Eventually we’ll get there, but not in a long long long time, far beyond the careers of the posters on this forum.
As one computer does not know when it is wrong, essentially you have to have several computer AI programs running at the same time, cross checking each other and taking out the odd one out.
these programs would have to be programmed by completely different teams/companies.
On the ground it gets interesting too with camera systems identifying threats on taxiways.
(dogs in India, etc, can I pass that aircraft without touching the tail, is it icy?)
the scope of such a project is so massive and the required infrastructure is so vast that for a long time it will be far cheaper and safer to keep a human in charge.