I know it'd get a lot of blowback from #AI involved researchers {if any of them gave a damn about my silly opinions}, but my suggestion is that all this AI hype be IGNORED TO DEATH until an actual AI can pass the Turing test. That means that the AI excusing itself for what it isn't or can't do is a NON-STARTER.
O look, CGPT15 rewrote my dissertation! [That just means the people on your committee who didn't try to defenestrate you are morons.]
No AI research should be made public UNTIL the Turing Test has been passed ... and we'll need statistics again: what /percentage/ of testers failed to out the machine?