As for the moral qualms regarding the use of drones for nuclear deterrence, this is a non-argument. A human would be controlling these drones on long-range missions and certainly be making the decision to use its weapons. Even if AI is to be incorporated into its flight control system, a sortie as critical as nuclear deterrence would almost definitely involve humans in both the flight and targeting decision aspects of the mission.
Sorry, but no.
Question 1 - Without the H-20 (or any manned aircraft with sufficient endurance), where would the human controllers even control the GJ-X from while still maintaining uninterrupted, reliable connection between the manned and unmanned elements?
Question 2 - Forgot about the S-70 incident over Ukraine? That S-70 lost contact with ground controllers during an operational test flight, forcing a Su-57 to shot it down after several failed attempts at reestablishing control with the drone as it was flying into Ukraine.
What if the same happens to a GJ-X carrying nuclear-tipped missiles during a deterrence patrol (whether that be from communication faults, or getting hacked through cyberwarfare (similar to the RQ-170 incident over Iran), or something else?)
Even with significant (if not full) AI-directed autonomous capability integrated into the GJ-X - What if any of the aforementioned possibilities occurred, and that the onboard AI assumed that the worst has happened (e.g. command aircraft has been downed/lost to enemy fire, if not the ground-based command post has been wiped out), and decided to conduct nuclear strikes on its own without prior authorization from humans? Hell - What if the onboard AI itself with controls over the nuclear missile release went rogue?
The thing is - Do you fully trust the AI to fly people safely from point A to point B? If not, then why would you trust AI to carry what could be the flashpoint that leads to nuclear doomsday on Earth?
In short, I see no justification to build a manned flying-wing-type bomber in the context of the GJ-X (if the GJ-X is indeed what this forum speculates it to be). With the money and human capital required of such an arguably-redundant project, the PLAAF could add many more GJ-Xs to its arsenal; this could mean a critical numerical advantage over the B-21/B-2 that would have profound impact on the military calculus of the 1/2/3IC/Asia-Pacific in general.
Having both manned H-20 and unmanned GJ-X with largely similar planform configurations (or with some deviations/variations) is certainly not mutually exclusive of one another.
Range and payload capacity aside - There's the factor where both the manned and unmanned elements share largely similar RCS category/scale, meaning enhanced survivability for the manned element in a contested/hostile environment. We've already seen the UADFs with largely similar planforms as the J-36 and J-XDS - I don't see how the same couldn't be achieved with the H-20 and GJ-X here.
There's also the point, similar to how some have already suggested above - The GJ-X, thanks to it being unmanned (and thus, should be reasonably cheaper to procure and operate) could have a tanker variant, which would enable mid-air refueling of other GJ-Xs (if not also the H-20s) during strike missions that are well past 2IC.
