All good stuff IO540. Perhaps a few ILAAFFT autopilot incidents might not go amiss.
1. After several years of using a fully coupled attitude based 3 axis A/P with altitude hold and capture ( separate Yaw damper) and automatic trim, I converted to a rate based 2 axis unit, again with altitude hold but no auto trim. In the cruise with the A/P having been engaged for some time, I decided it was time to descend, only for some obscure reason ( I think because the trim wheel was not periodically moving and thus giving me a constant reminder that the A/P was engaged) I forgot that it was indeed actually engaged and reduced power for the descent. I was surprised when instead of the nose dropping, it remained pegged to the horixon (altitude hold engaged) I therefore attempted to move the stick forward and found to my horror it was apparently "Locked." This was followed by winding in a fair dose of forward trim and a further power reduction in order to encourage a descent. Still the nose remained pegged on the horizon and airspeed began to fall rapidly. Realisation suddenly dawned and (ignoring the trim warning lights) I instinctively switched off the A/P resulting in an immediate severe bunt nose down which pitched myself and the front seat passenger up to the ceiling. Fortunately the event occurred in good VMC and immediate manual recovery was instinctive and uneventful. One only needs to imagine a reverse scenario however, where an attempt to climb preceded by a power increase, a tug on the stick, a large dose of aft trim and immediate A/p disengagement would result in a pitch up stall spin scenario which in IMC would have obvious repercussions. Lesson learned, Know your A/P inside out, read the operating manual, understand what mode it is in and avoid the "whats it doing now" situation.
2. Flying R/H seat with another (experienced) pilot in the left hand seat we reached 200 feet on the climb out when he announced that "Strange forces" were acting on the ailerons. I took the stick and immediately realised we had taken off with the A/P engaged. After a quick check of the engagement mode I disconnected and the flight continued uneventfully. The implications were obvious however. a) An unsatisfactory pre flight check had been carried out with failure to notice the A/P already engaged. b) Had the previous pilot actually therefore landed with the A/P engaged? (Far from unknown-imagine battling a cross wind landing against an engaged A/P and yaw damper) c) If altitude hold had been engaged, an attempted takeoff with the pilot commanding a climb against an A/P resolutely determined to maintain runway elevation poses an interesting end of runway situation.
3. Most A/P's have a test or self test facilty which together with an operational check must be carried out prior to departure. First time engagement of the A/P in the air without a ground test first is not a good idea and can result in some A/P's inducing some very dramatic attitude excursions in the air!
As others have said it is not usually the A/p which is the problem, It's the pilot!