A question not usually addressed in the back-and-forth on this subject is, "what technology are the advocates for single-pilot and autonomous flight ops relying upon?" Typically the advocates point to particular systems, like GPWS and its refinements, and assert that the human element is becoming or will become unnecessary.
That is, when the advocates get away with ignoring the second type of necessary technology - the computer software (or must we refer to code only as algorithms now?) capable of truly substituting for human flight crew members.
Case in point: show the effectiveness of code, operating today's flight control systems and those reasonably foreseen to be implemented within, say, five years, to figure out what was going wrong on the Delta flight out of LA a couple years ago in which the pilots first decided to dump fuel before returning to land. It was discussed in a thread on this forum, in part because of concern over where the fuel landed, and also among aviators about the sequence of, and reasons for, pilots' decisions and actions (though this SLF/attorney admits having forgotten the flight number).
Case in point: United flight out of Denver. P&W engine caught fire, due to a blade separation IIRC. All the P&W-powered United 777 aircraft were grounded. Flight returned to Denver safely. Show me the effectiveness of code.
Case in point, perhaps. American 191 out of Chicago ORD 25 May 1979. Engine separated and fell onto runway. Hydraulics severed, slats retracted, stall speed increased without any way for the flight crew to know these had occurred. (See NTSB final report, and as a veteran aviator poster on this forum has said in a different thread, it was a perfectly flyable airframe - if only they could have known what they were dealing with). Okay then, code advocates, fly the doomed, I mean fly the damn airplane.