Tesla has no safety life cycle process and as a result FSD is inherently unsafe, from the engineering point of view.
You can try to hand wave over that, but if you are an engineer working on the safety critical systems (it’s an actual engineering field), that’s all you need to know about FSD.
How will you discretely identify every problem to be solved in the domain of live real drivers in real world conditions? Applying a model of an industrial plant where machines interface with machines, and occasionally humans in prescribed and orderly functions, will not get you any closer to safety.
What will get you closer to safety is insanely large corpus of real world data being trained on in a continuous feedback loop.
You are not actually focusing on technical aspects. You are focusing on regulatory credentials and bureaucratic process. Focusing on technical aspects would mean looking at actual data from on road performance.
Tesla has no safety life cycle process and as a result FSD is inherently unsafe, from the engineering point of view.
You can try to hand wave over that, but if you are an engineer working on the safety critical systems (it’s an actual engineering field), that’s all you need to know about FSD.