I'm all for shitting on replacing people with AI, but in this case it's done with the agreement of the person, who is still able to give it, who can't talk anymore, and for a documentary. So sure, they could have done it with a voice-over actor, and maybe I'd have preferred it too, but I can't really say this feels wrong. At this point it feels a bit like Stephen Hawking using his voice synthesis software.
If the person was unable to agree and didn't write what is being told with "their" voice though? That'd be shit.