Unfortunately most airplane's/instructors don't have the capability to make an instrument incorrect...all we can do is cover it up or fail it*. That's why it's important to regularly review some instrument malfunction scenarios, along with Primary/Supporting instrument combinations**, so you have the best chance of dealing with the malfunction.
*Some creative instructors might be able to cover the AI with a nice picture of an AI in some wonky attitude, or things like that.
**I disagree with the FAA...Primary/Supporting and Control/Performance are not two different methods for flying instruments. Control/Performance is how you normally fly. Primary/Supporting is how you interpret and troubleshoot instrument indications.
This is one area where a GOOD simulator can be a benefit. Many can simulate an instrument misbehaving.
That can be pretty insidious. There's a belief among some that these glass screens either work perfectly or go black. But that wouldn't appear to be true based on what you have told us here?
A friend shared that he's had ALL of the displays drop offline in the 777 more times than he can count.
The AHRS type LCD panel instruments tend to drop out or go into "red X" mode MORE often than showing bad data, but I can't think of a CFI or Commercial pilot I've talked to who has NOT seen a "modern" LCD panel show something completely wrong that makes no sense at all and doesn't match reality.
An old tech spinning gyro will fail. The new tech motion sensor chip was the cheapest part that met spec for the FAA direct from China and a mass assembly line in a building with safety nets around it catch suicidal workers jumping from the roof after the western press caught wind of the first few going splat.
Maybe you got a good one. Maybe you got one that the QC was done by the guy or gal who was thinking about jumping off the roof all day.
(The well known story of the suicidal workers is from an Apple iPhone assembler, but I seriously doubt the low production avionics assembly lines are a whole lot better when it comes to quality of life. I dunno. Ask Garmin if the plant they use has suicide nets hanging from the outside of the building.)
I do know this. That motion sensor driving that computer generated (read: Everything has bugs in software that aren't caught by the manufacturer OR the FAA certification process or your avionics guy wouldn't be calling saying you need to come in and reflash the units every couple of years...) was built by the lowest bidder when Garmin sent out the RFQ.
Do not ever blindly trust ANY instrument. They're all machines, whether physically rotating gyroscopes or acceleration sensors in chip packages sending voltages to a microprocessor running buggy code.
I'd say anecdotally yes. They go black or show a failure more often than they show bad data, but they do show bad data once in a while. When it happens VMC, no harm. When it happens IMC or worse, IMC on an approach -- things can get confusing and bad real quick.
Cross checking occasionally with the redundancy backups is likely to save your butt someday if you've ingrained the habit. If you haven't built that habit, you might follow bad information until your erroneous aircraft control inputs have put you into a situation where the airplane is going into the ground before you can figure out the problem and ignore the bad data displayed.
Instruments are guilty until proven innocent by their friends.
