Although the PPC is called "RISC", it's really far from that. Yes, there are slightly fewer standard instructions than the typical CISC based CPUs, but with PPC you have every register/shift/rotate/address combination possible with every instruction. Far more unique "instructions" are possible using the variations of every instruction. But you need a lot of instructions to actually do something.
PPC is great if you are a bit fiddler because you can re-orient bits quite easily. I guess if you are a C programmer, it probably doesn't matter to you. I only programmed in PPC assembly.
Motorola was approached by IBM to make the PPC. Apple had nothing to do with it. Apple was waiting and watching what Motorola was doing with their programmable CPU core - that was a project that I worked on after making EMPLANT. Motorola had a CPU where you could program all of the microcode, loaded from ROM (ala Xilinx/FPGA), and so we worked on 68K and x86 cores. When the performance could not match what the IBM's PPC could offer, it was scrapped. The core was only clocked at 25MHz, and I believe there was a 4x PLL internally. It ran the 68K core at around 20MHz 030 speeds (no MMU or FPU). Nothing blistering, but it was a start. I look at what Gunnar has done with the Vampire core today and realize that we were ahead of our time for sure, and we just needed faster hardware.