Originally Posted by
etudiant
the pilot is baggage apart from the tactical decision making. The latter is driven by the sensor inputs, which can certainly be handled by an AI.
This is the key insight. We are well-used to hearing the argument for ‘sensor fusion’, which is that there is too much information coming into a F-35 for the single pilot to interpret it all, so a computer will do the processing needed to present a simple, clear air picture: red track bad, blue track good, yellow track unknown. While this is not quite the case yet, it is getting there. And it is one of the factors behind the huge expense of the aircraft.
The original rationale for sensor fusion was being able to dispense with a second crew member, and of course this was wholly acceptable to the pilot-dominated Air Force hierarchies. However, as AI and automated flight improve, there will eventually come a point when the only argument for retaining a pilot is the societally-driven need to keep a human in the loop on ethical decisions like employing weapons (on the reasonable assumption that electronic warfare will make beyond-line-of-sight remote control too unreliable). But, when fully implemented, sensor fusion will reduce the fighter pilot’s ethical input to ‘shooting red tracks OK, shooting other tracks bad’: and of course it will be a computer taking the in-cockpit decisions over which tracks to colour red. The remaining step (to actually shoot at them) is not such a big ethical leap as opponents of AI would like people to think.
I should add that this isn’t an endorsement of AI in all corners of the military need. Robust line-of-sight datalinks and/or an alternative cheap manned platform would offer an alternative for peacetime home air defence operations, allowing AI elements to be focussed on the war fighting requirement, where the need to avoid human losses is most pressing and the ‘societal ethics’ barrier is likely to be marginally lower. And the
relatively uncluttered air battle space makes it easier to foresee autonomous combat aircraft than (for instance) autonomous robot soldiers conducting house-to-house urban clearance operations with civilians present.
So I’m worried that we are doing Tempest 5-10 years too early, and thus including a cockpit, when the defining characteristic of 6th gen may well be the lack of a pilot. And this is exactly the kind of argument I think Dominic Cummings will raise during the UK’s forthcoming Integrated Review. The “start it now to keep BAES Warton busy after Typhoon” argument will cut little ice with him, one suspects.
Now, if the pilots had not pushed so hard for expensive/complex sensor fusion and had instead retained a single WSO as the crew of an automatically-piloted combat aircraft, there would not now be such an obvious path to a completely unmanned platform. Just saying... and I am a pilot. (Have I just hit on the answer to the MFTS debacle: stop training pilots to fly?!)