There are clearly coherent "moral" arguments to be made against mainsteam AI, in areas such as resource consumption, capitalist power, and so on. Some are correct; others, while in my opinion unpersuasive, are at least coherent.

But the article places more stress on arguments of the sort "It's evil to use AI because it doesn't work very well", and those don't seem very logical to me. Oh, SOME arguments of that kind make sense, e.g. in the area of autonomous weapons, but the author didn't focus on extreme cases such as those.