For the purposes of evaluating morality of actions, yes artillery shells with variable fusings are a primitive version of an autonomous weapon.

It is much more useful to think about the reverse situation: how does viewing future autonomous weapons as a sort of artillery shell with high CEP (or whatever feature you imagine the future weapons has) clarify the morality of its use?

In my opinion, no. Killing people with dumb artillery shells vs. smart microdrone swarms seems equally horrific morally. The primary question is about whether one is more dangerous on global scales than the other. Although I think the smarter drones may in fact be more dangerous globally, I haven't seen anyone define a clean criteria for distinguishing them. Without such clean criteria, it looks infeasible to enforce through international law.

Yes, killing is morally reprehensible. But given that we'll not end the practice of war anytime soon, how can you regulate all the senseless killing people are determined to carry out so that it at least some pretence at morality is upheld. That is the sort of problem I had in mind.

I mean, if that's your interpretation I suppose you could argue that bullets are autonomous once fired.

Am I correct in interpreting your sarcastic comment as meaning you think that pursuing some definition of "autonomous" is a nonstarter here?

I think it’s abundantly obvious to anyone looking at this with good faith that the autonomy refers to the targeting of the weapon and the decision to fire the weapon.

If you're arguing about whether some weapon is a WMD after the fact, good faith isn't really something we can count on. The whole point of precise definitions is eliminating subjectivity. Anyway, that definition includes cruise missiles, which can't be WMDs because they're in widespread use. This line of inquiry is dead in the water.

Well, I haven't been able to come up with a good definition and I don't think the criterion you've pointed to is very promising.