmentation stations at this time. Though somewhat less accurate, wide area augmentation(SBAS) is a suitable alternative for en-route and less critical approaches. Upgrading a small plane from traditional ground transmission navigation(NDB VOR ILS etc) to a basic WAAS capable GPSS nav system is about \$10000usd, one with a big screen and many features is \$50000usd, which is close to the cost of purchasing a complete airworthy old small airplane.
Military GPSS has an additional inherent correction signal from the standard constellation to deal with fluctuations in the ionosphere. But even the one-time use military GPSS equipment (eg cruise missiles) are backed up by internal IRS. To deal with both signal jamming and rapid maneuvers.