AI may fail on you since it's a piece of technical equipment. The engineers among us know a SPG with it's only 3 parts in hardware are less prone to fail as a sensor with 5 parts and containing software.
They also know having equipment never fail on you, only brings you closer to the one time there will be a failure.
Actually, real engineers know that neither of these statements are true.
Not all parts are equally likely to fail. You can't just count parts to determine failure probability or MTBF.
Some SPG's are more reliable than others, even with the exact parts count. Material choices, manufacturing quality, and design choices matter.
Not all failure modes are driven by time. Many failures are random. Some become less likely the longer it's been since failure. For example electronics often rapidly become more reliable the longer they go without failure early in their life, and then aren't likely to fail until much later in life when failure slowly becomes more likely again.
Not all failures carry equal risk. If the software fails, there is no air lost. If a hose fails, there could be significant loss.
Your parts counts are wrong. Just the hose on the SPG has more than 3 parts: a threaded fitting on each end, an o-ring on each end, a hose connecting the fittings, a ferrule on each end to connect each fitting to the hose. Just there we have 7 parts. The case for the SPG has a body, a transparent face, and a rubber vent plug on the back. There are 3 more parts, and we haven't even reached the SPG mechanism itself. And don't tell me those parts don't count, there is a reason divers care whether the body is plastic, brass, or steel, whether the face is plastic or glass, etc.