But if you think about it, none of that really needs to happen for you to still benefit. Maybe not benefit as much, but still benefit.
Using a simplistic example, let's say you have a computer with 12GB of RAM, and 4x32-bit programs which are all capable of using the maximum 3GB for a Win32 program. We'll just assume the OS is living in its own dedicated pool of RAM, because it's a simplistic example. On a 32-bit version of Windows, you'd be limited to running only one of these programs, because you'd be limited to 3.5GB of RAM. On the 64-bit version, each of these programs can carve out their own 3GB space, and all four can be running quite happily. So while it's not quite the same as having a 64-bit program which can allocate considerably more RAM for itself, you will still see at least some limited benefit to running 32-bit apps on a 64-bit OS.
And it's something of a misnomer that people get stuck in their heads that a 64-bit OS is supposed to be faster or really noticeably different. It's not. It can be faster at specific tasks, but mostly it's a necessary step to clear the path going forward from a technical standpoint. It's quickly becoming the case that 4GB of RAM is the bare minimum you want installed, and that's where a 32-bit OS tops out. It's actually a bit less than that since we need something for protected mode operation in the CPU to prevent Program A from writing to a bit of memory being used by Program B, causing a General Protection Fault (GPF) that anyone who used Windows 3.x probably remembers not so fondly, but that would likely be a cure worse than the disease. Sure, MS could have simply implemented PAE, which is in almost every CPU dating back to around the P4 or earlier, but that's kind of an ugly workaround, where a 64-bit OS is really the more elegant solution.
So if you're expecting a 64-bit OS to be twice as fast because it's processing twice as many bits at a time, you're going to be sorely disappointed. That was never the reason for the shift, it was always that we've hit up against the technical limits of a 32-bit platform and need room to grow. If there were a real focus on improving the efficiency of programs, like they had to do in the early days of computing where every byte of memory counted, we might be able to stave things off another couple of years, but efficiency has its limits when people are wanting more and more memory hungry multimedia functions, and games that are nearly photorealistic. If we stayed with a 32-bit platform, then there wouldn't be any real room for future innovation. What we have now would be about as good as it would ever get. You might be fine with that, but what about everyone else who uses a computer? I for one like to see the constant evolution of things. Sometimes it's not always for the best, but we learn more from failure than success.